Professional Documents
Culture Documents
Active Control of Off Shore Steel Jacket Platforms: Bao-Lin Zhang Qing-Long Han Xian-Ming Zhang Gong-You Tang
Active Control of Off Shore Steel Jacket Platforms: Bao-Lin Zhang Qing-Long Han Xian-Ming Zhang Gong-You Tang
Active Control
of Offshore
Steel Jacket
Platforms
Active Control of Offshore Steel Jacket Platforms
Bao-Lin Zhang • Qing-Long Han
Xian-Ming Zhang • Gong-You Tang
123
Bao-Lin Zhang Qing-Long Han
China Jiliang University Swinburne University of Technology
Hangzhou, Zhejiang, China Melbourne, VIC, Australia
This Springer imprint is published by the registered company Springer Nature Singapore Pte Ltd.
The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721,
Singapore
Preface
Offshore platforms are widely used to explore, drill, produce, store, and transport
ocean resources and are usually subjected to environmental loading, such as waves,
winds, ice, and currents, which may lead to failure of deck facilities, fatigue failure
of platforms, inefficiency of operation, and even discomfort of crews. In order to
ensure reliability and safety of offshore platforms, it is of great significance to
explore a proper way of suppressing vibration of offshore platforms. There are
mainly three types of control schemes, i.e., passive control, semi-active control, and
active control schemes, to deal with the vibration of offshore platforms. This book
provides a brief overview of these schemes and mainly presents recent advances of
active control schemes with optimal tracking control, sliding model control, delayed
feedback control, and network-based control.
Structure and readership This book consists of nine parts. An overview of
vibration control of offshore platforms is provided in Chap. 1, where passive control
schemes and several semi-active control schemes are briefly summarized; some
classical active control approaches, such as optimal control, robust control, and
intelligent control, are briefly reviewed; and recent advances of active control
schemes with optimal tracking control, integral sliding mode control, delayed
feedback control, and network-based control are deeply analyzed.
Dynamic models of offshore platforms Chapter 2 presents two dynamic models
of the offshore platforms and several mathematical lemmas used in this book. In
the first dynamic model, only the first dominant vibration mode of an offshore
steel jacket platform with an active mass damper (AMD) mechanism is taken into
account. The model is mainly utilized to design active controllers to attenuate
wave-induced vibration of the offshore platform. In the second dynamic model, the
first two dominant vibration modes of an offshore steel jacket platform subject to
an active tuned mass damper (TMD) mechanism are considered. By considering
parametric perturbations of the system and external disturbance, several uncertain
nonlinear models for the offshore platform are developed. Such models are utilized
to design active controllers to reduce vibration amplitudes of the offshore platform
subject to self-excited hydrodynamic forces and/or external disturbance.
v
vi Preface
while the safety of staff on the platform can be ensured. Chapters 8 and 9 focus on
dealing with network-based modeling and network-based control issues of offshore
platforms. In Chap. 8, for an offshore steel jacket platform with an active TMD
mechanism, a network-based state feedback control scheme is developed. Under
this scheme, the corresponding closed-loop system is modeled as a system with
an artificial interval time-varying delay, and a sufficient condition on the existence
of the network-based controller is obtained. The effects of network-induced time-
delays on the performance of the offshore platform are investigated.
Event-triggered H∞ reliable control in network environments In Chap. 9, a
network-based model of the offshore platform subject to external wave force
and actuator faults is presented, and an event-triggered H∞ reliable controller is
designed such that during the control implementation, only requisite sampled-data
are transmitted over networks. It is observed that the networked controllers are
capable of guaranteeing the stability of the offshore platforms and reducing the
required control cost. Moreover, the proposed network-based controllers are better
than some existing ones without network setting.
Acknowledgment We would like to acknowledge the collaborations with Profes-
sor Xinghuo Yu on the work of sliding mode control reported in this monograph. The
supports from the Key Project of Natural Science Foundation of Zhejiang Province
of China under Grants Z19F030002, the National Natural Science Foundation of
China under Grants 61773356, 61379029, and 61673357, the State Foundation for
Studying Abroad under Grant 201308330318, the Scientific Research Foundation
for the Returned Overseas Chinese Scholars, Ministry of Education of China under
Grant 2012-1707, the Natural Science Foundation of Zhejiang Province under
Grant Y1110036, the Academic Climbing Foundation of Youth Discipline Leaders
of Universities in Zhejiang Province under Grant PD2013190, the Australian
Research Council Discovery Projects under Grants DP1096780, DP0986376, and
DP160103567, and the Griffith University 2016 New Researcher Grant Scheme
under Grant 219128 are gratefully acknowledged. Finally, the close cooperation
with Springer as publisher and particularly with Dr. Jasmine Dou as Associate
Editor is gratefully acknowledged.
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Passive Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.1.1 Hysteretic and/or Viscoelastic Mechanisms . . . . . . . . . . . . . . . . . . . 2
1.1.2 Damping Isolation Mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.1.3 Dynamic Vibration Absorbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2 Semi-active Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.3 Active Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.3.1 Optimal Control. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.3.2 Robust Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.3.3 Sliding Mode Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.3.4 Delayed Feedback Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.3.5 Network-Based Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.4 Book Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2 Dynamic Models of Offshore Platforms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.1 Model of an Offshore Platform with AMD Mechanisms . . . . . . . . . . . . . 17
2.2 Model of an Offshore Platform with Active TMD Mechanisms. . . . . . 22
2.3 Some Related Mathematical Lemmas. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3 Optimal Tracking Control with Feedforward Compensation. . . . . . . . . . . 33
3.1 System and Problem Descriptions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.2 Design of Optimal Tracking Controllers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.3 Simulation Results and Discussions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.3.1 Performance of System with Optimal Tracking Controller . . . 39
3.3.2 Comparison of Optimal Controller and Tracking
Controller . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
3.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
3.5 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
ix
x Contents
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
Acronyms
A System matrix
A−1 Inverse of matrix A
AT Transpose of matrix A
A≥0 Symmetric positive semi-definite
A>0 Symmetric positive definite
A≤0 Symmetric negative semi-definite
A<0 Symmetric negative definite
det (A) Determinant of matrix A
diag(X1 , X2 , · · · , Xm ) Diagonal matrix with Xi as its ith diagonal element
I Identity matrix of appropriate dimensions
lim Limit
LMI Linear matrix inequality
N Positive integers
R Field of real numbers
Rn n-dimensional real Euclidean space
Rn×m Space of n × m real matrices
sgn(x) The sign of x
tr(A) Trace of matrix A
h The sampling interval of sensor
0n×m Zero matrix of dimension n × m
λ(A) Eigenvalue of matrix A
x(k) The state variable vector at time kT
|x| Absolute value (or modulus) of x
x Euclidean norm
P Induced norm supx=1 P x
∀ For all
∈ Belong to
→ Tend to, or mapping to (case sensitive)
xiii
xiv Acronyms
⊗
Matrix Kronecker product
sum
sup Supremum
inf Infimum
∗ Entries implied by symmetry
Chapter 1
Introduction
Offshore platforms are extensively used to explore, drill, produce, storage, and
transport ocean oil and/or gas resources in different depths. There are several
types of offshore platforms, such as self-elevating platforms, gravity platforms,
steel jacket platforms, tension leg platforms (TLPs), articulated leg platforms,
guyed tower platforms, spar platforms, floating production systems, and very large
floating structures. These platforms can be divided into fixed-bottom platforms
and buoyant platforms, which have their own particular purposes and different
configurations. To meet an increasing demand for marine sources of energy and
minerals, in the past several decades, a lot of research effort has been made on
offshore platforms. The related investigations are focused mainly on structure design
and monitoring, damage detection, fatigue analysis and reliability assessment,
mathematical modeling, and analysis of structures. Specifically, offshore platforms,
which are located in a very tough ocean environment over a long period of time, are
inevitably affected by environmental loading, such as waves, winds, ice, currents,
flow, and earthquakes [1, 2]. The environmental loading may lead to excessive
vibration of offshore platforms, thereby causing failure of deck facilities, fatigue
failure of structures, inefficiency of operation, and even discomfort of crews. Note
that reduction of vibration amplitude of an offshore platform by 15% can extend
service life over two times and can result in decreasing expenditure on maintenance
and inspection of structures [3]. Therefore, it is of great significance to explore
proper ways to reduce different types of vibrations of offshore platforms, and
comprehensive surveys of vibration control for offshore structures are provided by
Kandasamy et al. [4] and Zhang et al. [5].
Notice that a direct and simple way to mitigate vibration of offshore platforms
is to increase the stiffness of the platforms. As a result, natural frequencies can
be shifted away from resonating frequencies [6]. However, such schemes generally
require extra construction material, which unavoidably leads to increasing costs.
Thus, an alternative way is to choose a proper structural control method to reduce
structural vibration to an acceptable level [7, 8]. In the past several decades,
structural control schemes, such as passive control schemes [9], semi-active control
schemes [10], and active control schemes [11], are widely utilized to reduce
vibration of offshore platforms.
This chapter provides an overview of recent advances in vibration control of off-
shore platforms. Firstly, some passive control schemes including hysteretic mech-
anisms, viscoelastic mechanisms, damping isolation mechanisms, and dynamic
vibration absorbers utilized in vibration control are briefly outlined. Secondly, semi-
active control schemes are briefly surveyed. Thirdly, active control schemes are
reviewed in detail. Several active control schemes with optimal control, robust
control, intelligent control, sliding mode control, and sampled-data control are
presented. In particular, effectiveness and superiority of delayed feedback control
and network-based control, which are most recently developed active control
schemes, are deeply discussed. By purposely introducing time-delays into control
channel of offshore platforms, delayed feedback control schemes focus on the
controller design and investigate the effects of time-delays on active control for
the platforms. Network-based control schemes are concerned with network-based
modeling and controller design of offshore platforms in network settings.
Passive control does not require any other external energy, and control force is
generally derived from deformation of devices themselves, such as different pas-
sive energy dissipation mechanisms, damping isolation mechanisms, and dynamic
vibration absorbers [6]. Due to easiness and lower cost to implement and retrofit,
effectiveness to mitigate vibration and to maintain stability, and reliability of
offshore platforms, passive control has been implemented for offshore platforms
extensively and successfully in the last two decades. This section gives a brief review
of passive control for offshore platforms.
Hysteretic devices include metallic dampers and friction dampers, which dissipate
energy with no significant rate dependence. Viscoelastic devices are dependent on
frequency. The main types of viscoelastic devices are viscoelastic solid dampers
and viscoelastic fluid dampers. To mitigate vibration of offshore platforms and to
avoid damage of the offshore platforms, hysteretic and/or viscoelastic systems are
applied to the offshore platforms extensively. To mention a few, seismic response
of an offshore platform with shape memory alloy dampers is investigated in [12],
where influence of the number, location, and property of the shape memory alloy
dampers on vibration amplitudes of the offshore platform is discussed. It should be
1.1 Passive Control 3
Note that using damping dissipation devices to dissipate vibration energy generally
requires large relative deformation of the devices. However, relative deformation
of damping devices is not always large enough, which makes vibration reduction
unsatisfactory. In this case, dynamic vibration absorbers are feasible options. By
transferring some vibration energy to absorbers, an energy dissipation demand on
dominant vibration modes of offshore platforms is accomplished. The basic two
types of dynamic vibration absorbers are tuned mass dampers (TMDs) and tuned
liquid dampers (TLDs). In the past decades, TMD and TLD mechanisms have been
extensively implemented to mitigate vibration of offshore platforms.
well as surge motion [28] are investigated, respectively. It is shown that multiple
TMDs outperform a single TMD in terms of reducing oscillation amplitudes of
offshore platforms.
As aforesaid that TMD mechanisms have good control effect. However, such
methods generally add additional loads to offshore platforms. Therefore, applying
such schemes to deep-water platforms and flexible platforms is not always feasible.
To overcome disadvantage of TMDs where an additional mass is required to be the
mass body of a damper, an extended TMD and multiple extended TMD mechanisms
are proposed to reduce vibration of offshore platforms subject to earthquake and
wave loads, respectively [29, 30].
In [31], a TMD mechanism and a friction damper are applied to reduce wave-
induced vibration of an offshore platform, respectively. It is indicated that the
TMD and friction damper are both efficient in mitigating fatigue damage of the
offshore platform. It is also found that the former is more dependent on the dynamic
characteristics of the platform and the latter is more efficient for fixed steel jacket
platforms. In [32], by combining a TMD with a friction damper device, a hybrid
damping system is developed to control wave-induced and seismic vibrations and
fatigue damage of offshore platforms. In [33], based on the principle of TMD
mechanisms, an anti-vibration device is designed and installed on a similar model
of a jacket platform to control vertical and horizontal vibrations, respectively.
a control unit. In the past several decades, active control has gained increasing
attention both in theory and in practice, and various active control mechanisms
are proposed for offshore platforms, to mention a few, active mass damper (AMD)
mechanisms, active tendon mechanisms [11, 64], active tuned mass damper (TMD)
mechanisms [65, 66], propeller thruster mechanisms [64, 67–69], and even their
combinations [70]. Based on these mechanisms, a number of efficient active
schemes are reported in the literature, which are given in detail as follows.
It is well known that offshore platforms involve some uncertainties such as unknown
system parameters and structure flexibility. Most importantly, offshore platforms
are usually affected by various disturbances, such as waves, currents, winds, ice
1.3 Active Control 11
FFOC. It should be mentioned that vibration amplitudes of the offshore steel jacket
platform with the pure delayed H∞ controller are at the same level as the one with
a delay-free H∞ controller, while required control force by the former is much
smaller than the one by the latter. Similar results can also be found in [108], where a
pure delayed state feedback non-fragile control scheme is investigated for offshore
platforms with a TMD mechanism and controller perturbations. It is observed that
by choosing proper time-delays, vibration amplitudes of the offshore platforms with
a pure delayed control scheme are the same as or smaller than the ones with a delay-
free control scheme, while required control force by the delayed control scheme is
smaller than that by the delay-free control scheme.
Under the above control schemes, state/output feedback controllers are designed
in the continuous-time domain. When a continuous-time controller for an offshore
platform is implemented in practice, control signals are usually transmitted in a
digital form, which results in a sampled-data system. Hence, some sampled-data
control schemes are introduced to control offshore platforms. By taking input time-
varying delays, actuator faults, and linear fractional uncertainties into account, a
robust fault-tolerant sampled-data H∞ controller is developed to suppress wave-
induced vibration of an offshore steel jacket platform [109]. For an offshore
steel jacket platform subject to hydrodynamic forces, robust sampled-data control
schemes are presented to attenuate vibration of the platform with actuator faults and
parametric perturbations [110–113]. Simulation results indicates that the sampled-
data control schemes can significantly reduce wave-induced vibration and thereby
improve the control performance of the offshore platforms. Furthermore, it is found
that, compared with continuous-time controllers, sampled-data controllers may take
less control cost, while vibration amplitudes of offshore platforms can be reduced
to the similar level.
It is true that intentionally introducing proper time-delays can reduce vibration
amplitudes and required control force of an offshore platform. However, some
challenging issues need to be addressed. For instance, delayed feedback control
schemes are mainly based on simplified dynamic models of offshore steel jacket
platforms. As a result, only the first vibration mode or the first two vibration modes
are considered, while other higher vibration modes are ignored. Moreover, how
to choose a proper time-delay with positive effects on offshore platforms is still
challenging.
plicated and harsh ocean environmental loads. Thus, network-based control paves
an effective way to lower control costs and simplify installation and maintenance
of an offshore platform, while the safety of staff on the platform can be ensured. In
the recent years, several network-based modeling and active network-based control
schemes for offshore platforms are developed.
Based on an offshore platform with an active TMD mechanism [65, 72] or an
AMD mechanism [79], network-based dynamic modeling and controller design are
presented in [120] and [123], respectively. By inserting a communication network
over the offshore platform, a network-based model of the offshore platform is
established. Then a network-based state feedback controller is designed to suppress
the amplitudes of the offshore platform [120]. Under this scheme, the corresponding
closed-loop system of the offshore platform is modeled as a system with an interval
time-varying delay [121, 122]. Simulation results show that both the oscillation
amplitudes of the offshore platform and the required control force with a network-
based state feedback controller are smaller than those with a nonlinear controller
[65] and a dynamic output feedback controller [100]. Moreover, oscillation ampli-
tudes of the offshore platform with a network-based feedback controller are almost
the same as those with an integral sliding mode controller [99], while the required
control force is smaller than that required by an integral sliding mode controller.
In [123], a network-based control model is introduced for an offshore platform
with an AMD mechanism in the presence of actuator faults, where an event-
triggered mechanism is utilized to save the limited resources of the communication
networks [124, 125]. Based on a network-based dynamic model, an event-triggering
H∞ reliable controller is designed by employing the Lyapunov-Krasovskii func-
tional approach. It is found that for possible actuator failures, a network-based
controller is capable of guaranteeing the stability of the offshore platform. Com-
pared with a traditional H∞ controller, a network-based controller can suppress
vibration of the offshore platform to almost the same level while requiring less
control costs. Furthermore, with the event-triggered H∞ controller, limited network
resources can be saved significantly.
As mentioned above, network-based control can improve the control perfor-
mance of offshore platforms. However, using communication networks unavoidably
leads to some unfavorable factors, such as network-induced delays, packet dropouts,
quantization errors, and network congestion. By taking network-induced delays
into account, network-based control for offshore platforms is studied in [120]
and [123]. However, sustained attention to network-based control for offshore
platforms is still worthy to be paid. Specifically, by taking one or more factors
aforementioned into consideration, it is an interesting research topic to explore a
network-based dynamic model and develop effective network-based controllers for
floating offshore platforms in ocean environments.
1.4 Book Outline 15
In this chapter, two dynamic models of offshore platforms and several required
lemmas are introduced for investigating active control strategies in this book. In
the first dynamic model, only the first dominant vibration mode of an offshore steel
jacket platform with an AMD mechanism is taken into account [79]. This model
is utilized to design active controllers to attenuate wave-induced vibration of the
offshore platform. In the second dynamic model, the first and the second vibration
modes of an offshore steel jacket platform subject to an active TMD mechanism
are considered [65, 72]. By considering parametric perturbations of the system and
external disturbance, several uncertain nonlinear models for the offshore platform
are developed. Such models are used to design active controllers to reduce vibration
amplitudes of the offshore platform subject to self-excited hydrodynamic forces
and/or external disturbance.
where m1 , ω1 , and ξ1 are the modal mass, frequency, and damping ratio of
the offshore structure, respectively; m2 , ω2 , and ξ2 are the mass, frequency, and
damping ratio of the AMD, respectively; z1 (t) denotes the corresponding modal
coordinate which refers to the deck motion of the offshore structure, and z2 (t) is
displacement of the AMD; u(t) is the active control force, and f (t) is the external
wave forces.
Let
x1 (t) = z1 (t), x2 (t) = z2 (t), x3 (t) = ż1 (t), x4 (t) = ż2 (t) (2.2)
and denote
with
m2 ω22 m2 ω22
ā31 = −ω12 − , ā32 =
m1 m1
2m2 ξ2 ω2 2m2 ξ2 ω2
ā33 = −2ξ1 ω1 − , ā34 =
m1 m1
It is assumed that the wave force f (t) acting on the offshore structure comes
from the direction of x-axis, and the wave propagation is unidirectional [79, 82].
The power spectral density function of the wave elevation η(t) is given by the Joint
North Sea Wave Project (JONSWAP) wave spectrum as
−4
5Hs2 ω0 5 ω
Sη (ω) = exp −1.25 γ̄ β (2.5)
16ω0 ω ω0
ω2 = g m̄ tanh(m̄d0 ) (2.6)
where g denotes the gravitational acceleration. Then based on the linear wave
theory, for the wave elevation η(t) with wave frequency ω, the water particle
velocity υ(z, t), the acceleration υ̇(z, t), and the standard deviation συ (z) of the
velocity at location z can be written as
20 2 Dynamic Models of Offshore Platforms
where
cosh(mz) cosh(mz)
Tυη (ω, z) = ω , Tυ̇η (ω, z) = −j ω2 (2.10)
sinh(md0 ) sinh(md0 )
√
with j = −1.
Based on the linearized Morison equation [126, 127], along the cylindrical
structural members, the physical horizontal wave force p(z, t) per unit length is
in the form
1 8 1
p(z, t) = ρCd D̃ συ (z)υ(z, t) + ρπ Cm D̃ 2 υ̇(z, t) (2.11)
2 π 4
n
η(t) = ηj (t) (2.14)
j =1
where n is a specified positive integer and ηj (t) denotes the j th component of the
wave elevation with the form as
with Aj is the wave amplitude, ωj is the wave frequency, and ςj is the random
phase angle uniformly distributed between 0 and 2π .
Furthermore, denote
T
v(t) = η1 (t) η2 (t) · · · ηn (t)
T (2.16)
w(t) = v T (t) v̇ T (t)
and
⎧
⎨ G̃ = −diag{ω12 , ω22 , · · · , ωn2 }
n
(2.17)
⎩ H̃ = [1 1 · · · 1] T (ωj )
j =1
where
d0 1 8 1
T (ωj ) = ρCd D̃ συ (z)Tυη (ωj , z) + ρπ Cm D̃ Tυ̇η (ωj , z) φ(z)dz
2
0 2 π 4
Then, from (2.5), (2.11), and (2.12), the wave force f (t) can be modeled as the
output of a linear exogenous system as [81]
where
0̃ I˜
G= , H = H̃ I˜ 0̃ (2.19)
G̃ 0̃
with 0̃ and I˜ represent the n×n null matrix and the n×n identity matrix, respectively,
and matrices G̃ and H̃ are given by (2.17).
It should be specified that as an approximate representation of wave force, it can
be seen from (2.19) that the exogenous system (2.18) is stable but not asymptotically
stable, i.e., for any pole λ of the exogenous system, the following is true:
where σ (·) denotes the spectrum of a matrix. Specifically, it should be pointed herein
that the condition (2.20) is necessary for designing an optimal tracking controller in
Chap. 3.
22 2 Dynamic Models of Offshore Platforms
Consider an offshore steel jacket platform with three floors shown in Fig. 2.2 [11,
65, 72]. The platform is assembled from a concrete deck and 13 cylindrical steel
tube members with the dimensions reported in [11]. An active TMD mechanism,
which consists of a small mass, a spring and a viscous damper, is mounted on the
top of the offshore platform and is excited by a hydraulic servo mechanism. The
motion of the platform and the operation of the hydraulic servo influence the motion
of the damper. The hydraulic servo is driven by the active control force to reduce
vibration. The offshore platform is exposed to hydrodynamic force, which induces
a self-excited load term [65]. For simplicity, a monochromatic wave acting on the
offshore platform is considered and shown in Fig. 2.2, where h is the water depth,
H the wave height, λ the wave length, and Uow the current velocity at the water
surface. For details, one can refer to [11]. Since the first two modes of vibration
are the most dominant for controller design [65, 72, 79], in the sequel, we consider
the motions of the first two modes. When taking the parameter uncertainties of the
first and the second modes of the vibration and the coupled TMD, and external
disturbances acting on the two modes into consideration, the motion equations of
the first two modes of the vibration with the TMD can be described by
Fig. 2.2 An offshore steel jacket structure with a TMD [11, 65, 72]
2.2 Model of an Offshore Platform with Active TMD Mechanisms 23
⎧
⎪
⎪ z̈1 (t) = − [ω1 + Δω1 (t)]2 z1 (t) − 2[ξ1 + Δξ1 (t)][ω1 + Δω1 (t)]ż1 (t)
⎪
⎪
⎪
⎪ − φ1 [KT + ΔKT (t)][φ1 z1 (t) + φ2 z2 (t)] + φ1 [ζ1 (t) − u(t)]
⎪
⎪
⎪
⎪
⎪
⎪ − φ1 [CT + ΔCT (t)][φ1 ż1 (t) + φ2 ż2 (t)] + φ1 [KT + ΔKT (t)]zT (t)
⎪
⎪
⎪
⎪
⎪
⎪ + φ1 [CT + ΔCT (t)]żT (t) + f1 (z1 (t), z2 (t), t) + f2 (z1 (t), z2 (t), t)
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎨ z̈2 (t) = − [ω2 + Δω2 (t)] z2 (t) − 2[ξ2 + Δξ2 (t)][ω2 + Δω2 (t)]ż2 (t)
2
where K̂T , ĈT , and ξ̂i represent the maximum perturbations with respect to the
nominal values of KT , CT , and ξi , respectively, |ΔK̃T (t)| ≤ 1, |ΔC̃T (t)| ≤ 1, and
|Δξ̃i (t)| ≤ 1, i = 1, 2, T .
24 2 Dynamic Models of Offshore Platforms
we have
where
K̂T
ω̂T = , Δω̃T (t) = ΔK̃T (t) (2.26)
2KT
Similarly, the perturbation terms of the natural frequencies of the first and the second
vibration modes are given by
where ω̂i is a maximum perturbation with respect to the nominal value ωi and
|Δω̃i (t)| ≤ 1.
Then, by ignoring the higher-order infinitesimal o(Δω̃i (t)) and the second-order
terms Δξ̃i (t)·Δω̃i (t) and Δω̃i2 (t), i = 1, 2, T , it follows from (2.21), (2.22), (2.25),
and (2.27) that the dynamic model (2.21) can be written as
⎧
⎪
⎪ z̈1 (t) = − [ω12 + ω̄1 (t)]z1 (t) − [2ξ1 ω1 + ν̄1 (t)]ż1 (t) + φ1 [ζ1 (t) − u(t)]
⎪
⎪
⎪
⎪ − φ1 [KT + ΔKT (t)][φ1 z1 (t) + φ2 z2 (t)] + f1 (z1 (t), z2 (t), t)
⎪
⎪
⎪
⎪
⎪
⎪ − φ1 [CT + ΔCT (t)][φ1 ż1 (t) + φ2 ż2 (t)] + f2 (z1 (t), z2 (t), t)
⎪
⎪
⎪
⎪
⎪
⎪ + φ1 [KT + ΔKT (t)]zT (t) + φ1 [CT + ΔCT (t)]żT (t)
⎪
⎪
⎪
⎪
⎪
⎪ z̈ (t) = − [ω2 + ω̄ (t)]z (t) − [2ξ ω + ν̄ (t)]ż (t) + φ [ζ (t) − u(t)]
⎪
⎨ 2 2 2 2 2 2 2 2 2 2
where
ν̄i (t) = 2ωi [ξi ω̂i · Δω̃i (t) + ξ̂i · Δξ̃i (t)], i = 1, 2, T (2.29)
and
Denote
⎧
⎪
⎪ 0 0
⎨ ΔA0 (t) =
ΔKT ΔCT
(2.31)
⎪
⎪ 0 0
⎩ ΔAi (t) = , i = 1, 2, T
ω̄i (t) ν̄i (t)
where
A11 (t) = −φ12 · ΔA0 (t) − ΔA1 (t), A12 (t) = −φ1 φ2 · ΔA0 (t)
A22 (t) = −φ22 · ΔA0 (t) − ΔA2 (t), A13 (t) = φ1 · ΔA0 (t)
A21 (t) = A12 (t), A23 (t) = φ2 · ΔA0 (t)
A31 (t) = φ1 · ΔAT (t), A32 (t) = φ2 · ΔAT (t)
A33 (t) = −ΔAT (t)
Note that
ΔA0 (t) = Ē0 · diag{Δω̃T (t), ΔC̃T (t)}
(2.33)
ΔAi (t) = Ēi · diag{Δω̃i (t), Δξ̃i (t)} · H̄i , i = 1, 2, T
where
⎧
⎪
⎪ 0 0 0 0
⎨ Ē0 = , Ēi =
K̂T ĈT ω̂i ξ̂i
(2.34)
⎪
⎪
2
2ωi 2ξi ωi
⎩ H̄i = , i = 1, 2, T
0 2ωi
Denote
⎧
⎪
⎪ Ẽ = diag{Ē1 , Ē2 , ĒT }, Ẽ0 = Φ0 ⊗ Ē0
⎪
⎪ ⎡ ⎤
⎪
⎨ −H̄1 0 0
H̃ = ⎣ 0 −H̄2 0 ⎦ (2.35)
⎪
⎪
⎪
⎪ φ1 H̄T φ2 H̄T −H̄T
⎪
⎩ H̃ = diag{−I, −I, I }
0
26 2 Dynamic Models of Offshore Platforms
Let
⎧
⎪
⎨ M̃ = Ẽ Ẽ0
T T
Ñ = H̃ H̃0T (2.36)
⎪
⎩
F̃ (t) = diag{F̃1 (t), F̃0 (t)}
where
F̃1 (t) = diag{Δω̃1 (t), Δξ̃1 (t), Δω̃2 (t), Δξ̃2 (t), Δω̃T (t), Δξ̃T (t)}
F̃0 (t) = diag{Δω̃T (t), ΔC̃T (t), Δω̃T (t), ΔC̃T (t), Δω̃T (t), ΔC̃T (t)}
(2.37)
Then, the perturbation matrix ΔA(t) in (2.32) can be written as
Denote
x1 (t) = z1 (t), x2 (t) = ż1 (t), x3 (t) = z2 (t)
(2.40)
x4 (t) = ż2 (t), x5 (t) = zT (t), x6 (t) = żT (t)
and
Then the dynamic model (2.28) of the offshore platform can be written as
where x0 is the initial value, M̃, F̃ (t), and Ñ are denoted by (2.36), and the matrices
A, B, D, and D0 are given as
⎧ ⎡ ⎤
⎪
⎪
0 1 0 0 0 0
⎪ ⎢a
⎪
⎪
⎪ ⎢ 21 a22 a23 a24 a26 ⎥
a25 ⎥
⎪
⎪ ⎢ ⎥
⎪
⎪ ⎢ 0 0 0 1 0 0 ⎥
⎪
⎪ A=⎢ ⎥
⎪
⎪ ⎢ a41 a42 a43 a44 a46 ⎥
a45
⎪
⎪ ⎢ ⎥
⎪
⎪ ⎣ 0 0 0 0 0 1 ⎦
⎪
⎪
⎨ a61 a62 a63 a64 a65
a66
(2.42)
⎪
⎪ 1 T
⎪
⎪ B= 0 −φ1 0 −φ2 0
⎪
⎪ mT
⎪
⎪ T
⎪
⎪ 010000
⎪
⎪ D=
⎪
⎪
⎪
⎪
000100
T
⎪
⎪
⎪
⎪ 0 φ1 0 0 0 0
⎩ D0 =
0 0 0 φ2 0 0
with
It is assumed that in (2.41), the external disturbance term ζ (t) ∈ L2 [0, ∞]. As
shown in [65, 72], the self-excited wave force vector f (x, t) is uniformly bounded
and satisfies
As special cases of the model (2.41), several dynamic models of the offshore
platform are listed as follows.
In (2.21), if the external disturbance is not considered, the uncertain dynamic
model (2.41) reduces to the one in [101] as
Further, if setting ΔKT (t) ≡ 0 and ΔCT (t) ≡ 0, Δω̃i (t) ≡ 0 and Δξ̃i (t) ≡ 0, i =
1, 2, T , the dynamic model (2.44) reduces to the ones in [65, 72] and [100] as
In (2.21), if the uncertainties on damping and stiffness of the TMD are not
taken into account, and the perturbation terms Δω1 (t), Δω2 (t), and ΔωT (t) of
the nominal natural frequencies of the first two vibration modes and the TMD are
approximated by the additive uncertainty forms, the terms ν̄i (t) in (2.29) and ω̄i (t)
in (2.30) can be written as
ν̄i (t) = 2ξi ω̂i · Δω̃i (t) + 2ωi ξ̂i · Δξ̃i (t)
(2.46)
ω̄i (t) = 2ωi ω̂i · Δω̃i (t), i = 1, 2, T
In this case, the uncertain dynamic model (2.41) reduces to the one in [90] and [91]
as
and F̂ (t) = F̃1 (t) satisfying F̂ T (t)F̂ (t) ≤ I, ∀t ≥ 0, and Ẽ and F̃1 (t) are given
by (2.35) and (2.37), respectively, and
2ωi 2ξi
Ĥi = , i = 1, 2, T
0 2ωi
If the perturbations of the systematic parameters are not considered, the nominal
form of the offshore platform systems (2.41) and (2.47) is derived as
In this section, some main lemmas are provided to obtain the main results of this
book. The proof of these lemmas can be found in the literature.
S11 S12
Lemma 2.1 (Schur complement) For a symmetric matrix S = , where
∗ S22
S11 ∈ Rn×n . Then the following conditions are equivalent:
1. S < 0;
T S −1 S < 0;
2. S11 < 0, S22 − S12 11 12
−1 T
3. S22 < 0, S11 − S12 S22 S12 < 0.
Lemma 2.2 ([128]) Given a symmetric matrix Y and matrices H , F (t), E of
appropriate dimensions with F T (t)F (t) ≤ I , then
Y + εH H T + ε−1 E T E < 0
μi + νj = 0, i = 1, 2, · · · , n, j = 1, 2, · · · , m, (2.52)
W + U GV T + (U GV T )T < 0 (2.53)
30 2 Dynamic Models of Offshore Platforms
is solvable for matrix G with appropriate dimensions if and only if the following
holds
where
M1T + M1 −M1T + M2
M :=
∗ −M2T − M2
x(t) XY
ξ(t) := , ≥0
x(t − h) ∗ Z
with Y := M1 M2
Lemma 2.8 ([133]) For any constant matrix R ∈ Rn×n , R = R T , matrix S ∈
Rn×n , R = [ R∗ RS ] ≥ 0, scalars 0 ≤ τ1 ≤ τ (t) ≤ τ2 , and a vector function
ẋ : [−τ2 , −τ1 ] → Rn such that the following integration is well defined, it holds
that
2.3 Some Related Mathematical Lemmas 31
t−τ1
4
− (τ2 − τ1 ) ẋ T (s)R ẋ(s)ds ≤ − ∇k (2.58)
t−τ2 k=1
where
Lemma 2.9 ([134]) For any constant matrix R > 0 ∈ Rn×n , a scalar function
γ (t) with 0 < γ (t) ≤ γM , and a vector function
t ė : [−γM , 0] → Rn such that the
following integration is well defined, let t−γ (t) ė(s)ds = Mζ (t) where M ∈ Rn×m
and ζ (t) ∈ Rm . Then the following inequality holds for any matrix N ∈ Rn×m
t
− ėT (s)R ė(s)ds ≤ −ζ T (t) M T N + N T M − γ (t)N T R −1 N ζ (t)
t−γ (t)
(2.59)
Chapter 3
Optimal Tracking Control with
Feedforward Compensation
This chapter presents an optimal tracking control methodology for an offshore steel
jacket platform subject to external wave force. Based on a dynamic model of an
offshore steel jacket platform with an AMD mechanism and a linear exogenous
system model of the external wave force on the offshore platform, an optimal
tracking control scheme with feedforward compensation is proposed to attenuate
wave-induced vibration of the offshore platform. A feedforward and feedback
optimal tracking controller (FFOTC) can be obtained by solving an algebraic Riccati
equation and a Sylvester equation, respectively. It is demonstrated that the wave-
induced vibration amplitudes of the offshore platform under the FFOTC are much
smaller than the ones under the feedback optimal tracking controller (FOTC) and
the feedforward and feedback optimal controller (FFOC). Furthermore, the required
control force under the FFOTC is smaller than the ones under the FOTC and the
FFOC.
This chapter is organized as follows. Section 3.1 presents the linear quadratic
performance index optimal tracking control problem for the offshore steel jacket
platform. Section 3.2 investigates the design process of the feedforward and feed-
back optimal tracking controller. Some numerical examples are given in Sect. 3.3
to demonstrate the effectiveness of the proposed control scheme and its superiority
over some existing methods. Finally, Sect. 3.4 concludes the chapter, and Sect. 3.5
presents a brief note.
where
1000
C= (3.2)
0010
where M ∈ Rr×r and N ∈ R2×r are constant matrices with appropriate dimensions.
The quadratic average performance index is chosen as
T
1
J = lim eT (t)Qe(t) + uT (t)Ru(t) dt (3.4)
T →∞ 2T 0
T
1
J = lim x T C T QCx − 2x T C T QNz + zT N T QNz + uT Ru dt
T →∞ 2T 0
(3.6)
Applying the maximum principle to system (2.3) with the output Eq. (3.1) and
the performance index (3.6) yields the optimal tracking control law satisfying
where λ(t) ∈ R4×1 is an adjoint vector satisfying the following two-point boundary
value problem
ẋ(t) = Ax(t) − BR −1 B T λ(t) + Df (t), x(0) = x0
(3.8)
λ̇(t) = −C T QCx(t) + C T QNz(t) − AT λ(t), λ(∞) = 0
where P1 ∈ R4×4 is the symmetric feedback gain matrix and P2 ∈ R4×r and P3 ∈
R4×2n are feedforward gain matrices to be determined.
Then, the control law (3.7) can be written as
which indicates that the existence and uniqueness of the optimal tracking control
law (3.10) is equivalent to the existence and uniqueness of the feedback gain matrix
P1 and the feedforward gain matrices P2 and P3 .
Substituting (3.9) into the first equation of (3.8) and noting (2.18) yield
On the other hand, from the second equation of (3.8) and (3.9), we have
(P1 A + AT P1 − P1 BR −1 B T P1 + C T QC)x(t)
+ (AT P2 − C T QN + P2 M − P1 BR −1 B T P2 )z(t)
+ (AT P3 + P3 G + P1 DH − P1 BR −1 B T P3 )w(t) = 0 (3.14)
Notice the fact that (3.14) is true for any x(t), z(t), and w(t). Then, one yields
the Riccati equation as
P1 A + AT P1 − P1 BR −1 B T P1 + C T QC = 0 (3.15)
and
respectively.
By Assumption 1, there exists a unique positive-definite solution P1 to the Riccati
equation (3.15).
It is clear from optimal control theory that A − BR −1 B T P1 is a Hurwitz matrix.
Then, by Lemma 2.3, Assumption 2 and the condition (2.20) guarantee that the
Sylvester equations (3.16) and (3.22) have unique solutions P2 and P3 , respectively.
Based on the above analysis, we present the following proposition which
provides the design method of the feedforward and feedback optimal tracking
controller.
Proposition 3.1 Consider the optimal tracking control problem of the offshore
platform system (2.3) with (3.1) subject to the quadratic performance index (3.4).
Under Assumptions 1 and 2, there exists a unique feedforward and feedback optimal
tracking control law of the form (3.10), where the matrix P1 is the unique positive-
definite solution to the Riccati matrix equation (3.15), and the matrices P2 and P3
are the unique solutions to the Sylvester equations (3.16) and (3.22), respectively.
Remark 3.2 Suppose that the quadratic performance index (3.4) is modified as
T
1
J = lim x T (t)Q̄x(t) + uT (t)R̄u(t) dt (3.19)
T →∞ 2T 0
Remark 3.3 In terms of implementation, if the state w(t) of the exogenous sys-
tem (2.18) is unavailable in the practical engineering, under Assumption 3, one
can design a disturbance-observer-based feedforward and feedback near optimal
tracking control law as
where ŵ(t) is the state estimation of the exogenous system (2.18) and satisfies
˙
ŵ(t) = Gŵ(t) + Θ f (t) − H ŵ(t) , ŵ(0) = ŵ0 (3.24)
with ŵ0 is the initial state of the disturbance observer (3.24), andΘ is the designed
observer matrix with appropriate dimensions.
Correspondingly, for the feedforward and feedback optimal control law (3.20),
one can design a disturbance-observer-based feedforward and feedback near optimal
control law as
Table 3.1 Parameters of the offshore platform (2.3) and the wave force
Description Symbol Value Unit
Length of the offshore platform L 249 m
Diameter of the cylinder D̃ 1.83 m
Mass of the offshore platform m1 7,825,307 kg
Mass of the AMD m2 78,253 kg
Natural frequency of the offshore platform ω1 2.0466 rad/s
Natural frequency of the AMD ω2 2.0074 rad/s
Damping ration of the offshore platform ξ1 0.02 –
Damping ration of the AMD ξ2 0.2 –
Significant wave height Hs 7 m
Water depth d 218 m
Peak frequency of wave ω0 0.79 rad/s
Drag coefficient Cd 1.0 –
inertia coefficient Cm 1.5 –
Peakedness coefficient γ̄ 3.3 –
with and without feedforward compensation will be made first. Then, the feed-
forward and feedback optimal tracking control scheme will be compared with the
feedforward and feedback optimal control scheme [81] and the superiority of the
proposed optimal tracking control scheme with feedforward compensation will be
demonstrated.
In Fig. 2.1, the values of the masses, natural frequencies and the damping ratios
of the offshore platform with the AMD, and the wave-related parameters are from
[86], which are presented in Table 3.1. Based on these settings, the matrices A, B,
and D in (2.42) can be obtained as
⎧ ⎡ ⎤
⎪
⎪ 0 0 1.0000 0
⎪ ⎢
⎪
⎪
⎪ ⎢ 0 0 0 1.0000 ⎥
⎥
⎨A = ⎣
⎪
−4.2290 0.0403 −0.0899 0.0080 ⎦
−4.0297 0.8030 −0.8030 (3.26)
⎪
⎪ 4.0297
⎪
⎪ −4
T
⎪
⎪ B = 10 × 0 0 −0.0013 0.1278
⎪
⎩ T
D = 10−6 × 0 0 0.1278 0
T
The initial state x0 of the system is given as x0 = −0.02 0.03 0.01 0.03 . To
compute the wave force, let n = 7 in (2.14). Then based on the values given by
Table 3.1, one yields the matrices G̃ and H̃ in (2.19) as
6
x 10
8
2
Wave Force (N)
−2
−4
−6
−8
0 10 20 30 40 50 60 70 80
Time (s)
Furthermore, from (2.18), we can obtain the total wave force acting on the offshore
platform, which is shown in Fig. 3.1.
Due to the irregular nature of the wave force applied, both the peak values and the
root mean square (RMS) values of displacement, velocity of the offshore platform,
and the control force are investigated. Let Md , Mv , and Mu represent the peak
values of displacement, velocity of the offshore platform, and the required control
force, respectively, and Jd , Jv , and Ju denote the RMS values of displacement,
velocity of the offshore platform, and the control force, respectively, where
⎧
⎪
⎪ M = max{|x1 (t)|, t ∈ [0, T ]}, Jd = 1 T 2
⎪
⎨ d T 0 x1 (t)dt
1 T 2
⎪ Mv = max{|ẋ1 (t)|, t ∈ [0, T ]}, Jv = T 0 ẋ1 (t)dt
(3.27)
⎪
⎪
⎩ M = max{|u(t)|, t ∈ [0, T ]}, J = 1 T 2
u u T 0 u (t)dt
amplitudes of displacement and velocity and the required control force under no
control, FOTC, and FFOTC will be compared to demonstrate the superiority of the
proposed feedforward and feedback optimal tracking control scheme.
First, when no controller is applied to the offshore platform, it can be computed
that the peak values of the displacement and the velocity of the offshore platform
are 0.5295 m and 0.7327 m/s, respectively, and the RMS values of the displacement
and the velocity are 0.1921 m and 0.3108 m/s, respectively.
Then, we turn to design a feedforward and feedback optimal tracking controller
(FFOTC). For this, in the performance index (3.4), set
M = −0.2I2 , N = I2 (3.29)
Solving the Riccati equation (3.21), Sylvester equations (3.16) and (3.22) yield
⎧ ⎡ ⎤
⎪
⎪ 169750000 −72378 −20347000 −244140
⎪
⎪ ⎢ −72378
⎪
⎪ 259.84 39127 391.27 ⎥
⎪
⎪ P1 = ⎢
⎣ −20347000
⎥
⎪
⎪
⎪ 39127 11158000 92822 ⎦
⎨
−244140 391.27 92822 992.7
⎡ ⎤
⎪
⎪ −10105000 97979000
⎪
⎪
⎪
⎪ ⎢ −27000 −5000 ⎥
⎪
⎪ P2 = ⎢
⎣ −21262000
⎥ , P3 = Υ1 Υ2 Υ3
⎪
⎪
⎪ −4252000 ⎦
⎩
−211000 −42000
where
⎧ ⎡ ⎤
⎪
⎪ 757520 373220 128710 65080 40740
⎪
⎪ ⎢ −680
⎪
⎪
⎪ ⎢ −840 −440 −250 −160 ⎥ ⎥
⎪
⎪ Υ 1 = ⎣ −243510 −171050 −69580 −35900 −22280 ⎦
⎪
⎪
⎪
⎪
⎪
⎪ −2630 −1810 −730 −370 −230
⎪
⎪ ⎡ ⎤
⎪
⎪ 28190 20610 756970 791380 389170
⎪
⎨ ⎢ −110 −80 700 −560 −500 ⎥
Υ2 = ⎢ ⎥
⎣ −15410 −11430 −88820 −242420 −140740 ⎦
⎪
⎪
⎪
⎪
⎪
⎪ −150 −110 −1060 −2630 −1520
⎪
⎪ ⎡ ⎤
⎪
⎪ 216010 137380 95310 70090
⎪
⎪
⎪
⎪ ⎢ −320 −220 −150 −110 ⎥
⎪
⎪ Υ3 = ⎢ ⎥
⎪
⎪ ⎣ −81180 −52090 −36180 −26580 ⎦
⎪
⎪
⎩
−870 −560 −390 −290
3.3 Simulation Results and Discussions 41
0.6
No control
FOTC
FFOTC
0.4
Displacement of the Offshore Platform (m)
0.2
−0.2
−0.4
−0.6
0 10 20 30 40 50 60 70 80
Time (s)
Fig. 3.2 Displacement of the offshore platform under no control, FOTC, and FFOTC
1
No control
0.8 FOTC
FFOTC
0.6
Velocity of the Offshore Platform (m/s)
0.4
0.2
−0.2
−0.4
−0.6
−0.8
−1
0 10 20 30 40 50 60 70 80
Time (s)
Fig. 3.3 Velocity of the offshore platform under no control, FOTC, and FFOTC
Then an FFOTC (3.10) and a feedback optimal tracking controller (FOTC) with
the form (3.18) are obtained, respectively. Under no control, FFOTC, and FOTC,
the response curves of the displacement, the velocity of the system (2.3), and
the required control force are presented in Figs. 3.2, 3.3, and 3.4, respectively.
The tracking errors of the displacement and the velocity of the offshore platform
without control and under FFOTC and FOTC are depicted in Figs. 3.5 and 3.6,
42 3 Optimal Tracking Control with Feedforward Compensation
7
x 10
5
FOTC
4 FFOTC
2
Control Force (N)
−1
−2
−3
−4
−5
0 10 20 30 40 50 60 70 80
Time (s)
1
FOTC
FFOTC
0.8
0.6
Tracking Error of the Displacement
0.4
0.2
−0.2
−0.4
−0.6
−0.8
0 10 20 30 40 50 60 70 80
Time (s)
Fig. 3.5 Tracking error of the displacement under FOTC and FFOTC
respectively. From these two figures, one can see that FFOTC has a smaller tracking
error than FOTC. It also can be seen from Figs. 3.2, 3.3, and 3.4 that FFOTC can
significantly reduce the oscillation amplitudes of the displacement and the velocity
of the offshore steel jacket platform. In addition, the required control by FFOTC is
much smaller than the one by FOTC.
3.3 Simulation Results and Discussions 43
0.8
FOTC
FFOTC
0.6
0.4
Tracking Error of the Velocity
0.2
−0.2
−0.4
−0.6
−0.8
0 10 20 30 40 50 60 70 80
Time (s)
Fig. 3.6 Tracking error of the velocity under FOTC and FFOTC
In fact, it can be computed that under FOTC, the peak values of the displacement
and the velocity of the offshore platform are 0.2104 m and 0.2003 m/s, respectively,
and the maximum control force is about 4.2738 ×107 N. The RMS values of the
displacement and the velocity of the offshore platform are 0.1137 m and 0.0901 m/s,
respectively, and the RMS value of the control force is 2.4552 ×107 N. Under
FFOTC, the peak values of the displacement, the velocity of the offshore platform,
and the required control force are 0.0298 m, 0.0214 m/s, and 2.6898 ×107 N,
respectively, and the RMS values of the displacement, the velocity of the offshore
platform, and the control force are 0.0211 m, 0.0171 m/s, and 1.6241 ×107 N,
respectively.
To compare the two different control schemes clearly, the peak and RMS values
of the offshore platform and the required control force without control and under
FOTC and FFOTC are listed in Table 3.2. From this table, one can obtain the
following facts.
• Under FOTC, the peak values of the oscillation amplitudes of the displacement
and the velocity of the offshore platform are reduced by 60% and 72%,
respectively, while under FFOTC, there are further reduced by 94% and 97%,
respectively.
• From the point of view of the RMS values, it can be obtained that under FOTC,
the RMS values of the displacement and the velocity are reduced by 41% and
71%, respectively, while under FFOTC, there are even reduced by 89% and 94%,
respectively.
44 3 Optimal Tracking Control with Feedforward Compensation
Table 3.2 Peak and RMS values of displacement, velocity of the offshore platform, and the
required control force without control and under FOTC and FFOTC
Peak value RMS value
Controller
Md (m) Mv (m/s) Mu (107 N) Jd (m) Jv (m/s) Ju (107 N)
No control 0.5295 0.7327 – 0.1921 0.3108 –
FOTC 0.2104 0.2003 4.2738 0.1137 0.0901 2.4552
FFOTC 0.0298 0.0214 2.6898 0.0211 0.0171 1.6241
• Moreover, both the peak and RMS values of the control force required by FOTC
are 1.5 times as the ones by FFOTC. It indicated that FFOTC is more efficient
than FOTC to attenuate the wave-induced vibration of the offshore platform.
To compare the optimal tracking control scheme proposed in this paper with the
optimal control scheme in [81], based on the design process in [81], we first
design a feedforward and feedback optimal controller (FFOC). For this, discretizing
system (2.3) with a sampling period 0.1 yields
where
⎧ ⎡ ⎤
⎪
⎪ 0.9790 0.0002 0.09885 0.00005
⎪
⎪ ⎢ 0.0189 0.9805 0.00453 0.09545 ⎥
⎪
⎪ ⎢ ⎥
⎪
⎪ Ã = ⎣ −0.4179 0.0038 0.9701 0.00095 ⎦
⎪
⎪
⎪
⎨
0.3655 −0.3844 0.09522 0.90380
⎡ ⎤ ⎡ ⎤
⎪
⎪ −6.16 × 10−10 6.348 × 10−10
⎪
⎪
⎪ ⎢ 6.199 × 10−8 ⎥ ⎢ −11 ⎥
⎪
⎪
⎪ B̃ = ⎢ ⎥ , D̃ = ⎢ 1.876 × 10 ⎥
⎪
⎪ ⎣ −1.205 × 10−8 ⎦ ⎣ 1.263 × 10−8 ⎦
⎪
⎩
1.219 × 10−6 5.783 × 10−10
where
Ĝ11 Ĝ12
Ĝ = (3.32)
Ĝ21 Ĝ11
3.3 Simulation Results and Discussions 45
with
⎧
⎨ Ĝ11 = diag{1.0000, 1.0000, 0.9999, 0.9999, 0.9998, 0.9998, 0.9997}
Ĝ = diag{0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01}
⎩ 12
Ĝ21 = −diag{0.0013, 0.0050, 0.0113, 0.0201, 0.0314, 0.0453, 0.0616}
(3.33)
In the following discrete quadratic performance index [81]
1 T
N
Jˆ = lim [x (k)Q̂x(k) + u(k)R̂u(k)]dt (3.34)
N →∞ N
k=0
where
⎧
⎪
⎨ K̂1 = 10 × 1.4675 0.0311 7.9532 0.0045 , K̂2 = 1.0042
7
Applying the FFOC to the offshore platform, it can be computed that the peak
values of the oscillation amplitudes of the displacement and the velocity of the
offshore platform are 0.2189 m and 0.0244 m/s, respectively, and the RMS values
of the displacement and the velocity are 0.1037 m and 0.0112 m/s, respectively.
The peak and RMS values of the control force are about 4.3327 ×107 N and
2.0153 ×107 N, respectively, which are listed in Table 3.3. The responses of the
displacement and the velocity of the offshore platform and the required control
force by FFOC and FFOTC are depicted in Figs. 3.7, 3.8, and 3.9, respectively.
From Table 3.3 and the figures, one can see that:
• The oscillation amplitudes of the velocity of the offshore platform under FFOC
and FFOTC are almost in the same level, while the oscillation amplitudes of the
displacement of the offshore platform under FFOTC are much smaller than the
one under FFOC.
• Moreover, the control force required by FFOTC is smaller than that by FFOC.
To sum up, it can be clearly observed that the proposed optimal tracking con-
troller with feedforward compensation can significantly reduce the wave-induced
vibration of the offshore platform and thereby improve the control performance of
46 3 Optimal Tracking Control with Feedforward Compensation
Table 3.3 Peak and RMS values of displacement, velocity of the offshore platform and the
required control force under FFOC [81] and FFOTC
Peak value RMS value
Controller
Md (m) Mv (m/s) Mu (107 N) Jd (m) Jv (m/s) Ju (107 N)
FFOC 0.2189 0.0244 4.3327 0.1037 0.0112 2.0153
FFOTC 0.0298 0.0214 2.6898 0.0211 0.0171 1.6241
0.3
FFOC
0.25 FFOTC
0.2
Displacement of the Offshore Platform (m)
0.15
0.1
0.05
−0.05
−0.1
−0.15
−0.2
−0.25
−0.3
0 10 20 30 40 50 60 70 80
Time (s)
Fig. 3.7 Displacement of the offshore platform under FFOC and FFOTC
0.05
FFOC
0.04 FFOTC
Velocity of the the Offshore Platform (m/s)
0.03
0.02
0.01
−0.01
−0.02
−0.03
−0.04
−0.05
0 10 20 30 40 50 60 70 80
Time (s)
Fig. 3.8 Velocity of the offshore platform under FFOC and FFOTC
3.5 Notes 47
8
x 10
1
FFOC
0.8 FFOTC
0.6
0.4
Control Force (N)
0.2
−0.2
−0.4
−0.6
−0.8
−1
0 10 20 30 40 50 60 70 80
Time (s)
the offshore platform. Furthermore, the designed FFOTC is better than the FOTC
and FFOC [81] in two respects: vibration amplitudes of the displacement and the
velocity of the offshore platform and the required control force.
3.4 Conclusions
This chapter presents a feedforward and feedback optimal tracking control scheme
for the offshore platform. The gain matrices of the optimal tracking controller
can be obtained by solving an algebraic Riccati equation and Sylverter equations,
respectively. Simulation results have shown that under the feedforward and feedback
optimal tracking controllers, both wave-induced vibration amplitudes of the offshore
steel jacket platform and the required control force are significantly reduced.
3.5 Notes
Optimal control schemes have been extensively utilised to the active control of
offshore platforms. The earlier results concerning optimal control for offshore
platforms can be found in Kawano [71], Yoshida et al. [76], Abdel-Rohman [11],
Suneja et al. [75], Ahmad et al. [77], Terro [72], and Alves et al. [78]. The optimal
controllers in the frequency domain are designed by Yoshida et al. [76] and Mahadik
48 3 Optimal Tracking Control with Feedforward Compensation
et al. [73]. A nonlinear stochastic optimal control scheme for a jacket platform
is developed by Luo et al. [74]. By using feedforward compensation and optimal
control strategy, feedfoward and feedback optimal controllers are advocated by
Wang et al. [80], Ma et al. [81, 82], and Zhang et al. [85], where the key point
of controller design is that the wave force acting on offshore platforms is modeled
as an output of a linear exogenous system.
Based on [83], this chapter presents a feedforward and feedback optimal tracking
control scheme for a jacket-type offshore platform subject to wave force. Note that
discrete control schemes are easy to implement for computer control systems. For
the offshore steel jacket platform with control delays, a discrete feedforward and
feedback optimal tracking control scheme with memory is developed by Zhang et
al. [84]. As for the detailed design process and analysis of the control scheme, we
refer to Zhang et al. [84].
Indeed, optimal control is effective to reduce vibration of offshore platforms
and thereby can improve the performance of the offshore platform systems. How-
ever, optimal control generally requires the exact dynamic model of the offshore
platforms. Consequently, the practical implementation of controller and the control
effects are confined. It is still challenging to develop effective optimal controllers to
uncertain and general dynamic models of offshore platforms, which is an important
research topic in the future.
Note that offshore platforms are inevitably affected by ocean waves, ice, winds,
flow, and even earthquakes. The external loading generally results in random
features of dynamics of offshore platforms. Applying stochastic control theory to
active control of offshore platform is a natural and feasible way [74]. However,
few results about stochastic control for offshore platforms are by far available in
the literature. Therefore, for the offshore platform systems, some issues including
system modeling, filtering, and controller design in the stochastic control theory
framework deserve further exploration.
Chapter 4
Integral Sliding Mode H∞ Control
In this chapter, sliding mode H∞ control for an offshore steel jacket platform subject
to nonlinear self-excited wave force and external disturbance is developed. A sliding
mode H∞ controller is designed to reduce oscillation amplitudes of the offshore
platform. In the case that the dynamic model of the offshore platform is subject to
parameter perturbations, a robust sliding mode H∞ control scheme is proposed. It is
found through simulation results that compared with an H∞ controller and a sliding
mode controller, the sliding mode H∞ controller requires much less control force;
and the oscillation amplitudes of the offshore platform under the sliding mode H∞
controller are less than those under the sliding mode controller.
The rest of this chapter is organized as follows. In Sect. 4.1, the problem of
sliding mode H∞ controller design for an offshore platform system subject to
external disturbance is formulated. Section 4.2 presents the sliding surface design
and stability analysis of the resulting sliding motion. In Sects. 4.3 and 4.4, the design
results of the sliding mode H∞ controller and the robust sliding mode H∞ controller
are presented, respectively. The simulation results are given in Sect. 4.4 to illustrate
the usefulness and the advantages of the proposed methodology, conclusions are
given in Sect. 4.6, and a note is given in Sect. 4.7.
Consider the offshore platform system (2.50), where the external disturbance
on the offshore platform are considered. To design a controller to reduce the
oscillation amplitudes of the system, one should specify the control output so that
the performance index from the external disturbance to the control output can be
realized with the specified requirement. For this aim, the control output is given as
In what follows, a sliding mode H∞ control scheme is developed such that the
system (2.50) with (4.1) satisfies
(i) in the designed sliding surface, the resulting closed-loop system is asymptoti-
cally stable; and under the zero initial condition, the H∞ performance
where Λ = P A + AT P + P BK + K T B T P .
Proof We first consider the asymptotic stability of the system (4.6) with ζ (t) = 0
described by
where P > 0. Taking the derivative of V1 (x(t)) with respect to t along the
trajectory of system(4.8), noting that (2.43), and introducing a new vector α T (t) =
T T
x (t) f (x(t), t) , one obtains
where
Λ + μ2 I P D̄
Ψ = (4.11)
∗ −I
If the matrix inequality (4.7) holds, then by Schur complement, we have Ψ < 0,
which means that the system (4.8) is asymptotically stable.
In the following, we prove that the H∞ performance is guaranteed for nonzero
ζ (t) ∈ L2 [0, ∞] under zero initial condition.
Taking the derivative of V1 (x(t)) with respect to t along the trajectory of sliding
motion (4.6), and noticing that (2.43), one yields
where
⎡ ⎤
Λ + μ2 I + C1T C1 P D̄ P D̄0 + C1T D1
Ξ =⎣ ∗ −I 0 ⎦ (4.14)
∗ 0 D1 D1 − γ I
T 2
It is clear that if the matrix inequality (4.7) is feasible, then applying the Schur
complement yields Ξ < 0, which leads to
Integrating both sides of (4.15) from 0 to ∞, noting the fact that V1 (x(0)) = 0 under
zero initial condition, we have
∞
[ηT (t)η(t) − γ 2 ζ T (t)ζ (t)]dt < 0 (4.16)
0
which means the H∞ performance (4.2) is guaranteed. This completes the proof.
s(t)
u(t) = Kx(t) − (GB)−1 ρ(x(t), t) (4.17)
|s(t)|
Proposition 4.2 Consider the system (2.50) with (4.1). If the sliding surface is
given by (4.4), where K is the solution of matrix inequality (4.7), then the
reachability of sliding surface s(t) = 0 is ensured by the sliding mode H∞ control
law (4.17).
Proof To analyze the reachability, we choose the Lyapunov function candidate as
1 2
V2 (s(t)) = s (t) (4.19)
2
From (2.50), (4.4), and (4.17), it follows that
s(t)
ṡ(t) = GDf (x(t), t) + GD0 ζ (t) − ρ(x(t), t) (4.20)
|s(t)|
which implies that under the control law (4.17), the trajectories of the system (2.50)
will be driven onto the specified sliding surface s(t) = 0.
Note that Proposition 4.1 provides a stability condition for sliding motion (4.6).
It is clear that the condition is nonlinear due to the nonlinear term P BK. In
order to obtain the controller gain K in (4.17), pre- and post-multiplying (4.7) by
diag{P −1 , I, I, I, I }, respectively, and setting P̄ = P −1 and K̄ = KP −1 , we
have the following equivalent version of Proposition 4.1.
Proposition 4.3 Under Assumption 1, for given scalars μ > 0 and γ > 0, the
sliding motion (4.6) is asymptotically stable, and the H∞ performance (4.2) is
guaranteed for nonzero ζ (t) ∈ L2 [0, ∞] and the prescribed γ > 0 if there exist a
6 × 6 real matrix P̄ > 0 and a 1 × 6 real matrix K̄ such that
⎡ ⎤
Λ̄ D̄ D̄0 μP̄ P̄ C1T
⎢ ∗ −I 0 0 ⎥
⎢ 0 ⎥
⎢ ⎥
⎢ ∗ ∗ −γ 2 I 0 D1T ⎥ < 0 (4.22)
⎢ ⎥
⎣∗ ∗ ∗ −I 0 ⎦
∗ ∗ ∗ ∗ −I
where
Λ̄ = AP̄ + P̄ AT + B K̄ + K̄ T B T (4.23)
As two special cases of the sliding mode H∞ control law, based on Proposi-
tion 4.3, a sliding mode control law and an H∞ control law can be easily designed
by following Corollaries.
Corollary 4.1 For a given scalar μ > 0, if there exist a 6 × 6 real matrix P̄ > 0
and a 1 × 6 real matrix K̄ such that
⎡ ⎤
Λ̄ D̄ μP̄
⎣ ∗ −I 0 ⎦ < 0 (4.24)
∗ ∗ −I
then, under the sliding mode control law (4.17), the system (2.50) is asymptotically
stable in the sliding surface s(t) = 0; and the reachability of the sliding surface is
guaranteed. Moreover, the gain K of the control law is given by K = K̄ P̄ −1 .
Corollary 4.2 Under Assumption 1, for given scalars μ > 0 and γ > 0, if there
exist a 6 × 6 real matrix P̄ > 0 and a 1 × 6 real matrix K̄ such that
⎡ ⎤
Λ̄ D D0 μP̄ P̄ C1T
⎢ ∗ −I 0 0 ⎥
⎢ 0 ⎥
⎢ ⎥
⎢ ∗ ∗ −γ I 0 D1T ⎥ < 0
2 (4.25)
⎢ ⎥
⎣∗ ∗ ∗ −I 0 ⎦
∗ ∗ ∗ ∗ −I
the system (2.50) with (4.1) is asymptotically stable; and the H∞ performance (4.2)
is guaranteed for nonzero ζ (t) ∈ L2 [0, ∞] and the prescribed γ > 0. Moreover,
the gain K of the control law (4.26) is given by K = K̄ P̄ −1 .
In this subsection, we intend to design a robust sliding mode H∞ control law for an
uncertain system (2.47).
Suppose that the sliding surface function is designed as the same form as (4.4).
Then, an equivalent control can be written as
s(t)
u(t) = Kx(t) − (GB)−1 ρ1 (x(t), t) (4.29)
|s(t)|
where
where Λ̄ is given by (4.23). Then, the sliding motion (4.28) is robustly stable, and the
H∞ performance (4.2) is guaranteed for nonzero ζ (t) ∈ L2 [0, ∞] and a prescribed
γ > 0; the reachability of the sliding surface s(t) = 0 is guaranteed by the sliding
mode H∞ control law (4.29). Moreover, the gain matrix K of the control law (4.29)
is given by K = K̄ P̄ −1 .
Proof The proof are similar to those in Propositions 4.1 and 4.2 and thus are omitted
here.
Correspondingly, the corollaries stated below provide approaches to designing
robust sliding mode control law and robust H∞ control law, respectively.
Corollary 4.3 For a given scalar μ > 0, if there exist a 6 × 6 real matrix P̄ > 0, a
1 × 6 real matrix K̄, and a scalar ε > 0 such that
56 4 Integral Sliding Mode H∞ Control
⎡ ⎤
Λ̄ D̄ μP̄ εM̂¯ P̄ N̂ T
⎢ ⎥
⎢∗ −I 0 0 0 ⎥
⎢ ⎥
⎢∗ ∗ −I 0 0 ⎥ < 0, (4.32)
⎢ ⎥
⎣∗ ∗ ∗ −εI 0 ⎦
∗ ∗ ∗ ∗ −εI
then, under the robust sliding mode control law (4.29), the uncertain system (2.47)
is robustly stable in sliding surface s(t) = 0; and the reachability of sliding surface
is guaranteed; and the gain matrix K of the control law is given by K = K̄ P̄ −1 .
Corollary 4.4 Under Assumption 1, for given scalars μ > 0 and γ > 0, if there
exist a 6 × 6 real matrix P̄ > 0, a 1 × 6 real matrix K̄, and a scalar ε > 0 such that
⎡ ⎤
Λ̄ D D0 μP̄ P̄ C1T M̂¯ P̄ N̂ T
⎢ ∗ −I 0 ⎥
⎢ 0 0 0 0 ⎥
⎢ ⎥
⎢ ∗ ∗ −γ 2 I 0 D1T 0 0 ⎥
⎢ ⎥
⎢∗ ∗ ∗ −I 0 0 0 ⎥ < 0, (4.33)
⎢ ⎥
⎢∗ ∗ ∗ ∗ −I 0 0 ⎥
⎢ ⎥
⎣∗ ∗ ∗ ∗ ∗ −εI 0 ⎦
∗ ∗ ∗ ∗ ∗ ∗ −εI
then, under the robust H∞ control law in the form as (4.26), the uncertain
system (2.47) with (4.1) is robustly stable; and the H∞ performance (4.2) is
guaranteed for nonzero ζ (t) ∈ L2 [0, ∞] and the prescribed γ > 0. Moreover,
the gain matrix K of the control law is given by K = K̄ P̄ −1 .
In this section, we will first design a sliding mode H∞ controller (SMHC) formed
as (4.17) for the system (2.50) with (4.1) to show effectiveness of the proposed slid-
ing mode H∞ control scheme. The performances of the system under the SMHC,
H∞ controller (HIC) and sliding mode controller (SMC) are compared. Then,
for the uncertain system (2.47) with (4.1), a robust sliding mode H∞ controller
(RSMHC) defined in (4.29) will be given to improve the control performance.
An offshore platform with a TMD mechanism presented in Fig. 2.2 is simulated. The
parameters of the system and the waves are given in Table 4.1, which are derived
from [65] and [72]. With the setting in the table, the nonlinear wave force f (x, t) can
4.5 Simulation Results 57
In this subsection, under the SMC, HIC, and SMHC, the performance of the
system (2.50) with (4.1) will be analyzed.
Firstly, when no controller is applied to the system, the responses of the three
floors of the system are presented in Fig. 4.1. The oscillation amplitudes of the
first, second, and third floors peak to peak are 1.4159, 1.5270, and 1.6061 m,
respectively. The average response of the three floors peak to peak is 1.5164 m.
It can be found that the offshore platform is very dangerous to work. Secondly,
we consider the sliding mode control scheme. Set μ = 0.8, δ = 0.1 and
G = [500 1 1000 1 0 100]. By Corollary 4.1, we obtain the gain of controller
SMC as
When this SMC is applied to the system, the responses of the first, second, and third
floors and the required control force are shown in Fig. 4.2, from which one can see
that under the SMC, the oscillation amplitudes of the first, second, and third floors
peak to peak are 0.2333, 0.2537, and 0.2688 m, respectively. The control force peak
to peak by the SMC is about 1.8401 × 105 N.
Under the SMC, the variation of the sliding surface s(t) is illustrated in Fig. 4.3.
It can be found that the sliding function s(t) = 0 and the average value of the sliding
function s(t) are equal to zero. Theoretically, the sliding function should be zero,
Response of floor 1 (m)
1 1
0.5 0.5
0 0
-0.5 -0.5
-1 -1
0 50 100 150 0 50 100 150
Time (s) Time (s)
Response of floor 3 (m)
0.5
-0.5
-1
0 50 100 150
Time (s)
Fig. 4.1 Responses of the three floors of the system (2.50) without control
4.5 Simulation Results 59
0.1 0.1
0 0
-0.1 -0.1
-0.2 -0.2
0 50 100 150 0 50 100 150
Time (s) Time (s)
Response of floor 3 (m)
× 10 5
0.2 1.5
Fig. 4.2 Responses of the three floors of the nominal system (2.50) and the control force by SMC
500
400
300
Variation of sliding surface
200
100
-100
-200
-300
-400
-500
0 50 100 150
Time (s)
s(t) = 0. However, due to the self-excited nonlinear wave force and the external
disturbance, there always exists the deviations in the sliding function.
Thirdly, we study the H∞ control scheme. Set γ = 0.2. By Corollary 4.2, we
obtain the gain of an HIC as
Under the HIC, the peak-to-peak oscillation amplitudes of the first, second, and third
floors are 0.2040, 0.2220, and 0.2358 m, respectively. The range of the required
control force by the HIC is about 2.6952 × 105 N. The responses of the three floors
and the required control force are presented in Fig. 4.4.
Finally, we turn to the sliding mode H∞ control scheme. Let γ = 0.2. By
Proposition 4.3, the gain of an SMHC is obtained as
When the SMHC is applied to the system (2.50) with (4.1), the displacements
of the three floors and the required control force are presented in Fig. 4.5, and the
curve of the sliding function s(t) is given by Fig. 4.6. It can be seen that the peak-to-
peak oscillation amplitudes of the first, second, and third floors are 0.1998, 0.2177,
and 0.2317 m, respectively. The control force peak to peak by the SMHC is about
1.5449 × 105 N.
Response of floor 1 (m)
0.2 0.2
0.1 0.1
0 0
-0.1 -0.1
-0.2 -0.2
0 50 100 150 0 50 100 150
Time (s) Time (s)
Response of floor 3 (m)
× 10 5
0.2 2
Control force (N)
0.1 1
0 0
-0.1 -1
-0.2 -2
0 50 100 150 0 50 100 150
Time (s) Time (s)
Fig. 4.4 Responses of the three floors of the nominal system (2.50) and the control force by HIC
4.5 Simulation Results 61
0.1 0.1
0 0
-0.1 -0.1
-0.2 -0.2
0 50 100 150 0 50 100 150
Time (s) Time (s)
Response of floor 3 (m)
× 10 4
0.2 10
-0.2 -5
0 50 100 150 0 50 100 150
Time (s) Time (s)
Fig. 4.5 Responses of the three floors of the nominal system (2.50) and the control force by SMHC
400
300
200
Variation of sliding surface
100
-100
-200
-300
-400
-500
0 50 100 150
Time (s)
Table 4.2 Maximum control forces and oscillation amplitudes of three floors of the system (2.50)
under different controllers
γ Controller Floor 1 (m) Floor 2 (m) Floor 3 (m) Control force(105 N)
– No control 1.4159 1.5270 1.6061 –
– SMC 0.2333 0.2537 0.2688 1.8401
0.2 HIC 0.2040 0.2220 0.2358 2.6952
SMHC 0.1998 0.2177 0.2317 1.5449
0.3 HIC 0.2046 0.2226 0.2364 2.7102
SMHC 0.2027 0.2208 0.2352 1.5564
0.4 HIC 0.2036 0.2216 0.2354 2.0259
SMHC 0.2026 0.2204 0.2343 1.5137
0.5 HIC 0.2041 0.2221 0.2357 2.0242
SMHC 0.2043 0.2226 0.2371 1.6039
0.7 HIC 0.2040 0.2220 0.2357 2.0458
SMHC 0.2041 0.2221 0.2361 1.5480
In Table 4.2, in other cases of γ = 0.3, 0.4, 0.5 and 0.7, the oscillation
amplitudes of the system and the control force under the HIC and SMHC are
compared, where the performances of the system under the SMC and the case of
without control are also presented.
From Table 4.2 and Figs. 4.1, 4.2, 4.3, 4.4, and 4.5, one can see clearly that:
• The SMHC can reduce the oscillation amplitudes of the system to about 14%
of the oscillation amplitudes of the system without control while the SMC can
reduce the oscillation amplitudes of the system to about 17% of the oscillation
amplitudes of the system without control;
• Under the SMHC and the SMC, the average oscillation amplitude of the three
floors is about 0.22 and 0.25 m, respectively. Clearly, the oscillation amplitudes
of the three floors under the SMHC are less than those under the SMC. Moreover,
it can be found that the control force required by the SMHC is less than the one
by the SMC;
• Under the SMHC and the HIC, the average oscillation amplitude of the three
floors is about 0.22 m. However, it is not difficult to observe that the SMHC
requires less control force than the HIC.
In this subsection, the performance of the uncertain system (2.47) with (4.1) is
investigated when no controller, and a robust sliding mode controller (RSMC), a
robust H∞ controller (RHIC), and a robust sliding mode H∞ controller (RSMHC)
are applied, respectively.
4.5 Simulation Results 63
Depicted in Fig. 4.8 are the responses of the three floors and the required control
force when the RSMC is used to system (2.47), and presented by Fig. 4.9 is
the corresponding curve of the sliding function s(t). It can be obtained through
calculation that under the RSMC, the peak-to-peak oscillation amplitudes of the
first, second, and third floors are reduced from 1.4180, 1.5371, and 1.6181 to 0.2270,
0.2433, and 0.2549 m, respectively, and the control force peak to peak is about
4.5180 × 106 N.
Let γ = 0.15. By Corollary 4.4, we design an RHIC with the gain as
Under the RHIC, the peak-to-peak oscillation amplitudes of the first, second, and
third floors are 0.1934, 0.2096, and 0.2230 m, respectively; and the range of the
control force is about 8.5366 × 106 N, which can be observed from Fig. 4.10.
Then, in the case of γ = 0.15, by Proposition 4.4, we obtain the gain of an
RSMHC as
0.5 0.5
0 0
-0.5 -0.5
-1 -1
0 50 100 150 0 50 100 150
Time (s) Time (s)
Response of floor 3 (m)
0.5
-0.5
-1
0 50 100 150
Time (s)
Fig. 4.7 Responses of the three floors of the uncertain system (2.47) without control
Response of floor 1 (m)
0.2 0.2
0.1 0.1
0 0
-0.1 -0.1
-0.2 -0.2
0 50 100 150 0 50 100 150
Time (s) Time (s)
× 10 6
Response of floor 3 (m)
0.2 4
Control force (N)
0.1 2
0 0
-0.1 -2
-0.2 -4
0 50 100 150 0 50 100 150
Time (s) Time (s)
Fig. 4.8 Responses of the three floors of the uncertain system (2.47) and the control force by
RSMC
4.5 Simulation Results 65
8000
6000
Variation of sliding surface
4000
2000
-2000
-4000
-6000
-8000
0 50 100 150
Time (s)
0.2 0.2
0.1 0.1
0 0
-0.1 -0.1
-0.2 -0.2
0 50 100 150 0 50 100 150
Time (s) Time (s)
Response of floor 3 (m)
× 10 6
0.2 5
Control force (N)
0.1
0 0
-0.1
-0.2 -5
0 50 100 150 0 50 100 150
Time (s) Time (s)
Fig. 4.10 Responses of the three floors of the uncertain system (2.47) and the control force by
RHIC
66 4 Integral Sliding Mode H∞ Control
0.1 0.1
0 0
-0.1 -0.1
-0.2 -0.2
0 50 100 150 0 50 100 150
Time (s) Time (s)
Response of floor 3 (m)
× 10 6
0.2
0 0
-0.1
-1
-0.2
0 50 100 150 0 50 100 150
Time (s) Time (s)
Fig. 4.11 Responses of the three floors of the uncertain system (2.47) and the control force by
RSMHC
When the RSMHC is applied to the uncertain system, the peak-to-peak oscillation
amplitudes of the three floors are reduced to 0.1985, 0.2152 and 0.2278 m, and the
required control force peak to peak is about 1.9309 × 106 N. Figure 4.11 shows
the responses of the three floors of the uncertain system and the curve of the
control force, and Fig. 4.12 depicts the variation of the sliding surface s(t) under
the RSMHC.
For the sake of comparison, the oscillation amplitudes of the uncertain system
and the control force under the RSMC, RHIC, RSMHC, and no control are
summarized in Table 4.3, where for different values of γ , the simulation results
are compared between the RHIC and the RSMHC.
It is found from Table 4.3 that for the uncertain system (2.47) with (4.1),
applying the RSMC, RHIC, and RSMHC, the average oscillation amplitudes of
the three floors are reduced to 16%, 14%, and 14% of the oscillation amplitudes
of the system without control, respectively. In fact, under the RSMC, the average
oscillation amplitude of the three floors is about 0.24 m, while under the RHIC and
the RSMHC, the average oscillation amplitude of the three floors is about 0.22 m,
which indicates that the controlled average oscillation amplitude of the system under
the RSMHC is smaller than the one under the RSMC and almost the same as that
under the RHIC. In addition, it can be observed that the required control force under
the RSMHC is smaller than the ones under the RSMC and the RHIC.
4.6 Conclusions 67
2000
1500
Variation of sliding surface
1000
500
-500
-1000
-1500
-2000
0 50 100 150
Time (s)
Table 4.3 Maximum control forces and oscillation amplitudes of three floors of the uncertain
system (2.47) under different controllers
γ Controller Floor 1 (m) Floor 2 (m) Floor 3 (m) Control force (106 N)
– No control 1.4180 1.5371 1.6181 –
– RSMC 0.2270 0.2433 0.2549 4.5180
RHIC 0.1934 0.2096 0.2230 8.5366
0.15
RSMHC 0.1985 0.2152 0.2278 1.9309
RHIC 0.1993 0.2159 0.2286 2.1350
0.20
RSMHC 0.2027 0.2199 0.2326 1.7424
RHIC 0.2000 0.2160 0.2292 8.8975
0.25
RSMHC 0.2031 0.2202 0.2324 2.2739
RHIC 0.2014 0.2176 0.2306 2.6955
0.30
RSMHC 0.2041 0.2214 0.2339 1.9714
RHIC 0.2030 0.2193 0.2322 1.9707
0.45
RSMHC 0.2079 0.2251 0.2383 1.6647
4.6 Conclusions
This chapter presents a sliding mode H∞ control scheme for offshore platforms
subject to nonlinear wave forces and external disturbance. Applying this scheme,
controllers for the nominal system and the uncertain system have been designed
to reduce the oscillation amplitudes of the offshore platforms. Based on simulation
results, the designed sliding mode H∞ controller, sliding mode controller, and the
traditional H∞ controller are compared from two aspects: the oscillation amplitudes
of offshore platforms and the control cost. Simulation results have shown that the
68 4 Integral Sliding Mode H∞ Control
4.7 Notes
The main results of this chapter are derived from Zhang et al. [90], where both
parametric perturbations of the system and the external disturbance are consid-
ered. Specifically, if the external disturbance is ignored, the offshore platform
system (2.50) reduces to the one as (2.45) by Terro et al. [72] and Zribi et al. [65];
and the uncertain dynamic model (2.47) reduces to the one as (2.49), similar analysis
about the uncertain model can be found in Zhang et al. [99]. In this situation, an
integral sliding mode control scheme and a robust integral sliding mode control
scheme are developed for the nominal system (2.45) and the uncertain system (2.49),
respectively. For more discussions on integral sliding mode control for the offshore
platforms, one refers to [99]. It is found that the integral sliding mode control
schemes can reduce the internal oscillations of the offshore steel jacket platform
dramatically; and the performance of the offshore steel jacket platform under the
integral sliding mode control schemes is better than the ones under the nonlinear
control scheme [65] and the dynamic output feedback control scheme [100]. More
recently, Nourisola et al. [102, 103] have developed adaptive integral sliding mode
control schemes for the offshore steel jacket platforms. For an offshore platform
with an AMD mechanism, Zhang et al. [98] offer an optimal sliding mode controller,
which is better than the feedforward and feedback optimal controller proposed by
Ma et al. [81].
Note that the sliding mode control methods aforesaid are mainly based on the
simplified dynamic models of offshore steel jacket platform [65, 72, 79], where
either only the most dominant mode or the first and the second vibration modes of
the offshore platform with a single load are considered. However, offshore platforms
are of characteristics of combinations of multi-vibration modes. Moreover, offshore
platforms are usually subject to parametric perturbations, nonlinear dynamics, and
mixed effects of external disturbance, such as winds and waves, or earthquake, and
waves and flow. Therefore, it is significant to develop a nonlinear model for an
offshore platform such that its overall performance can be improved. In this sense,
a challenging problem is to establish a proper mathematical model that can exactly
reflect the dynamics of an offshore platform.
In fact, on the one hand, as an efficient, simple and robust control scheme,
sliding mode control is worthy to be further investigated for the complicated
nonlinear dynamic models of jacket-type offshore platform and even other types
of offshore platforms; on the other hand, note that neural networks are suitable for
4.7 Notes 69
approximating nonlinear systems and fuzzy logics are efficient to handle systematic
nonlinearities and uncertainties and easy to implement for structural systems.
Offshore platforms are typical nonlinear systems. However, except for several work
by Zhou and Zhao [92], Chang et al. [93], Kim [94, 95], and Cui and Hong
[96], the study on intelligent structure and/or intelligent controllers has not been
adequately explored for offshore platforms. Specifically, since it is very complicated
and impossible for offshore platforms to establish exact dynamic models. In this
situation, based on a great number of experimental and field data, to explore data-
based dynamic models and to develop effective data-based control schemes for
offshore platforms are of significance both in theory and in real implementations,
which is well worth investigation in the near future.
Chapter 5
Delayed Integral Sliding Mode Control
This chapter is concerned with active control for an offshore steel jacket platform
subject to wave-induced force and parameter perturbations. The offshore steel jacket
platform is shown in Fig. 2.2 [65, 72]. The dynamic model under consideration is
given by (2.44). By intentionally introducing a proper time-delay into the control
channel, a novel sliding mode control scheme is proposed. This scheme utilizes
mixed current and delayed states. It is shown through simulation results that this
scheme is more effective in both improving control performance and reducing
control force of the offshore platform than some existing ones such as delay-free
sliding mode control [99], nonlinear control [65], dynamic output feedback control,
and delayed dynamic output feedback control [100]. Furthermore, it is shown that
the introduced time-delay in this scheme can take values in different ranges, while
the corresponding control performance of the offshore platform is almost at the
similar level.
The rest of this chapter is organized as follows. Section 5.1 presents an scheme to
design a sliding mode controller with mixed current and delayed sates. An numerical
algorithm is given in Sect. 5.2 to solve the controller gain matrices. Section 5.3
provides simulation results to demonstrate the effectiveness of the sliding mode
control scheme with mixed current and delayed states, and to investigate the effects
of artificially introduced time-delays on the control for the offshore platform.
Finally, Sect. 5.4 concludes the chapter, and Sect. 5.5 provides a brief note.
In this section, for the system (2.44), we first present a new scheme to design the
sliding mode surface with mixed current and delayed states. Then, we provide an
approach to designing a desired sliding mode controller for the offshore platform.
where
Ψ11 = AP̄ + P̄ AT + B K̄ + K̄ T B T + Q̄ − R̄
Ψ12 = B K̄τ + R̄, Ψ14 = τ P̄ AT + τ K̄ T B T
Ψ22 = −Q̄ − R̄, Ψ24 = τ K̄τT B T , Ψ34 = τ D̄ T
Ψ44 = −P̄ R̄ −1 P̄ , Ψ46 = ετ Ē
where xt = x(t + s), s ∈ [−τ, 0], P > 0, Q > 0 and R > 0. Taking the
time derivative of V1 (xt ) along the trajectory of (5.3), after some simple algebraic
manipulation, we have
Applying Lemma 2.6 to the integral term in (5.5), together with (5.3), one has
V̇1 (xt ) ≤ ηT (t) Ξ (t) + Γ T (t)(τ 2 R)Γ (t) η(t)
where
⎡ ⎤
U (t) P BKτ + R P D̄
Ξ (t) := ⎣ ∗ −Q − R 0 ⎦
∗ ∗ −I
Γ (t) := [A + BK + Ē F̃ (t)Ñ BKτ D̄]
T
η(t) := x T (t) x T (t − τ ) f T (x, t)
74 5 Delayed Integral Sliding Mode Control
with
U (t) = P A + AT P + P BK + K T B T P + Q − R
+μ2 I + P Ē F̃ (t)Ñ + Ñ T F̃ T (t)Ē T P
If Ξ (t) + Γ T (t)(τ 2 R)Γ (t) < 0, then there exists a scalar κ > 0 such that V̇1 (xt ) ≤
−κx T (t)x(t) < 0 for x(t) = 0, which ensures the asymptotic stability of the sliding
mode in (5.3). Fortunately, Ξ (t) + Γ T (t)(τ 2 R)Γ (t) < 0 is implied by (5.4). To
show that, by Lemma 2.1
Ω1 = Q + H F̃ (t)E + ET F̃ (t)T H T
where
⎡ ⎤
Ũ P BKτ + R P D̄ τ (A+BK)T R
⎢ ∗ −Q − R 0 τ (BKτ )T R ⎥
Q := ⎢
⎣∗
⎥
⎦
∗ −I τ D̄ T R
∗ ∗ ∗ −R
H := col{P Ē, 0, 0, τ R Ē}
E := [Ñ 0 0 0 ]
Ũ := P A+AT P +P BK +K T B T P +Q−R+μ2 I
By Lemma 2.2, Ω1 < 0 if and only if there exists a scalar ε > 0 such that
Q + ε−1 ET E + εH H T < 0
where
Υ11 = P A + AT P + P BK + K T B T P + Q − R,
Υ12 = P BKτ + R, Υ14 = τ AT R + τ K T B T R,
Υ22 = −Q − R, Υ24 = τ KτT B T R,
Υ34 = τ D̄ T R, Υ46 = τ R Ē, Υ66 = −ε−1 I.
Now, we proceed to design a robust sliding mode controller with mixed current and
delayed states such that the reachability of the specified sliding surface s1 (t) = 0 is
ensured.
Based on the obtained controller gain matrices K and Kτ in Theorem 5.1,
a robust sliding mode controller with mixed current and delayed states can be
designed as
where sgn(·) is the sign function, and the switching gain ρ1 (x(t)) is given by
s˙1 (t) = GM̃ F̃ (t)Ñx(t) + GDf (x, t) − [η + (η0 + μ GD) x(t)] sgn(s1 (t))
(5.9)
76 5 Delayed Integral Sliding Mode Control
which implies that under the controller (5.6), the sliding surface s1 (t) = 0 is
reachable in finite time and the state trajectory of system (2.44) maintains on it
thereafter. This completes the proof.
If there is no uncertainty in the system’s matrices, then the system (2.44) reduces
to (2.45). In this case, we can design a sliding mode controller with mixed current
and delayed states as
platform, one way to avoid chattering is to replace sgn(s1 (t)) with the follow-
ing sigmoid-like function to “smooth” the behaviors of the controller in (5.6)
and (5.11)
s1 (t)
χ (s1 (t)) = (5.14)
|s1 (t)| + δ
Notice that from (5.4) and (5.13) the matrix inequalities are nonlinear due to the
block element Ψ44 . To solve the gain matrices K and Kτ from (5.4) or (5.13),
we now present a numerical algorithm by employing the cone complementary
linearization approach [137]. Since inequalities (5.4) and (5.13) are of the same
structure, we only focus on the calculation of the gain matrices K and Kτ
from (5.13).
Introduce a new matrix S > 0 such that S ≤ P̄ R̄ −1 P̄ , which is equivalent to
S −1 P̄ −1
≥0 (5.15)
∗ R̄ −1
The problem formulated by the conditions (5.16), (5.17), and (5.18) is a non-
convex feasibility problem (NCFP) due to the equality constraints in (5.18). In
the sequel, we convert the NCFP formulated by (5.16), (5.17), and (5.18) to
the following nonlinear minimization problem subject to a set of linear matrix
inequalities.
78 5 Delayed Integral Sliding Mode Control
To solve the above NMP, we give an algorithm. For this goal, we introduce some
notations as
" #
Γ = P̄, Q̄, R̄, S̄, S, L̄, M̄, K̄τ , K̄
(0)
Γ (0) = P̄ (0) , Q̄(0) , R̄ (0) , S̄ (0) , S (0) , L̄(0) , M̄ (0) , K̄τ , K̄ (0)
is not satisfied within a specified number of iterations, then exit; otherwise, set
k = k + 1 and go to Step 2.
In this section, we will show through simulation the effectiveness of the sliding
mode control scheme with mixed current and delayed states for system (2.45) and
the robust sliding mode control scheme with mixed current and delayed states for
uncertain system (2.44), respectively.
An offshore platform with a TMD mechanism presented in Fig. 2.2 is simulated.
The parameters of the system and the waves are given in Table 4.1. With the setting
in the table, the nonlinear wave force f (x, t) can be computed as Appendix A of
[65], and the matrices A and B in (2.44) and (2.45) can be obtained as (4.34).
5.3 Simulation Results 79
We consider the sliding mode control scheme with mixed current state and delayed
state. Set G = [1000 10 1000 1 1000 1], μ = 0.8, η = 0.1, δ = 0.005, and
τ = 0.01 s. By Algorithm 1, an SMC-MCDS of the form (5.11) can be obtained,
and the gain matrices K and Kτ are given as
Under SMC-MCDS, the response curves of the three floors and the
control force are presented in Fig. 5.1. It can be computed that the oscillation
amplitudes peak to peak of the first, second, and third floors are 0.1868, 0.2041,
and 0.2182 m, respectively, and the required peak to peak control force is
1.3745 × 105 N.
Then, under SMC, it can be seen from [99] that the oscillation amplitudes of
the first, second, and third floors are 0.2192, 0.2301 and 0.2383 m peak to peak,
respectively, and the control force required is about 2.1565 ×105 N. The oscillation
amplitudes of the three floors and the control force under SMC-MCDS and SMC
are listed in Table 5.1, which indicates that compared with SMC, SMC-MCDS has
some significant advantages as follows.
• The oscillation amplitudes of the three floors under SMC-MCDS are less than
those under SMC.
• The control force required by SMC-MCDS is smaller than the one by SMC. In
fact, the control force required by SMC is about 1.57 times as that by SMC-
MCDS.
80 5 Delayed Integral Sliding Mode Control
0.2 0.2
Response of floor 1 (m)
0 0
−0.1 −0.1
−0.2 −0.2
0 20 40 60 80 100 0 20 40 60 80 100
Time (s) Time (s)
5
x 10
0.2 1
Response of floor 3 (m)
0 0
−0.1 −0.5
−0.2 −1
0 20 40 60 80 100 0 20 40 60 80 100
Time (s) Time (s)
Fig. 5.1 Responses of the three floors of the system and the required control force under SMC-
MCDS
Table 5.1 The maximum oscillation amplitudes of three floors and control force under SMC and
SMC-MCDS
Controllers Floor 1 (m) Floor 2 (m) Floor 3 (m) Control force (105 N)
SMC [99] 0.2192 0.2301 0.2383 2.1565
SMC-MCDS 0.1868 0.2041 0.2182 1.3745
Table 5.2 The oscillation amplitudes of three floors and control force under SMC-MCDS for
different time-delays
Time-delay (s) Floor 1 (m) Floor 2 (m) Floor 3 (m) Control force (105 N)
0.01 0.1868 0.2041 0.2182 1.3745
0.07 0.1857 0.2035 0.2183 1.3605
0.13 0.1846 0.2029 0.2198 1.3655
0.15 0.1895 0.2016 0.2288 2.0296
0.16 0.2592 0.2241 0.2833 7.3935
0.42 0.3135 0.2579 0.3047 5.4760
0.43 0.1883 0.2037 0.2255 1.4739
0.48 0.1819 0.2024 0.2224 1.3781
0.54 0.1918 0.2103 0.2260 1.5044
0.56 0.2023 0.2201 0.2335 1.7694
0.57 0.2141 0.2322 0.2445 2.0809
0.60 0.3084 0.3246 0.3265 4.8688
0.25 0.26
0.24 SMC SMC
Amplitudes of floor 1 (m)
2.5
Amplitudes of floor 3 (m)
2.2
0.24
1.9
0.23
0.22 1.6
0.21 1.3
0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.01 0.03 0.05 0.07 0.09 0.11 0.13 0.15
Time−delay (s) Time−delay (s)
Fig. 5.2 Oscillation amplitudes of the system and the control force under SMC and SMC-MCDS
as a time-delay is in the range of 0.01–0.15 s
82 5 Delayed Integral Sliding Mode Control
0.25 0.25
0.24 SMC SMC
Amplitudes of floor 1 (m)
5
x 10
0.26 3
SMC SMC
SMC−MCDS SMC−MCDS
0.25 2.5
0.24 2
0.23 1.5
0.22 1
0.43 0.45 0.47 0.49 0.51 0.53 0.56 0.43 0.45 0.47 0.49 0.51 0.53 0.56
Time−delay (s) Time−delay (s)
Fig. 5.3 Oscillation amplitudes of the system and the control force under SMC and SMC-MCDS
as a time-delay is in the range of 0.43–0.56 s
Table 5.3 The oscillation amplitudes of three floors and control force under NLC, DOFC,
DDOFC, and SMC-MCDS
Controllers Floor 1 (m) Floor 2 (m) Floor 3 (m) Control force (105 N)
NLC [65] 0.3050 0.3050 0.3050 2.0
DOFC [100] 0.2329 0.2543 0.2705 4.0
DDOFC [100] 0.2034 0.2232 0.2391 0.6
SMC-MCDS 0.1868 0.2041 0.2182 1.4
Now, we compare SMC-MCDS with NLC [65], DOFC, and DDOFC [100],
respectively. Table 5.3 presents the maximum oscillation amplitudes of the three
floors and the required control force under the controllers mentioned above.
On the one hand, it can be seen from Table 5.3 that under NLC and DOFC,
the average oscillation amplitudes of the three floors are 0.3050 and 0.2526 m,
respectively, while under the SMC-MCDS, the average vibration amplitude is
reduced to about 0.2030 m. Moreover, the required control force by NLC and DOFC
is 1.4 times and 2.9 times as that by SMC-MCDS, respectively. It shows that
compared with NLC and DOFC, SMC-MCDS is capable of taking better vibration
reduction of the offshore platform at a much less control force.
5.3 Simulation Results 83
On the other hand, it can also be seen from Table 5.3 that DDOFC requires less
control force leading to almost the same oscillation amplitudes of the three floors
as those under the SMC-MCDS. However, from the implementation point of view,
the DDOFC is difficult to be implemented because the introduced time-delay should
be less than 0.11 s [100], while the physical time-delay is usually larger than 0.14 s
[138]. Nevertheless, the SMC-MCDS provides a larger range from 0.43 to 0.56 s
for the introduced time-delay to be chosen, which can be seen from Table 5.2 and
Figs. 5.2 and 5.3.
This subsection will focus on designing a robust sliding mode controller with mixed
current and delayed states (RSMC-MCDS) for the uncertain system (2.44). Some
comparisons between the designed RSMC-MCDS and the delay-free robust sliding
mode controller (RSMC) [99] and the robust dynamic output feedback controller
(RDOFC) will be made from the controlled oscillation amplitudes of the offshore
platform and the size of control force required. Then, the effects of time-delays on
the robust sliding mode control for the offshore platform will be investigated.
Suppose that ξ̂i = 0.02 (i = 1, 2, T ), ĈT = 0.05, K̂T = 1. Then from (2.26),
one has ω̂T = 0.0003. From (2.34), it is clear that
0 0 0 0
Ē0 = , Ēi = , i = 1, 2
1 0.05 0.001 0.02
0 0 −6.6102 −0.5454
ĒT = , HT =
0.0003 0.02 0 −3.6360
−6.6102−0.0182 −236.24 −0.1087
H̄1 = , H̄2 =
0 −3.6360 0 −21.737
To begin with, when no controller is acted on the uncertain system (2.44), Fig. 5.4
depicts the response curves of the three floors of the system. The oscillation
amplitudes peak to peak of the first, second, and third floors of the system are
1.4284, 1.5479, and 1.6252 m, respectively.
Then, under RDOFC [100], it can be computed that the oscillation amplitudes
peak to peak of the three floors are 0.7315, 0.7991, and 0.8504 m, respectively, and
the required control force peak to peak is about 1.4166 ×106 N.
84 5 Delayed Integral Sliding Mode Control
Response of floor 1 (m)
0 0
−0.5 −0.5
0 20 40 60 80 100 0 20 40 60 80 100
Time (s) Time (s)
1
Response of floor 3 (m)
0.5
−0.5
−1
0 20 40 60 80 100
Time (s)
Fig. 5.4 Responses of the three floors of the uncertain system without control
Third, we consider the delay-free robust sliding mode control scheme. Set μ =
0.8, η = 0.1, and G = [1000 10 1000 1 1000 100]. By Proposition 1 in [99], an
RSMC can be obtained, whose gain K is given by
Under RSMC, the response curves of the three floors and the control force are
plotted in Fig. 5.5. From this figure, the oscillation amplitudes peak to peak of the
first, second, and third floors are 0.2251, 0.2445, and 0.2581 m, respectively, and
the control force required is about 1.5942 × 106 N.
Finally, we turn to the robust sliding mode control scheme with mixed current and
delayed states. Let τ = 0.01 s. Solve the matrix inequality (5.4) to get an RSMC-
MCDS of the form (5.6) with
K = −2258.4 344.3 6403.4 309.4 −422.8 −1965.2
Kτ = −386.6 2.51 181.5 321.6 154.9 −349.8
Under RSMC-MCDS, the response curves of the three floors and the control force
are depicted in Fig. 5.6. The oscillation amplitudes of the three floors and the control
force required are listed in Table 5.4. From the table, one can see that the oscillation
amplitudes of the three floors are greatly reduced from 1.4284, 1.5479, and 1.6252
5.3 Simulation Results 85
0.2 0.2
Response of floor 1 (m)
0 0
−0.1 −0.1
−0.2 −0.2
0 20 40 60 80 100 0 20 40 60 80 100
Time (s) Time (s)
6
x 10
0.2 1
Response of floor 3 (m)
0.1 0.5
−0.1 −0.5
−0.2 −1
0 20 40 60 80 100 0 20 40 60 80 100
Time (s) Time (s)
Fig. 5.5 Responses of the three floors of the uncertain system under RSMC
0.2 0.2
Response of floor 1 (m)
0.1 0.1
0 0
−0.1 −0.1
−0.2 −0.2
0 20 40 60 80 100 0 20 40 60 80 100
Time (s) Time (s)
6
x 10
0.2 1
Response of floor 3 (m)
0.1 0.5
Control force (N)
0 0
−0.1 −0.5
−0.2 −1
0 20 40 60 80 100 0 20 40 60 80 100
Time (s) Time (s)
Fig. 5.6 Responses of the three floors of the uncertain system under RSMC-MCDS
86 5 Delayed Integral Sliding Mode Control
Table 5.4 The oscillation amplitudes of three floors and control force under no control, RDOFC,
RSMC, and RSMC-MCDS
Controllers Floor 1 (m) Floor 2 (m) Floor 3 (m) Control force (106 N)
No Control 1.4284 1.5479 1.6252 –
RDOFC [100] 0.7315 0.7991 0.8504 1.4166
RSMC [99] 0.2251 0.2445 0.2581 1.5942
RSMC-MCDS 0.2237 0.2427 0.2572 1.0459
Table 5.5 The oscillation amplitudes of three floors and control force under RSMC-MCDS for
different time-delays τ
τ (s) Floor 1 (m) Floor 2 (m) Floor 3 (m) Control force (106 N)
0.01 0.2237 0.2427 0.2572 1.0459
0.03 0.2239 0.2436 0.2577 1.0410
0.05 0.2242 0.2438 0.2579 1.0415
0.07 0.2240 0.2438 0.2580 1.0403
0.08 0.2244 0.2441 0.2584 1.0623
0.10 0.2249 0.2443 0.2586 1.0623
0.20 0.2268 0.2453 0.2593 1.0711
0.40 0.2283 0.2315 0.2447 1.0552
0.60 0.2291 0.2483 0.2626 1.0653
0.80 0.2293 0.2483 0.2623 1.0644
To investigate the effects of time-delays on robust sliding mode control for the
offshore platform, for various values of the time-delay τ , we calculate the maximum
oscillation amplitudes of the three floors and the control force required under
RSMC-MCDS. The obtained results are listed in Table 5.5. From this table, one
can see that if the time-delay takes values in the range of 0.01–0.80 s, the oscillation
5.3 Simulation Results 87
0.231 0.249
RSMC RSMC
Amplitudes of floor 1 (m)
0.224 0.243
0.01 0.11 0.21 0.31 0.41 0.51 0.61 0.71 0.8 0.01 0.11 0.21 0.31 0.41 0.51 0.61 0.71 0.8
Time−delay (s) Time−delay (s)
6
x 10
0.264 2
0.258
1.2
0.257
0.256 1
0.01 0.11 0.21 0.31 0.41 0.51 0.61 0.71 0.8 0.01 0.11 0.21 0.31 0.41 0.51 0.61 0.71 0.8
Time−delay (s) Time−delay (s)
Fig. 5.7 Oscillation amplitudes of the system and the control force under RSMC and RSMC-
MCDS as a time-delay is in the range of 0.01–0.80 s
amplitudes of the three floors under RSMC-MCDS are almost at the same level
as those under RSMC, while the control forces required under RSMC-MCDS are
less than those under RSMC. Specially, if the time-delay τ changes from 0.01 to
0.07 s, both the oscillation amplitudes of the three floors and the control force under
RSMC-MCDS are less than those under RSMC, which is confirmed by Figs. 5.7
and 5.8. In summary, the simulation results show that, by introducing some proper
time-delays into the control channel, the offshore platform can work in a more safe
environment. The proposed sliding mode controller with mixed current and delayed
states has following remarkable characteristics.
• Compared with SMC (RSMC) and DOFC (RDOFC) [100], SMC-MCDS
(RSMC-MCDS) requires much less control force, and the oscillation amplitudes
of the three floors of the offshore platform under SMC-MCDS (RSMC-MCDS)
are less than those under SMC (RSMC) or DOFC (RDOFC).
• Compared with NLC [65] and DOFC [100], it is found that the oscillation
amplitudes of the three floors of the offshore platform under SMC-MCDS are
less than those under NLC or DOFC. Moreover, the control force required by
SMC-MCDS is smaller than that by NLC or DOFC.
• Compared with DDOFC [100], SMC-MCDS provides more options for a time-
delay to be chosen such that the offshore platform works in a more safe
environment.
88 5 Delayed Integral Sliding Mode Control
0.227 0.248
RSMC RSMC
Amplitudes of floor 1 (m)
0.224 0.244
0.243
0.223
0.242
0.222 0.241
0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.01 0.02 0.03 0.04 0.05 0.06 0.07
Time−delay (s) Time−delay (s)
6
x 10
0.261 2
0.256 1.2
0.255 1
0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.01 0.02 0.03 0.04 0.05 0.06 0.07
Time−delay (s) Time−delay (s)
Fig. 5.8 Oscillation amplitudes of the system and the control force under RSMC and RSMC-
MCDS as a time-delay is in the range of 0.01–0.07 s
Remark 5.3 If one wants to implement the method proposed in this chapter, the
generalized coordinates z1 and z2 and their derivatives can be derived from the
corresponding displacements and velocities in physical units. The relation between
the generalized coordinates and the physical units can be found in Appendix B in
[72].
Remark 5.4 It should be pointed out that in this chapter, all the current and delayed
states are available for feedback. If some states and delayed states are not available
for feedback, one can consider a state observer to design a suitable delayed sliding
mode controller to attenuate the wave-induced vibration of the offshore platform.
5.4 Conclusions
In this chapter, a general uncertain model of the offshore platform has been
established by taking the parameter perturbations of the offshore platform with the
active tuned mass damper. Based on this uncertain model, a novel integral sliding
mode control scheme with mixed current and delayed states has been proposed for
the offshore platform subject to nonlinear self-excited wave force and parameter
perturbations. Then the sliding mode controller design has been cast to a convex
5.5 Notes 89
optimization problem with linear matrix inequality constraints. The effects of time-
delays on sliding mode control for an offshore steel jacket platform have been
investigated. As a result, by properly introducing the time-delays into the control
channel, the obtained sliding mode controller can significantly reduce both internal
oscillations of the offshore platform and required control force, which have been
confirmed by simulation results.
5.5 Notes
The delayed feedback control scheme is first applied to the active control for
offshore platform system by Zhang, Han and Han [100], where a delayed dynamic
output feedback controller is designed, and the positive effects of introduced
time-delays on dynamic output feedback control for the system are investigated.
However, the designed time-delay is too small from the point of view of controller
implementation. To solve this problem, the delayed sliding mode controller with
mixed current and delayed states are provided in this chapter, which is mainly
taken from [101]. An in-depth study of such an idea using delayed sliding mode
control can be found in Nourisola and Ahmadi [102] and Nourisola, Ahmadi, and
Tavakoli [103], where the delayed adaptive sliding mode control scheme are studied.
In [107], by intensionally introducing a time-varying delay into the control channel,
a delayed robust sliding mode H∞ controller is designed to reduce vibration of the
offshore platform subject to self-excited nonlinear wave force, external disturbance
and parametric uncertainties. Simulation results show that the delayed robust sliding
mode H∞ controller is better than the traditional robust sliding mode controller and
robust sliding mode H∞ controller [90].
From the reducing vibration of offshore platforms and saving the control cost
point of view, delayed sliding mode control feedback control strategies are effective
and have some distinct advantages. However, it is a key point to choose proper
time-delays and thereby to guarantee the control performance of offshore platforms.
Consequently, a critical value, an optimal value and/or interval of time-delays with
positive effects on control performance of offshore platforms should be further
investigated. Another issue of delayed sliding mode control is that if some states and
delayed states are not available, necessary alternatives deserve deeper investigation.
Chapter 6
Delayed Robust Non-fragile H∞ Control
In this chapter, the offshore platform under consideration is shown in Fig. 2.2
[65, 72]. The dynamic model of the offshore platform subject to the self-excited
hydrodynamic forces and the external disturbance is given (2.47). It is assumed that
the external disturbance ζ (t) ∈ L2 [0, ∞).
The control output equation is given by (4.1). To design a delayed robust non-
fragile H∞ control law with mixed current and delayed states such that under the
designed control law, the system (2.47) with ζ (t) = 0 is robustly stable; and under
the zero initial condition, the H∞ performance
of the resulting closed-loop system is guaranteed for nonzero ζ (t) and a prescribed
γ > 0.
The delayed robust non-fragile control law is designed as
where V , U1 , and U2 are known matrices with appropriate dimensions and G(t) is
an unknown time-varying matrix satisfying
In this section, some sufficient conditions for the existence of the delayed robust
non-fragile H∞ controllers are developed.
Substituting Eq. (6.2) into Eq. (2.47) yields
where
The following proposition provides a sufficient condition for the existence of the
delayed robust non-fragile control law (6.2).
Proposition 6.1 Under Assumption 1 of Chap. 4, for given scalars γ > 0, μ > 0,
and h > 0, the closed-loop system (6.5) with ζ (t) = 0 is robustly stable, and the
H∞ performance (6.1) is guaranteed, if there exist 6 × 6 matrices P > 0, Q > 0,
R > 0, M1 , M2 , Z1 ≥ 0, Z2 , Z3 ≥ 0, 1 × 6 matrices K1 and K2 , and scalars ε > 0
and ! > 0 such that
⎡ ⎤
Ψ11 Ψ12 P D P D0 Ψ15T C1T μI εP BV U1T !P M̂ N̂ T
⎢ ∗ T U2T 0 ⎥
⎢ Ψ22 0 0 Ψ25 0 0 0 0 ⎥
⎢ ⎥
⎢ ∗ ∗ −I 0 DT 0 0 0 0 0 0 ⎥
⎢ ⎥
⎢ ∗ ∗ ∗ −γ 2 I D0T D1T 0 0 0 0 0 ⎥
⎢ ⎥
⎢ ∗ ∗ ∗ ∗ Ψ55 0 0 εBV 0 ! M̂ 0 ⎥
⎢ ⎥
⎢ ∗ ∗ ∗ ∗ ∗ −I 0 ⎥
⎢ 0 0 0 0 ⎥<0 (6.6)
⎢ ∗ ∗ ∗ ∗ ∗ ∗ −I 0 ⎥
⎢ 0 0 0 ⎥
⎢ ⎥
⎢ ∗ ∗ ∗ ∗ ∗ ∗ ∗ −εI 0 0 0 ⎥
⎢ ⎥
⎢ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ −εI 0 0 ⎥
⎢ ⎥
⎣ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ −!I 0 ⎦
∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ −!I
and
⎡ ⎤
R M1 M2
⎣ ∗ Z1 Z2 ⎦ ≥ 0 (6.7)
∗ ∗ Z3
where
⎧
⎨ Ψ11 = P A + AT P + P BK1 + K1T B T P + Q + M1 + M1T + hZ1
Ψ = P BK2 − M1T + M2 + hZ2 , Ψ15 = A + BK1 (6.8)
⎩ 12
Ψ22 = −Q − M2 − M2T + hZ3 , Ψ25 = BK2 , Ψ55 = −h−1 R −1
First, we consider the robust stability of closed-loop system (6.5) with ζ (t) = 0.
Taking the derivative of V (xt ) with respect to t along the trajectory of system (6.5)
yields
where
⎧ ⎡ ⎤
⎪
⎪ ∇11 (t) ∇12 (t) P D
⎨ ∇(t) = ⎣ ∗ Ψ22 0 ⎦
(6.13)
⎪
⎪ ∗ ∗ −I
⎩
Γ (t) = Γ11 (t) Γ12 (t) D
with
⎧
⎪
⎪ ∇11 (t) = Ψ11 + μ2 I + P M̂ F̂ (t)N̂ + N̂ T F̂ T (t)M̂ T P
⎪
⎪
⎪
⎨ +P BV G(t)U1 + U1T GT (t)V T B T P
∇12 (t) = Ψ12 + P BV G(t)U2 (6.14)
⎪
⎪
⎪ Γ11 (t) = Ψ15 + M̂ F̂ (t)N̂ + BV G(t)U1
⎪
⎪
⎩
Γ12 (t) = Ψ25 + BV G(t)U2
Then, a sufficient condition for robust stability of system (6.5) with ζ (t) = 0 is that
there exist 6 × 6 matrices P > 0, Q > 0, R > 0, M1 , M2 , Z1 ≥ 0, Z2 , Z3 ≥ 0,
1 × 6 matrices K1 and K2 such that (6.7) and
which can be derived from the matrix inequality (6.6). It indicates that if the matrix
inequalities (6.6) and (6.7) hold, then the closed-loop system (6.5) with ζ (t) = 0 is
robustly stable.
We now prove that the H∞ performance (6.1) is guaranteed for nonzero ζ (t)
and a prescribed γ > 0. Taking the derivative of V (xt ) with respect to t along the
trajectory of (6.5), and combing with (4.1), after simple manipulations, one yields
where
⎧ T
⎪
⎪ ς (t) = x T (t) x T (t − h) f T (x, t) ζ T (t)
⎪
⎪ ⎡ ⎤
⎪
⎪ ∇11 (t) ∇12 (t) P D P D0
⎪
⎪
⎪
⎨¯ ⎢ ∗ Ψ22 0 0 ⎥
∇(t) = ⎢⎣ ∗
⎥
∗ −I 0 ⎦ (6.18)
⎪
⎪
⎪
⎪ ∗ ∗ ∗ −γ 2 I
⎪
⎪
⎪
⎪ Υ = C1 0 0 D 1
⎪
⎩
Γ¯ (t) = Γ11 (t) Γ12 (t) D D0
If the matrix inequalities (6.6) and (6.7) hold, by using Schur complement and
S-procedure again, one yields
¯
∇(t) + Υ T Υ + Γ¯ T (t)(hR)Γ¯ (t) < 0 (6.19)
which leads to
Since V (x(0)) = 0 under zero initial condition, integrating both sides of Eq. (6.20)
from 0 to ∞, we have
∞
[ηT (t)η(t) − γ 2 ζ T (t)ζ (t)]dt < 0 (6.21)
0
which means that the H∞ performance (6.1) is guaranteed. This completes the
proof. Notice that there exist nonlinear terms in matrix inequality (6.6). We give an
equivalent form of Proposition 6.1, where the sufficient condition for the existence
of the delayed robust non-fragile H∞ control law is formulated in the form of a
linear matrix inequality.
Proposition 6.2 Under Assumption 1 of Chap. 4, for given scalars γ > 0, μ > 0,
and h > 0, the closed-loop system (6.5) with ζ (t) = 0 is robustly stable, and the
H∞ performance (6.1) is guaranteed, if there exist 6 × 6 matrices P̄ > 0, Q̄ > 0,
R̄ > 0, M̄1 , M̄2 , Z̄1 ≥ 0, Z̄2 , Z̄3 ≥ 0, 1 × 6 matrices K̄1 , K̄2 , scalars ε > 0 and
! > 0 such that
⎡ ⎤
Φ11 Φ12 D D0 Φ15 Φ16 μP̄ Φ18 Φ19 ! M̂ P̄ N̂ T
⎢ ∗ Φ 0 ⎥
⎢ 22 0 0 Φ25 0 0 0 Φ29 0 ⎥
⎢ ⎥
⎢ ∗ ∗ −I 0 hD T 0 0 0 0 0 0 ⎥
⎢ ⎥
⎢ ∗ ∗ ∗ −γ 2 I hD0T D1T 0 0 0 0 0 ⎥
⎢ ⎥
⎢ ∗ ∗ ∗ ∗ −hR̄ 0 0 Φ58 0 h! M̂ 0 ⎥
⎢ ⎥
⎢ ∗ ∗ ∗ ∗ ∗ −I 0 0 0 ⎥
⎢ 0 0 ⎥<0 (6.22)
⎢ ∗ ∗ ∗ ∗ ∗ ∗ −I ⎥
⎢ 0 0 0 0 ⎥
⎢ ⎥
⎢ ∗ ∗ ∗ ∗ ∗ ∗ ∗ −εI 0 0 0 ⎥
⎢ ⎥
⎢ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ −εI 0 0 ⎥
⎢ ⎥
⎣ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ −!I 0 ⎦
∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ −!I
and
⎡ ⎤
2P̄ − R̄ M̄1 M̄2
⎣ ∗ Z̄1 Z̄2 ⎦ ≥ 0 (6.23)
∗ ∗ Z̄3
where
⎧
⎪
⎪ Φ11 = AP̄ + P̄ AT + B K̄1 + K̄1T B T + Q̄ + M̄1T + M̄1 + hZ̄1
⎪
⎪
⎪
⎨ Φ12 = B K̄2 − M̄1T + M̄2 + hZ̄2 , Φ15 = hP̄ AT + hK̄1T B T
Φ16 = P̄ C1T , Φ18 = εBV , Φ19 = P̄ U1T (6.24)
⎪
⎪
⎪
⎪ Φ22 = −Q̄ − M̄2 − M̄2 + hZ̄3 , Φ25 = hK̄2T B T
T
⎪
⎩Φ
29 = P̄ U2T , Φ58 = hεBV
diag{P −1 , P −1 , I, I, hI, I, I, I, I, I, I }
P̄ = P −1 , Q̄ = P −1 QP −1 , R̄ = R −1
K̄i = Ki P −1 , M̄i = P −1 Mi P −1 , i = 1, 2
Z̄j = P −1 Zj P −1 , j = 1, 2, 3
Notice that P̄ R P̄ ≥ 2P̄ − R̄. It is clear that if the matrix inequality (6.23) holds,
then we have Eq. (6.25). This completes the proof.
As a special case of the control law (6.2), we present a delay-free robust non-
fragile H∞ control law as
A sufficient condition for the existence of the control law (6.26) is given by the
following corollary, which can be obtained from Proposition 6.2.
Corollary 6.1 Under Assumption 1 of Chap. 4, for given scalars μ > 0 and
γ > 0, the closed-loop system (6.27) with ζ (t) = 0 is robustly stable, and the
H∞ performance (6.1) is guaranteed, if there exist a 6 × 6 matrix P̄ > 0, a 1 × 6
matrix K̄1 , and scalars ε > 0 and ! > 0 such that
⎡ ⎤
Σ D D0 P̄ C1T μ̄P εBV P̄ U1T ! M̂ P̄ N̂ T
⎢ ∗ −I 0 0 0 0 0 0 0 ⎥
⎢ ⎥
⎢ ∗ ∗ −γ 2 I D T 0 ⎥
⎢ 1 0 0 0 0 ⎥
⎢∗ ∗ ∗ −I 0 0 ⎥
⎢ 0 0 0 ⎥
⎢ ⎥
⎢∗ ∗ ∗ ∗ −I 0 0 0 0 ⎥<0 (6.28)
⎢ ⎥
⎢∗ ∗ ∗ ∗ ∗ −εI 0 0 0 ⎥⎥
⎢
⎢∗ ∗ ∗ ∗ ∗ ∗ −εI 0 0 ⎥
⎢ ⎥
⎣∗ ∗ ∗ ∗ ∗ ∗ ∗ −!I 0 ⎦
∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ −!I
98 6 Delayed Robust Non-fragile H∞ Control
where
Moreover, the gain matrices K1 and K2 in Eq. (6.2) are given by Ki = K̄i P̄ −1 ,
i = 1, 2.
Correspondingly, if a delay-free non-fragile H∞ control law in the form (6.26)
is utilized to control the nominal system (2.50), the resulting closed-loop system is
given as
Corollary 6.3 Under Assumption 1 of Chap. 4, for given scalars μ > 0 and γ > 0,
the closed-loop system (6.31) with ζ (t) = 0 is asymptotically stable, and the H∞
performance (6.1) is guaranteed, if there exist a 6 × 6 matrix P̄ > 0, a 1 × 6 matrix
K̄1 , and a scalar ε > 0 such that
⎡ ⎤
Σ D D0 P̄ C1T μP̄ εBV P̄ U1T
⎢∗ −I 0 ⎥
⎢ 0 0 0 0 ⎥
⎢∗ ∗ −γ 2 I D1T ⎥
⎢ 0 0 0 ⎥
⎢ ⎥
⎢∗ ∗ ∗ −I 0 0 0 ⎥<0 (6.32)
⎢ ⎥
⎢∗ ∗ ∗ ∗ −I 0 0 ⎥
⎢ ⎥
⎣∗ ∗ ∗ ∗ ∗ −εI 0 ⎦
∗ ∗ ∗ ∗ ∗ ∗ −εI
Moreover, the gain matrices K1 and K2 in Eq. (6.33) are given by Ki = K̄i P̄ −1 ,
i = 1, 2.
100 6 Delayed Robust Non-fragile H∞ Control
The nonlinear self-excited wave force f (x, t) can be computed as [65]. The
external disturbance acting on the first and second modes are simulated by uniformly
distributed random signals with maximum amplitudes 4.6 × 105 and 1.1 × 105 N,
respectively.
Then, by Corollary 6.3, one can obtain a non-fragile H∞ controller (NFHC) of the
form (6.26), where the gain matrix K1 is given as
K1 = 105 × −1.1860 0.2374 2.7705 −0.0647 −0.4528 −0.9704 .
Under the obtained NFHC, the oscillation amplitudes of the first, second, and third
floors of the nominal system (2.50) are 0.2013, 0.2185, and 0.2318 m peak to peak,
respectively. The maximum control force required is about 3.2524×105 N. Depicted
in Fig. 6.1 are the responses of the first, second, and third floors of the offshore
platform and the required control force, respectively.
6.3 Simulation Results and Discussions 101
0.15 0.15
0.05 0.05
0 0
−0.05 −0.05
−0.1 −0.1
−0.15 −0.15
0 50 100 150 0 50 100 150
Time (s) Time (s)
(a) 5 (b)
x 10
0.15 3
0.05
1
0
0
−0.05
−0.1 −1
−0.15 −2
0 50 100 150 0 50 100 150
Time (s) Time (s)
(c) (d)
Fig. 6.1 Responses of the system (2.50) and the control force under NFHC
When the DNFHC is applied to control the nominal system (2.50), the response of
the three floors and the control force are showed in Fig. 6.2. It can be seen that the
oscillation amplitudes of the first, second, and third floors peak to peak are 0.1989,
0.2168, and 0.2309 m, respectively, and the range of the control force peak to peak
is about 1.7288 × 105 N.
The oscillation amplitudes of the three floors and the required control force are
listed in Table 6.1, where the oscillation amplitudes of the three floors of the system
without control are presented [90]. It can be seen from Table 6.1 that under the
NFHC and the DNFHC, the average oscillation amplitudes of the three floors of the
system are reduced to 14.3% and 14.2% of that without control, respectively. The
reduction of the vibration of the three floors under the two non-fragile controllers
are almost at the same level. However, it is clear to see that the control force by
102 6 Delayed Robust Non-fragile H∞ Control
0.15 0.15
0.1 0.1
0.05 0.05
0 0
−0.05 −0.05
−0.1 −0.1
−0.15 −0.15
0 50 100 150 0 50 100 150
Time (s) Time (s)
(a) 5 (b)
x 10
0.15 3
0.1
0.05
1
0
0
−0.05
−0.1 −1
−0.15 −2
0 50 100 150 0 50 100 150
Time (s) Time (s)
Fig. 6.2 Responses of the system (2.50) and the control force under DNFHC
Table 6.1 Maximum oscillation amplitudes of the nominal system (2.50) and the range of the
control force under different controllers
Controllers Floor 1 (m) Floor 2 (m) Floor 3 (m) Control force (N)
No control 1.4159 1.5270 1.6061 –
NFHC 0.2013 0.2185 0.2318 325,240
DNFHC 0.1989 0.2168 0.2309 172,880
the NFHC is 1.88 times as that by the DNFHC. This indicates that by intentionally
introducing a time-delay into the control channel, the required control force can be
reduced.
To investigate the effects of introduced time-delays on the non-fragile H∞
control for the system, we apply DNFHC to control the nominal system, and let
time-delay h take different values. Then we calculate the peak to peak oscillation
amplitudes of the three floors and the range of the required control force, which are
listed in Table 6.2, from which it can be seen that:
• Under the DNFHC, as h ≤ 46.0 s, the average reduction of the vibration of three
floors of the nominal system is the almost same as the one under the delay-free
NFHC, while the required control force by the DNFHC is much smaller than the
one by the NFHC, which shows that time-delays properly introduced can make
a positive contribution to the non-fragile H∞ control for the offshore steel jacket
platform.
6.3 Simulation Results and Discussions 103
Table 6.2 Maximum oscillation amplitudes of the nominal system (2.50) and the range of the
control force under the delayed non-fragile H∞ controller (DNFHC) for different values of time-
delay h
h (s) Floor 1 (m) Floor 2 (m) Floor 3 (m) Control force (N)
0.05 0.1989 0.2169 0.2310 173,020
0.10 0.1990 0.2169 0.2310 172,930
1.00 0.1983 0.2164 0.2307 171,850
5.00 0.1977 0.2158 0.2301 170,870
15.0 0.1983 0.2164 0.2307 171,890
25.0 0.1988 0.2168 0.2311 172,580
35.0 0.1989 0.2168 0.2309 173,200
46.0 0.2003 0.2191 0.2339 198,630
47.0 0.2035 0.2217 0.2353 419,500
48.0 0.2339 0.2551 0.2720 484,630
• Under the DNFHC, the allowable time-delay can be obtained as 46.0 s. Com-
pared with the delayed dynamic output feedback controller (DDOFC) [100],
where the upper bound of the introduced time-delay with positive effects is less
than 0.11 s, the DNFHC can provide more options of the time-delay to improve
the control performance of the offshore platform.
Under the RNFHC, the responses of the three floors and the required control
force are showed in Fig. 6.3. It can be obtained that the peak-to-peak oscillation
amplitudes of the first, second, and third floors are 0.2068, 0.2237, and 0.2365 m,
respectively. The maximum control force is about 5.8260 × 105 N.
104 6 Delayed Robust Non-fragile H∞ Control
(a) (b)
0.15 0.15
0.1 0.1
0.05 0.05
0 0
−0.05 −0.05
−0.1 −0.1
−0.15 −0.15
0 50 100 150 0 50 100 150
Time (s) Time (s)
(c) 5 (d)
x 10
0.15 4
0.1
2
0.05
0 0
−0.05
−2
−0.1
−0.15 −4
0 50 100 150 0 50 100 150
Time (s) Time (s)
Fig. 6.3 Responses of the system (2.47) and the control force under RNFHC
Now, let h = 0.03 s and U2 = 12I6 . Then we design a delayed robust non-fragile
H∞ controller (DRNFHC) by Proposition 6.2, and the gain matrices K1 and K2 can
be computed as
K1 = 105 × −1.4732 0.3890 3.1569 −0.1310 −0.5498 −1.2545 ,
K2 = 102 × −4.9699 1.5278 3.4637 −0.8936 −2.1829 −4.6154 .
0.15 0.15
0.05 0.05
0 0
−0.05 −0.05
−0.1 −0.1
−0.15 −0.15
0 50 100 150 0 50 100 150
Time (s) Time (s)
(a) x 10
5 (b)
4
2
0.05
0 0
−0.05
−2
−0.1
−4
0 50 100 150 0 50 100 150
Time (s) Time (s)
(c) (d)
Fig. 6.4 Responses of the system (2.47) and the control force under DRNFHC
Table 6.3 Maximum oscillation amplitudes of the uncertain system (2.47) and the range of the
control force under different controllers
Controllers Floor 1(m) Floor 2(m) Floor 3(m) Control force(N)
No control 1.4180 1.5371 1.6181 –
RNFHC 0.2068 0.2237 0.2365 582,600
DRNFHC 0.2059 0.2235 0.2363 497,130
Let h = 0.2 s and μ = 0.8. Then, by Corollary 6.4, a delayed state feedback
controller (DSFC) in the form (6.33) is obtained, where
K1 = 104 × −0.2080 −0.0631 1.9832 0.4715 0.0738 −0.1780 ,
K2 = 7.3945 −0.6173 −37.3389 0.9779 3.0632 5.8249 .
When the obtained DSFC is applied to control the system (2.45), the responses
of the three floors of the system and the control force are presented in Fig. 6.5. It can
be computed that the oscillation amplitudes of the first, second, and third floors of
the system are reduced to 0.1873, 0.2054, and 0.2197 m, respectively, and the range
of the control force required is about 7.3460 × 104 N. Table 6.5 lists the range of the
control force and the oscillation amplitudes of the three floors of the system (2.45)
106 6 Delayed Robust Non-fragile H∞ Control
Table 6.4 Maximum oscillation amplitudes of the uncertain system (2.47) and the control force
under the delayed robust non-fragile H∞ controller (DRNFHC) for different values of time-delay h
h (s) Floor 1 (m) Floor 2 (m) Floor 3 (m) Control force (N)
0.05 0.2059 0.2235 0.2363 497,720
0.10 0.2059 0.2235 0.2363 497,140
0.50 0.2059 0.2235 0.2364 498,000
1.00 0.2059 0.2235 0.2364 497,310
5.00 0.2059 0.2234 0.2363 497,330
10.0 0.2059 0.2235 0.2363 497,780
15.0 0.2059 0.2235 0.2364 498,100
20.0 0.2058 0.2234 0.2363 497,630
30.0 0.2066 0.2234 0.2363 496,810
35.0 0.2059 0.2235 0.2364 498,270
40.0 0.2075 0.2235 0.2363 497,520
46.0 0.2059 0.2235 0.2364 559,480
47.0 0.2097 0.2261 0.2370 687,110
48.0 0.2058 0.2234 0.2363 675,460
49.0 0.3845 0.4269 0.4639 887,120
0.15 0.15
Second Floor (m)
0.1 0.1
First Floor (m)
0.05 0.05
0 0
−0.05 −0.05
−0.1 −0.1
−0.15 −0.15
0 50 100 150 0 50 100 150
Time (s) Time (s)
(a) 5 (b)
x 10
0.15 1.5
Control Force (N)
0.1 1
Third Floor (m)
0.05
0.5
0
0
−0.05
−0.1 −0.5
−0.15 −1
0 50 100 150 0 50 100 150
Time (s) Time (s)
(c) (d)
Fig. 6.5 Responses of the system (2.45) and the control force under DSFC
under the different controllers: nonlinear controller (NLC) [65], DDOFC [100], and
integral sliding mode controller (ISMC) [99]. From Table 6.5, one observes that:
6.5 Notes 107
Table 6.5 Maximum oscillation amplitudes of the system (2.45) and the range of the control
force with different controllers
Controllers Floor 1 (m) Floor 2 (m) Floor 3 (m) Control force (N)
NLC [65] 0.3050 0.3050 0.3050 200,000
DDOFC [100] 0.2329 0.2543 0.2705 400,000
SMC [99] 0.2192 0.2301 0.2383 215,700
DSFC 0.1873 0.2054 0.2197 73,460
• The oscillation amplitudes of the three floors of the offshore platform under
DSMC are much smaller than the ones under NLC, DDOFC, and ISMC.
Moreover, the range of the control force required by DSMC is much less than
the one by NLC, DDOFC and ISMC.
• It can be computed that under the obtained DSFC, the maximum allowable time-
delay with positive effects on the control performance of the system (2.45) can
reach 60 s, which is much larger than the one under DDOFC.
6.4 Conclusions
This chapter has developed a delayed non-fragile H∞ control scheme for the
offshore platform and investigated the positive effects of time-delays on the non-
fragile H∞ control for the system. The simulation results have shown that by
properly introduced time-delay into control channel, the control force by the delayed
non-fragile H∞ control scheme can be significantly reduced. Moreover, the obtained
allowable maximum time-delay with the delayed non-fragile H∞ control scheme
has been much larger than the one with delayed dynamic output feedback control
scheme.
6.5 Notes
The main results of this chapter are taken from [91], where the non-fragile H∞
control scheme is utilized to deal with perturbation of controller gain and external
disturbance, and both current and delayed states are used to design the active
controller for an offshore platform with TMD mechanisms. In Zhang, Han, and
Huang [108], a pure delayed non-fragile controller is proposed to stabilize the
offshore platform subject to self-excited wave force, and the positive effects of
time-delays on the performance of the system are further investigated. It should be
pointed out that in [108] only pure delayed sates are used as feedback signals. Such
an idea is also applied to an offshore steel jacket platform with AMD mechanisms,
and a pure delayed H∞ controller is designed to attenuate the vibration induced by
external wave force. As for the detailed discussion of the controller, one refers to
Zhang and Tang [89].
108 6 Delayed Robust Non-fragile H∞ Control
Notice that intensionally introducing proper time-delays can improve the control
performance of offshore platform systems. However, some challenging issues need
to be considered. For example, delayed feedback control schemes are mainly based
on simplified dynamic models of offshore steel jacket platforms, where only the
dominant vibration modes are considered, while other higher vibration modes are
ignored. To develop general dynamic models for offshore platforms and then to
design effective delayed feedback control schemes are still interesting topics for
active control of offshore platforms. In addition, the obtained results are mainly
based on theoretical analysis and simulations. From the implementation point of
view, how to utilize these theoretical results in practical offshore platforms should
be reevaluated.
Chapter 7
Delayed Dynamic Output Feedback
Control
Consider a simple offshore steel jacket platform in Fig. 2.2 [65, 72], the dynamic
model of the offshore platform is given by (2.45).
where C ∈ Rr×6 is a constant matrix. The corresponding state space model of the
offshore platform can be expressed as
⎧
⎪
⎪ ẋ(t) = Ax(t) + Bu(t) + Df (x, t)
⎨
y(t) = Cx(t) (7.2)
⎪
⎪
⎩
x(t0 ) = x0 , t0 ≥ 0
Moreover, if (7.6), (7.7) and (7.8) are feasible on matrix variables X and Y , then the
controller parameters K defined in (7.5) can be obtained by solving the following
LMI:
where
−1
y I I X
P := , MN T = I − XY (7.10)
NT 0 0 MT
which leads to
Clearly, if Φ < 0, then V̇ (ϑ(t)) ≤ −λmin (−Φ)ξ T (t)ξ(t) ≤ −λmin (−Φ)ϑ T (t)
ϑ(t) < 0 for ϑ(t) = 0, which guarantees the asymptotic stability of system (7.4).
In order to solve the controller parameters, we rewrite Φ < 0 as
where
P A0 + AT0 P + μ2 E T E P E T D
Φ0 :=
∗ −I
P 0 H LT
Σ := , Π := , Λ := .
0 I 0 0
By Lemma 2.4, the inequality (7.14) is feasible for the matrix variable K if and only
if
where M, N ∈ R6×6 and the symbol “%” denotes an irrelevant matrix. Choosing
⎡ ⎤ ⎡ ⎤
W1 0 W2 0
Π⊥ = ⎣ 0 0⎦ , Λ⊥ = ⎣ 0 0⎦
0 I 0 I
we have
−1 −1 P −1 0 P A0 + AT0 P + μ2 E T E P E T D P −1 0
Σ Φ0 Σ =
0 I ∗ −I 0 I
A0 P −1 + P −1 AT0 + μ2 P −1 E T EP −1 E T D
=
∗ −I
7.1 Dynamic Output Feedback Control 113
⎡ ⎤
AX + XAT + μ2 XX AM + μ2 XM D
=⎣ ∗ μ2 M T M 0 ⎦
∗ ∗ −I
It is clear that the inequality (7.16) is equivalent to LMI (7.7). On the other hand,
by Lemma 2.5, there exists a P > 0 satisfying (7.17) if and only if x − y −1 ≥ 0,
which is equivalent to (7.8). This completes the proof.
Now, based on Proposition 7.1, we design a dynamic output feedback con-
troller for system (2.45) by using the output information of the generalized
coordinates of vibrational modes 1 and 2 and the horizontal displacement of the
TMD. Let
⎡ ⎤
100000
C = ⎣0 0 1 0 0 0⎦ . (7.18)
000010
The parameters of the offshore platform system (2.45) are given as Table 4.1.
Applying Proposition 7.1, for μ = 0.8, the obtained dynamic output feed-
back controller, which is denoted by DOFC1, is given in (7.3) with AK , BK ,
CK , DK as
114 7 Delayed Dynamic Output Feedback Control
⎡ ⎤
−0.1245 0.0650 0.0148 0.0058 0.0226 −0.0013
⎢−0.0777 −0.4729 −0.0385 0.1210 0.1478 −0.0034⎥
⎢ ⎥
⎢ ⎥
⎢ 0.7365 1.1961 0.3529 −0.4959 −0.4414 0.0076 ⎥
AK := 10 × ⎢
5
⎥
⎢ 4.9878 0.5868 2.6570 −1.9726 −0.6940 −0.0126⎥
⎢ ⎥
⎣−3.6257 0.1269 −1.8758 0.9852 −0.1391 0.0202 ⎦
−5.7661 −2.5562 −0.0695 3.8665 5.9187 −0.1423
⎡ ⎤
0.0004 −0.0119 0.0023
⎢−0.0652 −0.0085 0.0231 ⎥
⎢ ⎥
⎢ ⎥
⎢ 0.1581 −0.1373 −0.0748 ⎥
BK := 106 × ⎢ ⎥
⎢−0.0215 −1.6387 −0.1712⎥
⎢ ⎥
⎣ 0.1560 1.2877 0.0349 ⎦
−1.1523 −1.0792 0.8028
CK := 108 × −3.3396 2.7054 0.4868 −0.0658 0.3546 −0.0310
DK := 108 × 1.3256 −3.1587 0.2229
Under DOFC1, the responses of the first, the second, and the third floors are
depicted in Figs. 7.1, 7.2 and 7.3, from which one can see that the controlled
responses of the first, the second, and the third floors with oscillation magnitudes
of peak to peak have been reduced from 1.3793, 1.4946, and 1.5471 m to 0.2329,
0.2543, and 0.2705 m, respectively.
Notice that the Euclidean norm of the gain matrix A K BK
CK DK of the DOFC1 is
calculated as 5.5340 × 108 . One can see that the DOFC1 is of a large gain in the
sense of Euclidean norm, which demands a large control force.
0.15
0.1
0.05
FirstFloor Response (m)
−0.05
−0.1
−0.15
−0.2
0 20 40 60 80 100 120 140 160 180
Time (s)
Fig. 7.1 Response of the first floor under the control of DOFC1
7.2 Design of a Delayed Dynamic Output Feedback Controller 115
0.15
0.1
SeondFloor Response (m)
0.05
−0.05
−0.1
−0.15
−0.2
0 20 40 60 80 100 120 140 160 180
Time (s)
Fig. 7.2 Response of the second floor under the control of DOFC1
0.15
0.1
0.05
ThirdFloor Response (m)
−0.05
−0.1
−0.15
−0.2
0 20 40 60 80 100 120 140 160 180
Time (s)
Fig. 7.3 Response of the third floor under the control of DOFC1
We now design a dynamic output feedback controller of form (7.3) based on the
system (7.20). The resulting closed-loop system is
where
Let
Then
t
V̇ (t, ϑt ) = ηT (t)[Ψ1 + h2 Γ T RΓ ]η(t) − h ϑ̇ T (θ )E T RE ϑ̇(θ )dθ (7.26)
t−h
where
⎡ ⎤
ψ11 P H KL2 P E T D
Ψ1 := ⎣ ∗ −Q 0 ⎦
∗ ∗ 0
with
where
⎡ 2 T ⎤
μ E E − E T RE E T R 0
Ψ2 = ⎣ ∗ −R 0 ⎦
∗ ∗ −I
where Ψ0 := Ψ1 + Ψ2 , i.e.
⎡ ⎤
Θ11 P H KL2 + E T R P E T D
Ψ0 = ⎣ ∗ −Q − R 0 ⎦
∗ ∗ −I
Ψ0 + h2 Γ T RΓ < 0 (7.31)
7.2 Design of a Delayed Dynamic Output Feedback Controller 119
where
One can see that matrix inequality (7.32) is still nonlinear on matrix variables,
which is a non-convex feasible problem. Now, we will convert this non-convex
feasible problem into a nonlinear minimization problem subject to a set of LMIs.
Define J := diag{I, X}. Then pre- and post-multiplying both sides of Ω in (7.32)
by J T and its transpose, respectively, yields
Ω11 W2T XRX
J T ΩJ = <0 (7.34)
∗ −XQX − XRX
Introducing two new matrix variables S > 0 and Z > 0 such that
XRX ≥ S, XQX ≥ Z
120 7 Delayed Dynamic Output Feedback Control
Notice that
−W2T XRXW2 W2T XRX W2T
=− XRX W2 −I
∗ −XRX −I
W2T
≤− S W2 −I
−I
−W2T SW2 W2T S
= .
∗ −S
where
where
Proposition 7.4 Let W1 and [W2T W3T ]T be the orthogonal complements of C T and
[B T B T ]T , respectively. For given scalars μ > 0 and h > 0, the dynamic output
feedback control problem for system (7.20) is solvable if there exist 6 × 6 matrices
X > 0, Y > 0, Q > 0, R > 0, R̄ > 0, X̄ > 0, S > 0, S̄ > 0, Z > 0, Z̄ > 0,
and Q̄ > 0 such that (7.8), (7.33) and
⎡ ⎤
Υ W T S (W2T + W3T )D W2T X μW2T X
⎢ ∗ −Z2− S 0 ⎥
⎢ 0 0 ⎥
⎢ ⎥
⎢∗ ∗ −I 0 0 ⎥<0 (7.37)
⎢ ⎥
⎣∗ ∗ ∗ −Q̄ 0 ⎦
∗ ∗ ∗ ∗ −I
R X̄ Q X̄
≥ 0, ≥0 (7.38)
X̄ S̄ X̄ Z̄
R R̄ = I, XX̄ = I, S S̄ = I, Z Z̄ = I, QQ̄ = I (7.39)
where
Υ := W2T (AX + XAT − S)W2 + W2T XAT W3 + W3T AXW2 − h−2 W3T R̄W3 .
¯
Minimize Tr(X̄X + R̄R + Q̄Q + Z̄Z + SS) (7.40)
Subject to (7.8), (7.33), (7.37), (7.38) and
R I X I Q I
≥ 0, ≥ 0, ≥ 0, (7.41)
I R̄ I X̄ I Q̄
Z I SI
≥ 0, ≥ 0. (7.42)
I Z̄ I S̄
The following iterative algorithm can be used to solve the above NMP.
122 7 Delayed Dynamic Output Feedback Control
where ε is a prescribed sufficiently small positive scalar, are satisfied, then set
l = l + 1, and go to Step 2. If one of the conditions (7.32) and (7.43) is not
satisfied within a specified number of iterations, then exit.
Finally, if the nonlinear minimization problem (7.40) is feasible on the matrix
variables R, Q, and so on, then the desired dynamic output feedback controller of
form (7.3) can be obtained by solving the LMI (7.22) on the matrix variable K in
(7.5) with the known R, Q, and P of form (7.10).
⎤
⎡
0.0075 −0.0002 0.0007
⎢ 0.0364 0.0005 0.0032 ⎥
⎢ ⎥
⎢ ⎥
⎢−23.4249 −43.5381 0.3757 ⎥
BK = ⎢ ⎥
⎢−20.5615 −43.6538 0.6321 ⎥
⎢ ⎥
⎣ 76.6976 37.5001 4.6865 ⎦
131.2817 −384.3427 33.3547
CK = 104 × −7.2556 0.4957 1.3130 −1.4967 −0.0622 −0.1302
DK = 2.8932 −0.0098 0.2151
Under DOFC2, the responses of the first, the second, and the third floors are
shown in Figs. 7.4, 7.5 and 7.6, from which one can see that the controlled responses
of the first, the second, and the third floors with oscillation magnitudes of peak to
peak have been reduced from 1.3793, 1.4946, and 1.5471 m to 0.2034, 0.2232, and
0.2391 m, respectively.
In this section, based on the simulation results in Sects. 7.2 and 7.3, we make a
comparison between DOFC1 and DOFC2. Then, we investigate the effect of a time-
delay on the dynamic output feedback control of the offshore platform.
Firstly, we compare controller gains of controllers DOFC1 and DOFC2 in the
sense of the Euclidean norm. We calculate the Euclidean norms of the gain matrix
0.15
0.1
FirstFloor Response (m)
0.05
−0.05
−0.1
0 20 40 60 80 100 120 140 160 180
Time (s)
Fig. 7.4 Response of the first floor under the control of DOFC2 with h = 0.02
124 7 Delayed Dynamic Output Feedback Control
0.15
0.1
SecondFloor Response (m)
0.05
−0.05
−0.1
0 20 40 60 80 100 120 140 160 180
Time (s)
Fig. 7.5 Response of the second floor under the control of DOFC2 with h = 0.02
0.15
0.1
0.05
ThirdFloor Response (m)
−0.05
−0.1
−0.15
−0.2
0 20 40 60 80 100 120 140 160 180
Time (s)
Fig. 7.6 Response of the third floor under the control of DOFC2 with h = 0.02
AK BK
CK DK of controllers DOFC1 and DOFC2 as 5.5340×108 and 7.5415×104 ,
respectively. One can see clearly that the Euclidean norm of the gain of the controller
DOFC1 is about 7338 times the Euclidean norm of the gain of the controller
DOFC2, which means that a much larger control force is needed by the controller
DOFC1. Thus, we can conclude that, for the offshore steel platform, in order to
reduce the oscillation amplitudes of peak to peak of three floors to a certain range,
the introduction of a small time-delay into the output measurement can significantly
decrease the demanded control force, which is demonstrated by Figs. 7.7 and 7.8.
7.3 Comparison Between Different Controllers 125
5
x 10
5
3
Control Forces
−1
0 20 40 60 80 100 120 140 160 180
Time (s)
4
x 10
12
10
6
Control Forces
−2
−4
0 20 40 60 80 100 120 140 160 180
Time (s)
Table 7.1 The oscillation amplitudes of peak to peak of the three floors of the offshore platform
under no control, the control of DOFC1, and the control of DOFC2
Controllers Floor 1 (m) Floor 2 (m) Floor 3 (m)
No control 1.3793 1.4946 1.5471
DOFC1 0.2329 0.2543 0.2705
DOFC2 0.2034 0.2232 0.2391
Table 7.2 The oscillation h (s) Floor 1 (m) Floor 2 (m) Floor 3 (m)
amplitudes of peak to peak of
the three floors of the offshore 0.001 0.2037 0.2238 0.2400
platform under the control of 0.02 0.2034 0.2232 0.2391
DOFC2 for different values 0.04 0.2035 0.2229 0.2382
of time-delay h 0.08 0.2055 0.2238 0.2371
0.10 0.2076 0.2245 0.2394
0.10035 0.2365 0.2259 0.2674
0.11 1.2841 0.5904 1.5477
0.12 1.9355 0.8169 2.3421
0.2 5.1043 2.3301 5.7420
0.8
0.6
0.4
FirstFloor Response (m)
0.2
−0.2
−0.4
−0.6
−0.8
0 20 40 60 80 100 120 140 160 180
Time (s)
Fig. 7.9 Response of the first floor under the control of DOFC2 with h = 0.11
0.3
0.2
SecondFloor Response (m)
0.1
−0.1
−0.2
−0.3
−0.4
0 20 40 60 80 100 120 140 160 180
Time (s)
Fig. 7.10 Response of the second floor under the control of DOFC2 with h = 0.11
control. In this case, the controller will make the offshore platform work in an
unsafe environment, which is not expected.
In summary, it is shown that, when intentionally introducing a proper small
time-delay, one can design a dynamic output feedback controller such that (i) the
controller is of a small gain in the sense of Euclidean norm and (ii) the internal
oscillations of the offshore platform can be dramatically reduced.
128 7 Delayed Dynamic Output Feedback Control
0.8
0.6
0.4
ThirdFloor Response (m)
0.2
−0.2
−0.4
−0.6
−0.8
0 20 40 60 80 100 120 140 160 180
Time (s)
Fig. 7.11 Response of the third floor under the control of DOFC2 with h = 0.11
7.4 Conclusions
This chapter has investigated the effects of a time-delay on dynamic output feedback
control of an offshore platform subject to a nonlinear wave-induced self-excited
hydrodynamic force. It is found that by intentionally introducing a time-delay into
the output channel, the controller exhibits a small gain in the sense of Euclidean
norm, which in turn demands a correspondingly small control force. The internal
oscillation of the offshore platform is thereby dramatically reduced.
7.5 Notes
This chapter focuses on designing a delayed feedback controller for the offshore
platform by only using the measurable output signals, and the main results, to
name a few, are obtained by Zhang and Han [100]. In Nourisola et al. [103], an
adaptive output feedback integral sliding mode controller and a delayed adaptive
output feedback integral sliding mode controller have been designed for the offshore
platform. It should be mentioned that in [100] and [103], the delayed dynamic output
feedback controller and delayed adaptive output feedback sliding mode controller
are compared to the delay-free dynamic output feedback controller and delay-free
adaptive output feedback sliding mode controller, respectively. The positive effects
of artificially introduced time-delays on the active control for the offshore platform
are analyzed. Simulation results show that purposefully introduced proper time-
delays can reduce the vibration amplitudes of the system and the control cost as
well.
7.5 Notes 129
Note that the above output feedback controllers in [100, 103] and the state
feedback controllers provided by previous chapters are mainly designed in the
continuous-time domain. When a continuous-time controller for an offshore plat-
form is implemented in practice, control signals are usually transmitted in a digital
form, which results in a sampled-data system. More recently, Several sampled-data
control schemes are introduced to the area of active control for offshore platforms.
Comprehensive coverage on this subject can be found in Sakthivel et al. [109, 110],
Sivaranjani et al. [111], Zhang et al. [112], and Huang, Cai, and Xiang [113]. It is
observed that the sampled-data control schemes can reduce wave-induced vibration
of the platform significantly. Moreover, compared to continuous-time controllers,
sampled-data controllers may take less control cost. In fact, the above sampled-
data control schemes are all based on the input delay approach. In this sense,
such control schemes can also be classified as a delayed feedback control strategy,
and consequently, sampled-data control is a feasible and effective alternative
for the active control for the offshore platform. Specifically, to explore output
feedback or observer-based sampled-data control scheme for offshore platforms is
of significance both in theory and in real implementations, which is well worth
investigation in the near future.
Chapter 8
Network-Based Modeling and Active
Control
The network-based modeling and active control for an offshore steel jacket platform
with an active tuned mass damper mechanism is investigated in this chapter. A
network-based state feedback control scheme is developed. Under this scheme, the
corresponding closed-loop system is modeled by a system with an artificial interval
time-varying delay. Then, a delay-dependent stability criterion for the corresponding
closed-loop system is derived. Based on this stability criterion, a sufficient condition
on the existence of the network-based controller is obtained. It is found through
simulation results that (i) both the oscillation amplitudes of the offshore platform
and the required control force under the network-based state feedback controller
are smaller than those under the nonlinear controller [65] and the dynamic output
feedback controller [100]; (ii) the oscillation amplitudes of the offshore steel jacket
platform under the network-based feedback controller are almost the same as the
ones under the integral sliding mode controller [99], while the required control force
by the former is smaller than the one by the latter.
The rest of this chapter is organized as follows. A network-based dynamic
model of the offshore steel jacket platform is developed in Sect. 8.1. Section 8.2
is devoted to design a network-based state feedback control scheme and present
a stability criterion for the network-based offshore platform. Section 8.3 provides
some simulation results to illustrate the effectiveness of the developed control
strategies. Finally, Sect. 8.4 presents the conclusion of this chapter, and Sect. 8.5
provides a brief note.
An offshore steel jacket platform model with an active TMD mechanism shown in
Fig. 2.2 [65, 72] is considered for controller design. The dynamic equation is given
by (2.45). Let g(x, t) := Df (x, t). Then, the system (2.45) can be expressed as
Assumption 1 The sensor is clock-driven, the controller and actuator are event-
driven.
Assumption 2 The full state variables are available for measurement, and the state
measurements are transmitted with a single packet, and no packet dropout occurs in
the transmission.
Assumption 3 The state signal x(t) is sampled as x(lh) every other h seconds,
the signal x(lh) and its time stamp l are encapsulated as a sampled-data packet
(l, x(lh)) for transmission, l = 1, 2, · · · , and h > 0 is the sampling period.
To design a network-based state feedback control law u(t) as
where K is a 1×6 gain matrix to be determined, h > 0 denotes the sampling period,
and τk = τksc + τkca is the network-induced delay, which denotes the time from the
instant kh when sensor nodes sample sensor data from a plant to the instant when
actuator transfer data to the plant.
For simplicity, denote
Notice that
where
τ1 = τm , τ2 = τM + h (8.8)
Then, from (8.3), (8.4), (8.5), and (8.6), one yields the closed-loop system
ẋ(t) = Ax(t) + BKx(t − d(t)) + g(x, t), x(0) = x0
(8.9)
t ∈ [kh + τk , (k + 1)h + τk+1 ), k ∈ N
134 8 Network-Based Modeling and Active Control
In this section, a stability criterion is derived for the network-based offshore steel
jacket platform system (8.9). Then, based on the stability criterion, a sufficient
condition on the existence of the state feedback controller (8.3) is obtained.
For simplicity, denote
Further, denote
⎧
⎪
⎪ E1 := [I 0 0 0 0 0], E2 := [0 I 0 0 0 0]
⎨
E3 := [0 0 I 0 0 0], E4 := [0 0 0 I 0 0]
(8.13)
⎪ E5 := [0 0 0 0
⎪ I 0], E6 := [0 0 0 0 0 I ]
⎩
Eij := Ei − Ej , i, j = 1, · · · , 6
Notice
⎧
⎨ x(t) = E1 η(t), x(t − d(t)) = E2 η(t)
x(t − τ1 ) = E3 η(t), x(t − τ2 ) = E4 η(t) (8.14)
⎩
x(t − ρ(t)) = E5 η(t), g(x, t) = E6 η(t)
Then, based on (8.11) and (8.14), the closed-loop system (8.9) can be written as
Proposition 8.1 Given scalars τ1 and τ2 with τ2 > τ1 ≥ 0, μ > 0, the closed-loop
system (8.9) is asymptotically stable for d(t) satisfying (8.7) if there exist 6 × 6 real
matrices P > 0, Q1 > 0, Q2 > 0, R1 > 0, R2 > 0, S1 > 0, S2 > 0, Z, 6 × 36
matrix X, and 1 × 6 matrix K such that
⎡ ⎤
Λ11 + δΨ11 Λ12 + δΨ12 μE1T
⎣ ∗ −τ12 R1 − δ 2 R2 − δQ1 0 ⎦ < 0 (8.16)
∗ ∗ −I
⎡ ⎤
Λ11 Λ12 δXT μE1T
⎢ ∗ −τ 2 R1 − δ 2 R2 0 0 ⎥
⎢ 1 ⎥<0 (8.17)
⎣ ∗ ∗ −δQ1 0 ⎦
∗ ∗ ∗ −I
R2 Z
≥0 (8.18)
∗ R2
where δ = τ2 − τ1 and
4
V (xt , ẋt ) = Vj (xt , ẋt ) (8.22)
j =1
where
t
V4 (xt , ẋt ) =(τ2 − d(t)) ẋ (s)Q1 ẋ(s)ds + ψ (t)Q2 ψ(t)
T T
t−ρ(t)
with xt = x(t + θ ), θ ∈ [−τ2 , 0], P > 0, Si > 0, Ri > 0 and Qi > 0 (i = 1, 2),
and ψ(t) = x(t) − x(t − ρ(t)).
Note that for real matrices P > 0, Si > 0, Ri > 0, and Qi > 0 (i = 1, 2), on
the one hand, there exist !1 > 0 and !2 > 0 such that
where W is the space of functions xt (s) and ẋt (s) with the norm xt W as
In fact, due to the fact that in Lyapunov-Krasovskii functional (8.22), the terms
V1 , V2 , and V3 are absolutely continuous on [0, ∞]; in addition, V4 is absolutely
continuous for t = kh + τk and does not increase along {kh + τk , k ∈ N}
since it is nonnegative before {kh + τk , k ∈ N} and becomes zero just after these
points.
In what follows, to prove the asymptotic stability of system (8.9), we only need
to prove that V̇ (xt , ẋt ) < 0 holds for any x(t) = 0 with t = kh + τk , k ∈ N. For
this, taking the derivative of the Lyapunov-Krasovskii functional (8.22) along the
trajectory of system (8.9) and noticing the fact in (8.7) yields
4
V̇ (xt , ẋt ) = V̇j (xt , ẋt ) (8.26)
j =1
where
with
t
χ1 (t) = − τ1 ẋ T (s)R1 ẋ(s)ds (8.31)
t−τ1
t−τ1
χ2 (t) = − δ ẋ T (s)R2 ẋ(s)ds (8.32)
t−τ2
t
χ3 (t) = − ẋ T (s)Q1 ẋ(s)ds (8.33)
t−ρ(t)
Note that
T
E13 S1 E13 + E34
T
S2 E34 + E1T S1 E3 + E3T S1 E1 + E3T S2 E4
+ E4T S2 E3 − (E1T S1 E1 − E3T S1 E3 + E3T S2 E3 − E4T S2 E4 )
= 2E3T S1 E3 + 2E4T S2 E4 ≥ 0 (8.34)
Similarly, by applying Lemma 2.9, the following inequality is true for any 6 × 36
matrix X:
χ3 (t) ≤ ηT (t)[E51
T
X + XT E51 + ρ(t)XT Q−1
1 X]η(t) (8.38)
χ3 (t) ≤ ηT (t)[E51
T
X + XT E51 + (d(t) − τ1 )XT Q−1
1 X]η(t) (8.39)
Consequently, from (8.26) to (8.30), (8.35), (8.36), (8.37), (8.39), and (8.40), one
yields
where Λ11 and Λ12 are defined in (8.19) and (8.20), respectively, and
Applying the Schur Complement to (8.46) and (8.45), we arrive at (8.16) and (8.17),
respectively. The proof is completed.
Based on Proposition 8.1, a sufficient condition on the existence of the network-
based state feedback controller for the system (8.9) is proposed and stated as the
following proposition.
Proposition 8.2 Given scalars τ1 and τ2 with τ2 > τ1 ≥ 0, ν1 > 0, ν2 > 0, μ > 0,
and ϑ > 0, the offshore platform system (8.1) under the networked state feedback
controller (8.3) is asymptotically stable, if there exist 6 × 6 real matrices P̄ > 0,
Q̄1 > 0, Q̄2 > 0, R̄1 > 0, R̄2 > 0, S̄1 > 0, S̄2 > 0, Z̄, 6 × 36 matrix X̄, and 1 × 6
matrix K̄ such that
8.2 Stability Analysis and Network-Based Controller Design 139
⎡ ⎤
Λ̄11 + δ Ψ̄11 Λ̄12 μE1T P̄
⎣ ∗ −Ψ̄22 0 ⎦<0 (8.47)
∗ ∗ −I
⎡ ⎤
Λ̄11 Λ̄12 δ X̄T μE1T P̄
⎢ ∗ −Λ̄22 0 0 ⎥
⎢ ⎥<0 (8.48)
⎣ ∗ ∗ −δ Q̄1 0 ⎦
∗ ∗ ∗ −I
R̄2 Z̄
≥0 (8.49)
∗ R̄2
where
:= diag{P −1 , P −1 , P −1 , P −1 , P −1 , I }
Υ1 := diag{, (τ12 R1 + δ 2 R2 + δQ1 )−1 , I }
Υ2 := diag{, (τ12 R1 + δ 2 R2 )−1 , P −1 , I }
Υ3 := diag{P −1 , P −1 }
Then, let
P̄ = P −1 , K̄ = KP −1 , Q̄1 = P −1 Q1 P −1 , Q̄2 = ϑP −1
−1
X̄ = P X, −1
Ȳ = Q X, Z̄ = P −1 ZP −1
−1 −1 −1 −1
S̄i = P Si P , R̄i = P Ri P , i = 1, 2
140 8 Network-Based Modeling and Active Control
−P̄ (τ12 R̄1 + δ 2 R̄2 )−1 P̄ ≤ −2ν1 P̄ + ν12 (τ12 R̄1 + δ 2 R̄2 )
and
−P̄ (τ12 R̄1 + δ 2 R̄2 + δ Q̄1 )−1 P̄ ≤ −2ν2 P̄ + ν22 (τ12 R̄1 + δ 2 R̄2 + δ Q̄1 )
we arrive at (8.47), (8.48), and (8.49), respectively. This completes the proof.
In this subsection, both the constant network-induced delay and the time-varying
network-induced delay are considered, and several network-based state feedback
controllers (NSFCs) will be designed, and the performance of the offshore platform
under the NSFCs will be presented.
For comparison, we first give the performance of the offshore platform without
control. As stated in [99], when no controller is utilized to control the offshore
platform, the oscillation amplitudes of the first, second, and third floors peak to peak
are 1.3738, 1.4489, and 1.5634 m, respectively. The average oscillation amplitude
of the three floors is about 1.4620 m. In this situation, the offshore platform works
in a dangerous environment.
8.3 Simulation Results 141
Set μ = 0.8, ϑ = 0.1, and ν1 = ν2 = 10, and let the sampling period h = 0.01 s.
Suppose that the network-induced delay is a constant, i.e., τk ≡ τ, k ∈ N. In this
situation, by Proposition 8.2, set τ = 0.01, 0.02, 0.03, 0.04, and 0.05 s, respectively,
we can design different NSFCs with the form (8.3), respectively. For convenience,
the obtained controllers are denoted as NSFC1, NSFC2, NSFC3, NSFC4, and
NSFC5, respectively, and the corresponding gain matrices are listed in Table 8.1.
In what follows, unless otherwise specified, all related values of oscillation
amplitudes of the offshore platform and the control force are given in a sense of peak
to peak. Applying the above NSFCs to the offshore platform system, respectively,
the oscillation amplitudes of the first, second, and third floors of the platform and the
ranges of the control force are listed in Table 8.2, where the reduction percentages of
the average oscillation amplitudes of the offshore platform with NSFCs and without
control are also presented.
From this table, it is observed that NSFC1 can reduce the average oscillation
amplitudes of the three floors to about 13% of that of the system without control
[99]. Compared with the oscillation amplitudes of the system under other con-
trollers: NSFC2 to NSFC5, the average oscillation amplitudes under NSFC1 is
the smallest. In this case, the oscillation amplitudes of the three floors are 0.1784,
0.1969, and 0.2120 m, respectively, which can be seen from Figs. 8.2, 8.3, and 8.4.
However, it is not difficult to see from Table 8.2 and Fig. 8.5 that the control force
required by NSFC1 is about 7.6699 × 104 N, which is the largest. It indicates that
if the control cost is specifically taken into account, one may choose NSFC5 owing
to its smallest control force required, though the average vibration amplitudes of the
Table 8.1 The values of the constant network-induced delay τ , corresponding NSFCs and the
gain matrices
τ (s) Controller Gain matrix K
0.01 NSFC1 104 × [−0.2964 − 0.1050 2.7860 0.3923 0.0288 − 0.2613]
0.02 NSFC2 104 × [−0.2230 − 0.0819 1.6986 0.3564 0.0555 − 0.1996]
0.03 NSFC3 103 × [−1.6350 − 0.6521 9.1914 3.3411 0.7785 − 1.4671]
0.04 NSFC4 103 × [−0.9972 − 0.4329 0.6516 3.0963 0.9469 − 0.9254]
0.05 NSFC5 103 × [−1.0763 − 0.2700 1.4937 1.6761 0.4357 − 1.0926]
Table 8.2 The oscillation amplitudes of the floors, reduction percentages of amplitudes, and
ranges of control force under different NSFCs via constant network-induced delays
Controllers Floor 1 (m) Floor 2 (m) Floor 3 (m) Reduction (%) Control force (104 N)
No control[99] 1.3738 1.4489 1.5634 – –
NSFC1 0.1784 0.1969 0.2120 13.3 7.6699
NSFC2 0.1862 0.2053 0.2209 13.4 7.3318
NSFC3 0.1928 0.2124 0.2284 14.5 6.9681
NSFC4 0.2176 0.2391 0.2562 16.3 6.4260
NSFC5 0.2504 0.2746 0.2933 18.7 6.0955
142 8 Network-Based Modeling and Active Control
0.2
0.15
0.1
First Floor Response (m)
0.05
−0.05
−0.1
−0.15
−0.2
0 10 20 30 40 50 60 70 80 90 100
Time (s)
Fig. 8.2 The response of the first floor under NSFC1 with τk ≡ 0.01 s
0.2
0.15
0.1
Second Floor Response (m)
0.05
−0.05
−0.1
−0.15
−0.2
0 10 20 30 40 50 60 70 80 90 100
Time (s)
Fig. 8.3 The response of the second floor under NSFC1 with τk ≡ 0.01 s
platform is the largest. In fact, in this case, the oscillation amplitudes of the three
floors are increased from 0.1784, 0.1969, and 0.2120 m under NSFC1 to 0.2504,
0.2746, and 0.2933 m, respectively. The average oscillation amplitudes of the three
floors under NSFC5 is about 18.7% of that of the offshore platform without control.
Therefore, choosing a network-based controller among NSFC1 to NSFC5, there
exists a tradeoff between control performance and cost.
8.3 Simulation Results 143
0.2
0.15
0.1
Third Floor Response (m)
0.05
−0.05
−0.1
−0.15
−0.2
0 10 20 30 40 50 60 70 80 90 100
Time (s)
Fig. 8.4 The response of the third floor under NSFC1 with τk ≡ 0.01 s
4
x 10
10
6
Control Force (N)
−2
−4
0 10 20 30 40 50 60 70 80 90 100
Time (s)
Fig. 8.5 The response of the control force under NSFC1 with τk ≡ 0.01 s
The values of μ, ϑ, ν1 , ν2 , and h are set as the ones as Sect. 8.3.1.1. Then, in
each case, by Proposition 8.2, we can design corresponding NSFC, respectively. The
obtained network-based controllers are denoted by NSFC6 to NSFC11, respectively,
and the values of gain matrix K are presented by Table 8.3. When the obtained
controllers NSFC6 to NSFC11 are utilized to control the offshore platform, the
oscillation amplitudes of the first, second, and third floors, the reduction percentages
of the average oscillation amplitudes of the three floors with NSFCs and without
control, and the ranges of the control force are given in Table 8.4.
Table 8.3 The bounds τm and τM of the time-varying network-induced delay τk , corresponding
NSFCs and the gain matrices
τm (s) τM (s) Controller Gain matrix K
τM − τm = 0.01 (s)
0.01 0.02 NSFC6 104 × [−0.2324 −0.0864 1.8577 0.3647 0.0531 −0.2065]
0.02 0.03 NSFC7 103 × [−1.6052 −0.6461 8.8364 3.3530 0.7898 −1.4439]
0.03 0.04 NSFC8 103 × [−0.9980 −0.4346 0.6806 3.1036 0.9465 −0.9292]
τM − τm = 0.02 (s)
0.01 0.03 NSFC9 103 × [−1.6343 −0.6547 9.2164 3.3560 0.7805 −1.4688]
0.02 0.04 NSFC10 103 × [−0.9912 −0.4318 0.5843 3.1071 0.9479 −0.9228]
τM − τm = 0.03 (s)
0.01 0.04 NSFC11 103 × [−0.9925 −0.4321 0.6120 3.0988 0.9469 −0.9239]
Table 8.4 The oscillation amplitudes of the floors, reduction percentages of amplitudes, and
ranges of control force under different NSFCs via time-varying network-induced delays
Controllers Floor 1 (m) Floor 2 (m) Floor 3 (m) Reduction (%) Control force (104 N)
No control[99] 1.3738 1.4489 1.5634 – –
NSFC6 0.1833 0.2022 0.2176 13.8 7.4175
NSFC7 0.1933 0.2130 0.2290 14.5 6.9496
NSFC8 0.2177 0.2392 0.2564 16.2 6.4309
NSFC9 0.1928 0.2124 0.2284 14.5 6.9737
NSFC10 0.2181 0.2396 0.2568 16.3 6.4306
NSFC11 0.2181 0.2396 0.2567 16.3 6.4394
8.3 Simulation Results 145
From this table, one can see that the designed NSFCs can be classified two
groups. The first one includes NSFC8, NSFC10, and NSFC11, under which the
average oscillation amplitudes of the three floors can be reduced to about 16% of
the ones without control, and the average range of the control force by NSFC8,
NSFC10, and NSFC11 is less than 6.5 ×104 N. The second one involves NSFC6,
NSFC7, and NSFC9, and under these three controllers, the average oscillation
amplitudes of the system can be reduced to even about 14% of those when no
controller is used, while the average range of the control force required is 7.2
×104 N, which is larger than the one by the NSFCs in the first group. Specifically,
as representative controllers of the two groups, under NSFC 7 and NSFC11, the
response curves of the three floors and the control force are demonstrated by
Figs. 8.6, 8.7, 8.8, 8.9, 8.10, 8.11, 8.12, and 8.13, respectively.
Note the fact from Tables 8.2 and 8.4 that under different NSFCs, if the controlled
oscillation amplitudes of the three floors of the offshore platform are smaller,
the control force required is generally larger. For instance, under NSFC1, the
oscillation amplitudes of the floors are the smallest, while the control force is
the largest, and under NSFC5, the required control cost is the smallest, while
the control performance is not the best. Therefore, to compare the network-based
control scheme with some existing control schemes without network setting under
0.2
0.15
0.1
First Floor Response (m)
0.05
−0.05
−0.1
−0.15
−0.2
0 10 20 30 40 50 60 70 80 90 100
Time (s)
Fig. 8.6 The response of the first floor under NSFC7 with 0.02 ≤ τk ≤ 0.03 s
146 8 Network-Based Modeling and Active Control
0.2
0.15
0.1
Second Floor Response (m)
0.05
−0.05
−0.1
−0.15
−0.2
0 10 20 30 40 50 60 70 80 90 100
Time (s)
Fig. 8.7 The response of the second floor under NSFC7 with 0.02 ≤ τk ≤ 0.03 s
0.2
0.15
0.1
Third Floor Response (m)
0.05
−0.05
−0.1
−0.15
−0.2
0 10 20 30 40 50 60 70 80 90 100
Time (s)
Fig. 8.8 The response of the third floor under NSFC7 with 0.02 ≤ τk ≤ 0.03 s
4
x 10
10
6
Control Force (N)
−2
−4
0 10 20 30 40 50 60 70 80 90 100
Time (s)
Fig. 8.9 The response of the control force under NSFC7 with 0.02 ≤ τk ≤ 0.03 s
0.2
0.15
0.1
First Floor Response (m)
0.05
−0.05
−0.1
−0.15
−0.2
0 10 20 30 40 50 60 70 80 90 100
Time (s)
Fig. 8.10 The response of the first floor under NSFC11 with 0.01 ≤ τk ≤ 0.04 s
0.2
0.15
0.1
Second Floor Response (m)
0.05
−0.05
−0.1
−0.15
−0.2
0 10 20 30 40 50 60 70 80 90 100
Time (s)
Fig. 8.11 The response of the second floor under NSFC11 with 0.01 ≤ τk ≤ 0.04 s
0.2
0.15
0.1
Third Floor Response (m)
0.05
−0.05
−0.1
−0.15
−0.2
0 10 20 30 40 50 60 70 80 90 100
Time (s)
Fig. 8.12 The response of the third floor under NSFC11 with 0.01 ≤ τk ≤ 0.04 s
• Comparison of NSFC3, NSFC7 with NLC, DOFC, and SMC The oscillation
amplitudes of the three floors of the offshore platform under NSFC3 and NSFC7
are smaller than the ones under NLC, DOFC, and SMC. Moreover, the ranges of
the control force required by NSFC3 and NSFC7 are much smaller than the ones
by NLC, DOFC and SMC. In fact, under NLC, DOFC and SMC, the average
oscillation amplitudes of the three floors are reduced to about 21%, 17% and
8.3 Simulation Results 149
4
x 10
10
4
Control Force (N)
−2
−4
−6
0 10 20 30 40 50 60 70 80 90 100
Time (s)
Fig. 8.13 The response of the control force under NSFC11 with 0.01 ≤ τk ≤ 0.04 s
Table 8.5 The oscillation amplitudes of the floors, reduction percentages of amplitudes, and
ranges of control force under controllers without and with network settings
Controllers Floor 1 (m) Floor 2 (m) Floor 3 (m) Reduction (%) u (104 N)
NLC [65] 0.3050 0.3050 0.3050 20.9 20.000
DOFC [100] 0.2329 0.2543 0.2705 17.3 40.000
SMC [99] 0.2192 0.2301 0.2383 15.7 21.565
NSFC3 0.1928 0.2124 0.2284 14.5 6.9681
NSFC7 0.1933 0.2130 0.2290 14.5 6.9496
NSFC11 0.2181 0.2396 0.2567 16.3 6.4394
16% of those of the platform without control, respectively, while the average
oscillation amplitudes of the floors can be reduced even to about 14% under
NSFC3 and NSFC7 of their original value. In addition, the ranges of the control
force by NLC, DOFC and SMC are about 2.9, 5.8 and 3.1 times as the ones by
NSFC3 and NSFC7.
• Comparison of NSFC11 with NLC, DOFC, and SMC On the one hand, both
the oscillation amplitudes of the three floors and required control force under
NSFC11 are much smaller than the ones under NLC and DOFC, respectively.
This shows that NSFC11 is better than NLC and DOFC from the perspective of
the control performance and cost. On the other hand, the oscillation amplitudes of
the three floors under NSFC11 are almost in the same level as the ones by SMC,
while the control force required by SMC is 3.3 times as the one by NSFC11,
which indicates that to obtain the same control performance, the control cost
required by NSFC11 is smaller the one by SMC.
150 8 Network-Based Modeling and Active Control
Table 8.6 The oscillation amplitudes of the floors, reduction percentages of amplitudes, and
ranges of control force under NSFC3 as the network-induced delay τk ≡ τ varies
τ (s) Floor 1 (m) Floor 2 (m) Floor 3 (m) Reduction (%) Control force (104 N)
0.01 0.1925 0.2121 0.2281 14.3 6.9665
0.02 0.1927 0.2123 0.2282 14.4 6.9627
0.03 0.1928 0.2124 0.2284 14.4 6.9680
0.04 0.1929 0.2125 0.2285 14.5 6.9843
0.05 0.1930 0.2126 0.2286 14.6 7.0014
0.06 0.1931 0.2127 0.2287 14.6 6.9958
8.3 Simulation Results 151
Table 8.7 The oscillation amplitudes of the floors, reduction percentages of amplitudes, and
ranges of control force under NSFC7 as τk varies on interval [τm , τM ]
τm (s) τM (s) Floor 1 (m) Floor 2 (m) Floor 3 (m) Reduction (%) Control force (104 N)
0.01 0.02 0.1932 0.2128 0.2288 14.5 6.9485
0.03 0.1933 0.2130 0.2290 14.5 6.9592
0.04 0.1934 0.2131 0.2291 14.5 6.9729
0.05 0.1935 0.2132 0.2292 14.5 6.9875
0.06 0.1935 0.2132 0.2292 14.5 7.0319
0.07 0.1936 0.2133 0.2293 14.5 7.4784
0.02 0.02 0.1932 0.2128 0.2288 14.5 6.9419
0.03 0.1933 0.2130 0.2290 14.5 6.9558
0.04 0.1934 0.2131 0.2291 14.5 6.9742
0.05 0.1935 0.2132 0.2292 14.5 6.9903
0.06 0.1936 0.2132 0.2292 14.5 7.0027
0.07 0.1938 0.2133 0.2293 14.5 13.191
0.03 0.03 0.1933 0.2130 0.2290 14.5 6.9474
0.04 0.1934 0.2131 0.2291 14.5 6.9734
0.05 0.1935 0.2132 0.2292 14.5 6.9891
0.06 0.1936 0.2133 0.2293 14.5 7.0332
0.07 0.1937 0.2134 0.2300 14.5 14.230
0.04 0.04 0.1934 0.2131 0.2291 14.5 6.9634
0.05 0.1935 0.2132 0.2292 14.5 6.9902
0.06 0.1936 0.2133 0.2293 14.5 7.0099
0.07 0.1936 0.2139 0.2339 14.5 74.391
force required tends to become large significantly. It means that the larger the
network-induced delay, the more control cost NSFC3, NSFC7, and NSFC11 will
take.
• The allowable maximum network-induced delay The allowable maximum
network-induced delay under NSFC3, NSFC7, and NSFC11 is 0.06, 0.07, and
0.08 s, respectively. It means that if the network-induced delay is larger than
0.06 s for NSFC3, 0.07 s for NSFC7, and 0.08 s for NSFC11, respectively, the
controlled oscillation amplitudes of the three floors or the required control force
may increase significantly, in these cases, the obtained controllers may not work.
Also, it should be pointed herein that the allowable maximum network-induced
delay under NSFC11 is larger than the one under NSFC3, which means that
NSFC11 is more practical than both NSFC3 and NSFC7.
In summary, based on the developed network-based dynamic model of the
offshore platform, for the proper network-induced delays, the network-based state
feedback control scheme can significantly attenuate the vibration of the offshore
platform. In addition, the proposed network-based control scheme is superior to
some existing control strategies without network-setting, i.e., nonlinear control [65],
dynamic output feedback control [100], and integral sliding mode control [99].
152 8 Network-Based Modeling and Active Control
Table 8.8 The oscillation amplitudes of the floors, reduction percentages of amplitudes, and
ranges of control force under NSFC11 as τk varies on interval [τm , τM ]
τm (s) τM (s) Floor 1 (m) Floor 2 (m) Floor 3 (m) Reduction (%) Control force (104 N)
0.01 0.02 0.2175 0.2390 0.2561 16.3 6.3898
0.03 0.2178 0.2393 0.2564 16.3 6.4177
0.04 0.2181 0.2396 0.2567 16.3 6.4287
0.05 0.2184 0.2399 0.2570 16.3 6.4448
0.06 0.2186 0.2401 0.2572 16.3 6.4763
0.07 0.2188 0.2403 0.2574 16.3 6.5687
0.08 0.2192 0.2406 0.2578 16.4 7.3361
0.09 0.2203 0.2409 0.2606 16.5 80.256
0.02 0.03 0.2178 0.2393 0.2564 16.3 6.4154
0.04 0.2181 0.2396 0.2567 16.3 6.4272
0.05 0.2184 0.2399 0.2570 16.3 6.4448
0.06 0.2186 0.2402 0.2573 16.3 6.4707
0.07 0.2189 0.2404 0.2575 16.3 6.5943
0.08 0.2192 0.2406 0.2577 16.4 7.4523
0.09 0.2216 0.2409 0.2601 16.5 137.32
0.03 0.04 0.2181 0.2396 0.2567 16.3 6.4331
0.05 0.2184 0.2399 0.2570 16.3 6.4442
0.06 0.2187 0.2402 0.2573 16.3 6.4777
0.07 0.2189 0.2405 0.2576 16.4 6.5940
0.08 0.2193 0.2407 0.2578 16.4 7.4242
0.09 0.2315 0.2425 0.2311 17.0 415.94
8.4 Conclusions
In this chapter, a network-based dynamic model of the offshore platform has been
established and a network-based control scheme for the offshore platform has been
developed. A new Lyapunov-Krasovskii functional has been constructed to derive
a delay-dependent sufficient condition, which makes the system asymptotically
stable. Based on the condition, the gain matrix of the network-based state feedback
controller has been obtained. The simulation results have demonstrated that the
proposed network-based controller is effective to improve the control performance
of the offshore platform and reduce the required control cost. Moreover, the
proposed network-based state feedback controllers are better than some existing
ones without network setting.
8.5 Notes
Note that networked control system is connected over networked media, it is a viable
and efficient way to deal with remote control problems. In recent two decades, net-
worked control system has received increasing research interest, a comprehensive
8.5 Notes 153
coverage of this topic can be found in Ge et al. [117], Zhang et al. [119], and the
references therein. So far, there have been considerable research results concerning
the modeling, stability analysis and control, and filtering problems for the networked
system [118]. In addition, network-based control strategies have been applied to
several engineering applications.
Due to the fact that offshore structures are often far from land and always
affected by very complicated and harsh ocean environmental loads. From the
saving low control cost point of view, simplifying installation and maintenance,
improving reliability, and increasing safety of the staff on the structure, it is a
feasible and efficient way to control the offshore structure over the network media.
Based on [120], this chapter firstly provides a networked dynamic model and
then presents a network-based active controller for the offshore platform with a
TMD mechanism. The superiority of network-based controller over some existing
ones without network setting such as the nonlinear controller [65], the dynamic
output feedback controller [100], and the integral sliding mode controller [99]
is investigated. However, in [120], only network-induced delays are investigated,
while other network-induced factors including packet dropouts, packet disorders,
quantization errors, and network congestion are ignored. Therefore, by taking into
account one or more of network-induced delays, packet dropouts, packet disorders,
quantization errors, and network congestion, to establish more general network-
based dynamic models and design effective network-based controllers for offshore
platform systems is still a challenging topic.
Chapter 9
Event-Triggered H∞ Reliable Control
in Network Environments
Fig. 9.1 Control diagram for a network-based offshore structure with an AMD [123]
τk := tk − ik h, k = 1, 2, · · · (9.3)
It is clear that ϕk ≥ 1. Then, similar to [100], we denote the interval [tk , tk+1 ] as
'
ϕk
[tk , tk+1 ] = ϑj (9.9)
j =1
where
where
Similarly, we define a state error function δ(t) over [tk , tk+1 ) as follows:
⎧
⎪
⎪ x(ik h) − x(ik h), t ∈ ϑ1
⎪
⎨ x(ik h) − x(ik h + h), t ∈ ϑ2
δ(t) = .. .. (9.13)
⎪
⎪ . .
⎪
⎩
x(ik h) − x(ik h + (ϕk − 1)h), t ∈ ϑϕk
where
Γ1 (t) = A r(t)BK 0 0 r(t)BK D (9.18)
T
α(t) = x(t) x(t − ρ(t)) x(t − τm ) x(t − η̄) δ(t) f (t) (9.19)
of the closed-loop system can be ensured for the wave force f (t) ∈ L2 [0, ∞) and a
prescribed γ > 0.
9.2 Design of an Event-Triggered H∞ Reliable Controller 159
where
⎡ ⎤
Θ r̄P BK R1 0 r̄P BK P D
⎢ ∗ −2R + S + S T R2 − S R2 − S T 0 ⎥
⎢ 2 0 ⎥
⎢ ⎥
⎢∗ ∗ Q2 − Q1 − R1 − R2 S T 0 0 ⎥
Δ11 =⎢ ⎥
⎢∗ ∗ ∗ −Q2 − R2 0 0 ⎥
⎢ ⎥
⎣∗ ∗ ∗ ∗ −Ω 0 ⎦
∗ ∗ ∗ ∗ ∗ −γ 2 I
(9.23)
⎡ ⎤ ⎡ ⎤
0 τm AT (η̄ − τm )AT C1T εP B 0
⎢ I r̄τ K T B T r̄(η̄ − τ )K T B T 0 ⎥ ⎢ 0 r̄K T ⎥
⎢ m m ⎥ ⎢ ⎥
⎢ ⎥ ⎢ ⎥
⎢0 0 0 0 ⎥ ⎢ 0 0 ⎥
Δ12 =⎢ ⎥ , Δ13 = ⎢ ⎥ (9.24)
⎢0 0 0 0 ⎥ ⎢ 0 0 ⎥
⎢ ⎥ ⎢ ⎥
⎣ I r̄τm K T B T r̄(η̄ − τm )K T B T 0 ⎦ ⎣ 0 r̄K T ⎦
0 τm D T (η̄ − τm )D T D1T 0 0
where
V̇ (t, x(t)) = V̇1 (t, x(t)) + V̇2 (t, x(t)) + V̇3 (t, x(t)) (9.29)
where
Note that matrices R2 and S satisfy the constraint (9.21). Then, by Lemma 2.8,
we have
t−τm
− (η̄ − τm ) ẋ T (s)R2 ẋ(s)ds
t−η̄
where
⎡ ⎤
Θ r(t)P BK R1 0 r(t)P BK PD
⎢ ∗ −2R + S + S T − − T 0 ⎥
⎢ 2 R 2 S R2 S 0 ⎥
⎢ ⎥
⎢ ∗ ∗ Q 2 − Q 1 − R 1 − R2 S T 0 0 ⎥
Π1 (t) = ⎢ ⎥
⎢∗ ∗ ∗ −Q2 − R2 0 0 ⎥
⎢ ⎥
⎣∗ ∗ ∗ ∗ −Ω 0 ⎦
∗ ∗ ∗ ∗ ∗ 0
Π2 (t) = Γ2T (σ Ω)Γ2 + τm2 Γ1T (t)R1 Γ1 (t) + (η̄ − τm )2 Γ1T (t)R2 Γ1 (t)
Now, we turn to prove that the system (9.17) is asymptotically stable. For this,
set f (t) = 0 in (9.17) and denote
Λ1 (t) = A r(t)BK 0 0 r(t)BK
T
β(t) = x(t) x(t − ρ(t)) x(t − τm ) x(t − η̄) δ(t)
where
⎡ ⎤
Θ r(t)P BK R1 0 r(t)P BK
⎢ ∗ −2R + S + S T R2 − S R2 − S T ⎥
⎢ 2 0 ⎥
⎢ ⎥
Ξ1 (t) = ⎢ ∗ ∗ Q2 − Q1 − R1 − R2 S T 0 ⎥
⎢ ⎥
⎣∗ ∗ ∗ −Q2 − R2 0 ⎦
∗ ∗ ∗ ∗ −Ω
(9.40)
Ξ2 (t) = ΛT2 (σ Ω)Λ2 + τm2 ΛT1 (t)R1 Λ1 (t) + (η̄ − τm )2 ΛT1 (t)R2 Λ1 (t)
with Λ2 = 0 I 0 0 I .
To guarantee that the system (9.38) is asymptotically stable, we require that there
exist 4 × 4 matrices P > 0, Q1 > 0, Q2 > 0, R1 > 0, R2 > 0, S, Ω > 0, and a
1 × 4 matrix K such that
where
⎡ ⎤
Θ r̄P BK R1 0 r̄P BK
⎢ ∗ −2R + S + S T R2 − S R2 − S T 0 ⎥
⎢ 2 ⎥
⎢ ⎥
Φ11 =⎢∗ ∗ Q2 − Q1 − R1 − R2 ST 0 ⎥
⎢ ⎥
⎣∗ ∗ ∗ −Q2 − R2 0 ⎦
∗ ∗ ∗ ∗ −Ω
⎡ ⎤
0 τm AT (η̄ − τm )AT
⎢ I r̄τ K T B T r̄(η̄ − τ )K T B T ⎥
⎢ m m ⎥
⎢ ⎥
Φ12 = ⎢0 0 0 ⎥
⎢ ⎥
⎣0 0 0 ⎦
I r̄τm K T B T r̄(η̄ − τm )K T B T
9.2 Design of an Event-Triggered H∞ Reliable Controller 163
T
1 = B T P 0 0 0 0 0 τm B T (η̄ − τm )B T
2 = 0 r̄K 0 0 r̄K 0 0 0
By S -procedure and Schur complement, for any ε > 0, the matrix inequality
(9.45) is equivalent to
⎡ ⎤
Φ ε1 T2
⎣ ∗ −εI 0 ⎦ < 0 (9.46)
∗ ∗ −εI
Due to the fact that the inequalities (9.21) and (9.22) hold, it is clear that the above
inequality is true, which means that the system (9.17) is asymptotically stable.
Now, we focus on proving that the H∞ performance (9.20) is guaranteed for the
external wave force f (t) under
zero initial condition.
Let Γ3 = C1 0 0 0 0 D1 . Then, from (9.1) and (9.37), one yields
where
⎡ ⎤
Θ r(t)P BK R1 0 r(t)P BK P D
⎢ ∗ −2R + S + S T R2 − S R2 − S T 0 ⎥
⎢ 2 0 ⎥
⎢ ⎥
⎢∗ ∗ Q2 − Q1 − R1 − R2 S T 0 0 ⎥
Π (t) = ⎢ ⎥
⎢∗ ∗ ∗ −Q2 − R2 0 0 ⎥
⎢ ⎥
⎣∗ ∗ ∗ ∗ −Ω 0 ⎦
∗ ∗ ∗ ∗ ∗ −γ 2 I
+ Γ2T (σ Ω)Γ2 + τm2 Γ1T (t)R1 Γ1 (t) + (η̄ − τm )2 Γ1T (t)R2 Γ1 (t) + Γ3T Γ3
(9.48)
In what follows, we prove that Π (t) < 0 is true, if matrix inequalities (9.21) and
(9.22) hold. In fact, by Schur complement, Π (t) < 0 if and only of the following
inequality is true:
Δ11 Δ12
+ r1 (t)Σ1 Σ2 + r1 (t)Σ2T Σ1T < 0 (9.49)
∗ Δ22
where
T
Σ1 = B T P 0 0 0 0 0 0 τm B T (η̄ − τm )B T 0
Σ2 = 0 r̄K 0 0 r̄K 0 0 0 0 0
By S -procedure and the Schur complement again, the known conditions (9.21) and
(9.22) guarantee that the inequality (9.49) holds. Then from (9.47), one yields
Note that V (t, x(0)) = 0 under zero initial condition. Integrating both sides of
(9.50) from 0 to ∞ yields
∞
[zT (t)z(t) − γ 2 f T (t)f (t)]dt < 0 (9.51)
0
Ψ1 = diag{P −1 , P −1 } (9.52)
Ψ2 = diag{P −1 , P −1 , P −1 , P −1 , P −1 , I, I, I, I, I, I, I } (9.53)
9.2 Design of an Event-Triggered H∞ Reliable Controller 165
where Δ23 and Δ33 are given by (9.26) and (9.27), respectively, and
⎡ ⎤
Θ̄ r̄B K̄ R̄1 0 r̄B K̄ D
⎢ ∗ −2R̄ + S̄ + S̄ T R̄2 − S̄ R̄2 − S̄ T 0 ⎥
⎢ 2 0 ⎥
⎢ ⎥
⎢∗ ∗ Q̄2 − Q̄1 − R̄1 − R̄2 S̄ T 0 0 ⎥
Δ̄11 =⎢ ⎥
⎢∗ ∗ ∗ −Q̄2 − R̄2 0 0 ⎥
⎢ ⎥
⎣∗ ∗ ∗ ∗ −Ω̄ 0 ⎦
∗ ∗ ∗ ∗ ∗ −γ 2 I
(9.56)
⎡ ⎤ ⎡ ⎤
0 τm P̄ AT (η̄ − τm )P̄ AT P̄ C1T εB 0
⎢ P̄ r̄τ K̄ T B T r̄(η̄ − τ )K̄ T B T 0 ⎥ ⎢ 0 r̄ K̄ T ⎥
⎢ m m ⎥ ⎢ ⎥
⎢ ⎥ ⎢ ⎥
⎢0 0 0 0 ⎥ ⎢ 0 0 ⎥
Δ̄12 =⎢ ⎥ , Δ̄13 = ⎢ ⎥ (9.57)
⎢0 0 0 0 ⎥ ⎢ 0 0 ⎥
⎢ ⎥ ⎢ ⎥
⎣ P̄ r̄τm K̄ T B T r̄(η̄ − τm )K̄ T B T 0 ⎦ ⎣ 0 r̄ K̄ T ⎦
0 τm D T (η̄ − τm )D T D1T 0 0
with
Then, the closed-loop system (9.17) with f (t) = 0 is asymptotically stable, and the
H∞ performance (9.20) is guaranteed for the wave force f (t) ∈ L2 [0, ∞] and the
prescribed γ . Moreover, the matrix K in (9.16) is given by K = K̄ P̄ −1 .
Due to the nonlinear terms such as σ̄ P̄ Ω̄ −1 P̄ , P̄ R̄1−1 P̄ and P̄ R̄2−1 P̄ in (9.58), it
is difficult to compute the variables P̄ , Q̄1 , Q̄2 , R̄1 , R̄2 , S̄, Ω̄, K̄, and ε and thereby
to obtain the gain matrix K. Noting the fact that
− σ̄ P̄ Ω̄ −1 P̄ ≤ σ̄ Ω̄ − 2σ̄ P̄
where Δ̄11 , Δ̄12 , and Δ̄13 are defined in (9.56) and (9.57), Δ23 and Δ33 are given
by (9.26) and (9.27), respectively, and
then, the closed-loop system (9.17) with f (t) = 0 is asymptotically stable, and the
H∞ performance (9.20) is guaranteed for the wave force f (t) ∈ L2 [0, ∞] and the
prescribed γ . Moreover, the matrix K in (9.16) is given by K = K̄ P̄ −1 .
If there is no any fault in the actuator, one yields an event-triggered H∞ controller
as
In this situation, the matrix K can be solved via the following corollary, which is a
special case of Proposition 9.3.
Corollary 9.1 For given scalars τm and τM satisfying 0 ≤ τm ≤ τM , γ > 0 and
h > 0, if there exist 4 × 4 matrices P̄ > 0, Q̄1 > 0, Q̄2 > 0, R̄1 > 0, R̄2 > 0, S̄,
Ω̄ > 0, a 1 × 4 matrix K̄ and a positive scalar ε such that (9.54) and
Υ11 Υ12
<0 (9.63)
∗ Δ̂22
9.2 Design of an Event-Triggered H∞ Reliable Controller 167
where the matrix Θ̄ is defined by (9.59). Then, under the controller (9.62), the system
(2.3) with f (t) = 0 is asymptotically stable, and the H∞ performance (9.20) is
guaranteed for the wave force f (t) ∈ L2 [0, ∞] and the prescribed γ . Moreover,
the gain matrix K in (9.62) is determined by K = K̄ P̄ −1 .
It should be pointed out that the above results are based on the dynamic model of
the offshore structure, where the first dominant vibration mode is considered. Note
that other higher vibration modes may have some effects on the performance and
stability of the offshore structure. If we take the effects of higher vibration modes
on the first mode and the AMD into account and treat them as the unknown but
bounded nonlinear perturbations, then the corresponding dynamic equations of the
offshore structure can be described by
⎧
⎪
⎪ m1 z̈1 (t) = −c1 ż1 (t) − k1 z1 (t) + k2 (z2 (t) − z1 (t)) + c2 (ż2 (t) − ż1 (t))
⎨
+f (t) − u(t) + g1 (t, z1 (t), z2 (t), ż1 (t), ż2 (t))
⎪
⎪ m z̈ (t) = −c2 (ż2 (t) − ż1 (t)) − k2 (z2 (t) − z1 (t))
⎩ 2 2
+g2 (t, z1 (t), z2 (t), ż1 (t), ż2 (t))
(9.65)
where m1 , m2 , k1 , k2 , c1 , c2 , z1 (t), z2 (t), u(t), and f (t) are defined in the system
(2.1) and g1 (t, z1 , z2 , ż1 , ż2 ) and g2 (t, z1 , z2 , ż1 , ż2 ) present the effects of higher
vibration modes on the first mode and the AMD, respectively.
It is supposed that the perturbation terms g1 (t, z1 , z2 , ż1 , ż2 ) and g2 (t, z1 , z2 ,
ż1 , ż2 ) satisfy
T
giT (t, z1 , z2 , ż1 , ż2 )gi (t, z1 , z2 , ż1 , ż2 ) ≤ ςi2 z1 z2 ż1 ż2 z1 z2 ż1 ż2
gi (t, 0, 0, 0, 0) = 0, i = 1, 2
(9.66)
168 9 Event-Triggered H∞ Reliable Control in Network Environments
where ςi > 0, i = 1, 2.
Define
⎡ 1 ⎤T
00 0
⎢ m ⎥ g1 (t, x(t))
D0 = ⎣ 1
1 ⎦ , g(t, x(t)) = g2 (t, x(t)) (9.67)
00 0
m2
⎡ ⎤ ⎡ ⎤
0 τm P̄ AT (η̄ − τm )P̄ AT P̄ C1T εB 0 ς P̄
⎢ P̄ r̄τ K̄ T B T r̄(η̄ − τ )K̄ T B T 0 ⎥ ⎢ 0 r̄ K̄ T 0 ⎥
⎢ m m ⎥ ⎢ ⎥
⎢0 0 ⎥ ⎢ 0 0 0 ⎥
⎢ 0 0 ⎥ ⎢ ⎥
⎢ ⎥ ⎢ ⎥
Δ̂12 =⎢0 0 0 0 ⎥ , Δ̂13 = ⎢ 0 0 0 ⎥
⎢ ⎥ ⎢ ⎥
⎢ P̄ r̄τm K̄ T B T r̄(η̄ − τm )K̄ T B T 0 ⎥ ⎢ 0 r̄ K̄ T 0 ⎥
⎢ ⎥ ⎢ ⎥
⎣ 0 τm D T (η̄ − τm )D T D1T ⎦ ⎣ 0 0 0 ⎦
0 τm D0 T (η̄ − τm )D0 T 0 0 0 0
(9.72)
⎡ ⎤T
0 ετm B T ε(η̄ − τm )B T 0
Δ̂23 = ⎣0 0 0 0⎦ (9.73)
0 0 0 0
In this section, the related parameters of an offshore structure are first given. Then,
event-triggered H∞ controllers are designed, respectively, for the system without
actuator faults and the system with actuator faults. Under the obtained controllers,
the performance of the structure is investigated.
Suppose that the parameters of the offshore platform [79] is set as Table 3.1.
Based on the setting in the table, one yields the matrices A and B in system (2.3)
as (3.26). Choose displacements of the offshore structure and the AMD as the
controlled outputs, in this case, the matrices C1 and D1 in (9.1) are given as
1000 0.01
C1 = , D1 =
0010 0
Based on (2.18), the irregular wave force can be computed and presented in
Fig. 9.2. In what follows, we focus on designing event-triggered H∞ controllers,
respectively, for the system without actuator faults and the system with actuator
faults. We set the H∞ performance index γ as 0.1 and the sampling period h as 0.01
second (s). On the one hand, to demonstrate the effectiveness of the proposed control
schemes, the controlled vibration amplitudes of the displacement and acceleration
of the offshore structure are computed, respectively; on the other hand, to measure
170 9 Event-Triggered H∞ Reliable Control in Network Environments
5
x 10
5
1
Control Force (N)
−1
−2
−3
−4
−5
0 10 20 30 40 50 60
Time (s)
the performance of the event-triggered scheme, the transmission rate (TR) of the
packets [100] is investigated.
It can be obtained that the maximum values of the displacement and acceleration
of the structure without control are 0.2829 m and 0.7987 m/s2 , respectively, and the
root mean square (RMS) values of them are 0.1025 m and 0.3088 m/s2 , respectively.
Under the obtained ETHC, the maximum values of the displacement and
acceleration of the offshore structure are reduced from 0.2829 m and 0.7987 m/s2
to 0.2221 m and 0.4627 m/s2 , respectively; the RMS values of them are reduced
from 0.1025 m and 0.3088 m/s2 to 0.0859 m and 0.1934 m/s2 , respectively. One
can conclude that under the designed controller, the vibration amplitudes of the
displacement and acceleration of the structure are effectively reduced. The response
curves of the offshore structure without control and with the ETHC are presented in
Figs. 9.3 and 9.4. To suppress the vibration to such levels, the peak and RMS values
of the required force are 4.7759 ×105 N and 1.9171 ×105 N, respectively. Depicted
in Fig. 9.5 is the curve of the control force by the ETHC, and in Fig. 9.6 is the
diagram of the release time intervals. In fact, one can obtain that under the ETHC,
the TR on [0, 60 s] is 0.0247, which means that the performance of the structure can
be maintained while 97.53% network resources are saved.
To compare the event-triggered H∞ control scheme with the traditional H∞ con-
trol scheme without network setting, we require to study the vibration amplitudes of
the structure with the H∞ control scheme. For this purpose, set the H∞ performance
index γ as 0.1, which is equal to the one for the ETHC. Thus, by Corollary 9.1 in
[89], the gain matrix of an H∞ controller (HIC) can be obtained as
K = 106 × 4.4268 0.1639 0.9081 −0.0709
Under this controller, the peak and RMS values of the structure responses and
the ranges of the required force are given by Table 9.1, where Md , Ma , and
Mu denote the peak values of displacement, acceleration of the offshore platform,
0.3
No Control
Event−triggered H ∞ Controller
0.2
Displacement of Offshore Platform (m)
0.1
−0.1
−0.2
−0.3
0 10 20 30 40 50 60
Time (s)
Fig. 9.3 The displacement responses of system under no control and ETHC
172 9 Event-Triggered H∞ Reliable Control in Network Environments
0.8
No Control
Event−triggered H ∞ Controller
0.6
0.4
Acceleration of Offshore Platform (m/s )
2
0.2
−0.2
−0.4
−0.6
−0.8
0 10 20 30 40 50 60
Time (s)
Fig. 9.4 The acceleration responses of system under no control and ETHC
5
x 10
5
1
Control Force (N)
−1
−2
−3
−4
−5
0 10 20 30 40 50 60
Time (s)
and the required control force, respectively; Jd , Ja , and Ju represent the RMS
values of displacement, acceleration of the offshore platform, and the control force,
respectively (3.27). The table indicates that both the maximum and the RMS values
9.3 Simulation Results 173
3.5
3
Event−based Release Instants and Release Interval
2.5
1.5
0.5
0
0 10 20 30 40 50 60
Time (s)
Table 9.1 The peak and RMS values of displacement, acceleration of the offshore structure, and
the range of the control force under ETHC and HIC
Peak value RMS value
Controllers (s) Md (m) Ma (m/s) Mu (105 N) Jd (m) Ja (m/s) Ju (105 N)
No control 0.2829 0.7987 – 0.1025 0.3088 –
HIC [89] 0.2171 0.4188 14.122 0.0841 0.1760 5.7285
ETHC 0.2221 0.4627 4.7759 0.0859 0.1934 1.9171
of the vibration amplitudes of the structure under the HIC are slightly smaller than
those under the ETHC. However, it is not difficult to see that the required force by
the HIC is larger than the one by the ETHC. In fact, the maximum and the RMS
values of the force by the HIC are nearly three times as those by the ETHC.
Remark 9.2 Compared with the proposed event-triggered H∞ control scheme,
some existing control schemes without network environments, such as the H2
control [79], the feedforward and feedback optimal control [81, 82], the optimal
tracking control [84], and the delayed H∞ control [89], may provide smaller
vibration amplitudes of the structure. However, the required control force by the
proposed event-triggered H∞ control scheme is much less than the one by some
existing control schemes without network environments. Specifically, notice that
the designed ETHC is of several advantages of the network-based control as well
as saving communication resources. Therefore, choosing either the event-triggered
H∞ control scheme or other control schemes without network setting is a tradeoff
between the vibration amplitudes of the structure and the required control force.
174 9 Event-Triggered H∞ Reliable Control in Network Environments
In the next subsection, in the case of the offshore structure with actuator
faults, the performance of the system under event-triggered H∞ reliable controller
(ETHRC) is investigated.
0.3
No Control
Event−triggered H ∞ Reliable Controller
0.2
Displacement of Offshore Platform (m)
0.1
−0.1
−0.2
0 10 20 30 40 50 60
Time (s)
Fig. 9.7 The displacement responses of system under no control and ETHRC1
9.3 Simulation Results 175
0.8
No Control
Event−triggered H ∞ Reliable Controller
0.6
0.4
Acceleration of Offshore Platform (m/s )
2
0.2
−0.2
−0.4
−0.6
−0.8
0 10 20 30 40 50 60
Time (s)
Fig. 9.8 The acceleration responses of system under no control and ETHRC1
5
x 10
5
2
Control Force (N)
−1
−2
−3
−4
0 10 20 30 40 50 60
Time (s)
1.6
1.4
Event−based Release Instants and Release Interval
1.2
0.8
0.6
0.4
0.2
0
0 10 20 30 40 50 60
Time (s)
the required force are about 4.1065 ×105 N and 1.4334 ×105 N, respectively. It is
found that if there exist actuator faults in the offshore structure, the performance of
the structure can be ensured. In addition, the TR for the ETHRC1 is 0.0478, which
indicates that during the control process, only 4.78% communication resources
are transmitted to the ZOH through the communication network, while 97.53%
resources are saved. Figure 9.10 presents a diagram of release time intervals versus
release instants.
K = 106 × 3.4059 −0.0032 0.0939 −0.0497
0.3
No Control
Event−triggered H ∞ Reliable Controller
0.2
Displacement of Offshore Platform (m)
0.1
−0.1
−0.2
−0.3
0 10 20 30 40 50 60
Time (s)
Fig. 9.11 The displacement responses of system under no control and ETHRC2
178 9 Event-Triggered H∞ Reliable Control in Network Environments
0.8
No Control
Event−triggered H Reliable Controller
∞
0.6
0.4
Acceleration of Offshore Platform (m/s )
2
0.2
−0.2
−0.4
−0.6
−0.8
0 10 20 30 40 50 60
Time (s)
Fig. 9.12 The acceleration responses of system under no control and ETHRC2
5
x 10
6
2
Control Force (N)
−2
−4
−6
0 10 20 30 40 50 60
Time (s)
Based on the above simulation results, one can see that the following statements
are true:
• If there is no any actuator fault in the control process, the designed event-
triggered H∞ controller can effectively attenuate the vibration of the offshore
9.3 Simulation Results 179
0.9
Event−based Release Instants and Release Interval
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60
Time (s)
Table 9.2 The peak and RMS values of displacement, acceleration of the offshore structure, the
control force, and the TR under ETHC with τm = 0.02 s and τM varies
Peak value RMS value
τM (s) Md (m) Ma (m/s) Mu (105 N) Jd (m) Ja (m/s) Ju (105 N) TR (%)
0.03 0.2213 0.4601 4.8412 0.0860 0.1926 1.9567 2.52
0.07 0.2215 0.4601 4.8436 0.0859 0.1927 1.9496 2.48
0.15 0.2221 0.4574 4.8619 0.0856 0.1928 1.9381 2.50
0.25 0.2231 0.4584 4.8530 0.0854 0.1943 1.9290 2.58
0.40 0.2240 0.4701 4.7654 0.0850 0.1946 1.9223 2.63
0.70 0.2366 0.5325 5.2742 0.0868 0.2228 1.9964 2.87
0.90 0.2306 0.6739 5.4838 0.0917 0.2676 2.2753 3.42
1.00 0.2645 0.6659 6.3192 0.0959 0.2991 2.6886 3.47
1.01 0.2924 0.8535 8.1240 0.0960 0.3072 2.5185 3.78
structure, and reduce the control cost dramatically. If there exist actuator faults
in the system, the event-triggered H∞ reliable controller can also stabilize the
platform and thus improve performance of the offshore structure.
• Under the event-triggered H∞ controllers, the network resources can be saved
significantly, thereby a good network service can be guaranteed and satisfactory
performance of the offshore platform can be ensured.
180 9 Event-Triggered H∞ Reliable Control in Network Environments
Table 9.3 The peak and RMS values of displacement, acceleration of the offshore structure, the
control force, and the TR under ETHRC1 for different values of constant network-induced delay τ
Peak value RMS value
τM (s) Md (m) Ma (m/s) Mu (105 N) Jd (m) Ja (m/s) Ju (105 N) TR (%)
0.02 0.2268 0.5211 4.1062 0.0917 0.2367 1.4359 4.70
0.08 0.2273 0.5210 4.1132 0.0916 0.2378 1.4228 4.72
0.14 0.2279 0.5219 4.1265 0.0916 0.2394 1.4117 4.70
0.25 0.2295 0.5318 4.1711 0.0918 0.2437 1.3975 4.83
0.40 0.2329 0.5647 4.2919 0.0927 0.2539 1.3969 4.77
0.60 0.2416 0.6565 4.6376 0.0960 0.2850 1.4763 4.92
0.70 0.2498 0.7404 4.9565 0.1006 0.3213 1.6141 5.25
Table 9.4 The peak and RMS values of displacement, acceleration of the offshore structure, the
control force, and the TR under ETHRC2 with τm = 0.01 s, and the values of τM are different τ
Peak value RMS value
τM (s) Md (m) Ma (m/s) Mu (105 N) Jd (m) Ja (m/s) Ju (105 N) TR (%)
0.03 0.2146 0.5287 5.8426 0.0916 0.2340 2.1104 4.48
0.12 0.2150 0.5338 5.8346 0.0915 0.2349 2.0827 4.55
0.25 0.2159 0.5443 5.8184 0.0915 0.2377 2.0518 4.88
0.40 0.2170 0.5509 5.9327 0.0918 0.2426 2.0429 4.67
0.57 0.2183 0.5627 6.1172 0.0917 0.2462 2.0611 4.73
0.65 0.2241 0.5919 6.1092 0.0923 0.2548 2.0888 4.75
0.75 0.2236 0.6391 6.8572 0.0951 0.2784 2.1667 5.35
0.80 0.2344 0.7243 7.5623 0.0977 0.3018 2.4874 6.13
9.4 Conclusions
9.5 Notes
This chapter is mainly based on Zhang and Han [123]. Another result of network-
based controller design of offshore platform is reported in [120]. In fact, in [120],
a design approach of network-based state feedback controller for an offshore steel
jacket platform subject to self-excited nonlinear hydrodynamic force is provided,
9.5 Notes 181
and the effects of network-induced delay on the active control for the platform is
investigated. However, the main concern of [123] is that, based on an offshore steel
jacket platform subject to external wave force, to deal with the limited resources
of communication networks and actuator faults [85]. As is known that unexpected
actuator faults and fatigue damage often appear in the implementation of the
offshore platform systems. Such faults and even fatigue damage generally lead to
poor performance and even instability of the platforms. To prevent fatigue damage
of offshore platforms and to protect operation and staff on offshore platforms subject
to a wide range of environmental loading, it is very important for offshore platforms
to improve the ability to detect, diagnose, and tolerate malfunctions of the control
system. To develop reliable fault diagnosis and fault-tolerant controllers for the
offshore platforms under network settings to cope with potential failures in actuators
and sensors is one of significant issues.
Another issue is that vibration control of offshore platforms lies at an inter-
section of different research areas including structure vibration theory, control
theory, communication theory, civil engineering, mechanical engineering, and ocean
engineering. Consequently, collaborative research in vibration control of offshore
platforms is required from different engineering and scientific fields.
Compared with passive and semi-active control, active control has several
distinct advantages. However, if power supply cannot be guaranteed during the harsh
ocean environment, the active controller will not work [4]. In this situation, semi-
active and hybrid control mechanisms with passive components are still feasible
options for the vibration control of the offshore platforms. Therefore, to develop
semi-active and hybrid control mechanisms with high reliability, desirable control
effects and low control cost requires further study.
References
19. Kareem, A.: Mitigation of wind induced motion of tall buildings. J. Wind Eng. Ind. Aerod.
11(1–3), 273–284 (1983)
20. Alves, R.M., Batista, R.C.: Active/passive control of heave motion for TLP type of offshore
platforms. In: Proceedings of the International Offshore and Polar Engineering Conference,
Brest, France, pp. 332–338 (1999)
21. Wang, S., Li, H., Ji, C., et al.: Energy analysis for TMD-structure systems subjected to impact
loading. China Ocean Eng. 16(3), 301–310 (2002)
22. Chandrasekaran, S., Bhaskar K., Lino, H., et al.: Dynamic response behaviour of multi-legged
articulated tower with & without TMD. In: Proceedings of the International Conference on
Marine Technology, Dhaka, Bangladesh. pp. 131–136 (2010)
23. Yue, Q., Zhang, L., Zhang, W., et al.: Mitigating ice-induced jacket platform vibrations
utilizing a TMD system. Cold Reg. Sci. Technol. 56(2–3), 84–89 (2009)
24. Abe, M., Igusa, T.: Tuned mass dampers for structures with closely spaced natural frequen-
cies. Earthq. Eng. Struct. Dyn. 24, 247–261 (1995)
25. Taflanidis, A.A., Angelides, D.C., Scruggs, J.T.: Robust design optimization of mass dampers
for control of tension leg platforms. In: Proceedings of the International Offshore and Polar
Engineering Conference, Vancouver, Canada, pp. 92–99 (2008)
26. Lu, J., Mei, N., Li, Y., et al.: Vibration control of multi-tuned mass dampers for an offshore
oil platfrom. China Ocean Eng. 16(3), 321–328 (2002)
27. Taflanidis, A.A., Angelides, D.C., Scruggs, J.T.: Simulation-based robust design of mass
dampers for response mitigation of tension leg platforms. Eng. Struct. 31(4), 847–857 (2009)
28. Chandrasekaran, S., Kumar, D., Ramanathan, R.: Dynamic response of tension leg platform
with tuned mass dampers. J. Naval Architect. Mar. Eng. 10(2), 1813–8235 (2013)
29. Ma, R., Wang, J., Zhao, D.: Simulation of vibration control of offshore platforms under
earthquake loadings. In: Proceedings of the International Conference on Offshore Mechanics
and Arctic Engineering, Berlin, Germany, pp. 597–601 (2008)
30. Zhao, D., Cai, D.M., Ma, R.J.: Vibration control of offshore platforms using METMD system
under the random ocean wave forces. In: Proceedings of the Seventh ISOPE Pacific/Asia
Offshore Mechanics Symposium, Dalian, China, pp. 60–65 (2006)
31. Golafshani, A.A., Gholizad, A.: Passive devices for wave induced vibration control in
offshore steel jacket platforms. Trans. A Civ. Eng. 16(6), 443–456 (2009)
32. Jafarabad, A., Kashani, M., Parvar, M.R.A., et al.: Hybrid damping systems in offshore jacket
platforms with float-over deck. J. Constr. Steel Res. 98, 178–187 (2014)
33. Ma, R., Zhang, H., Zhao, D.: Study on the anti-vibration devices for a model jacket platform.
Mar. Struct. 23(4), 434–443 (2010)
34. Vel.ičko, J., Gaile L.: Overview of tuned liquid dampers and possible ways of oscillation
damping properties improvement. In: Proceedings of the International Scientific and Practical
Conference on Environment, Technology, Resources, Rezekne, Latvia, pp. 233–238 (2015)
35. Vandiver, J.K., Mitome, S.: Effect of liquid storage tanks on the dynamic response of offshore
platforms. Appl. Ocean Res. 1(2), 67–74 (1979)
36. Li, H., Ma, B.: Seismic response reduction for fixed offshore platform by tuned liquid damper.
China Ocean Eng. 11(2), 119–125 (1997)
37. Chen, X., Wang, L., Xu, J.: TLD technique for reducing ice-induced vibration on platforms.
J. Cold Reg. Eng. 13(3), 139–152 (1999)
38. Jin, Q., Li, X., Sun, N., et al.: Experimental and numerical study on tuned liquid dampers
for controlling earthquake response of jacket offshore platform. Mar. Struct. 20(4), 238–254
(2007)
39. Spillane, M.W., Rijken, O.R., Leverette S.J.: Vibration absorbers for deep water TLP’s.
In: Proceedings of the International Offshore and Polar Engineering Conference, Lisbon,
Portugal, pp. 210–217 (2007)
40. Bian, X.S., Leverette, S.J., Rijken, O.R.: A TLP solution for 8000 ft water depth. In:
Proceedings of the International Conference on Ocean, Offshore and Arctic Engineering,
Shanghai, China, pp. 255–262 (2010)
References 185
41. Rijken, O., Spillane, M., Leverette, S.J.: Vibration absorber technology and conceptual
design of vibration absorber for TLP in ultradeep water. In: Proceedings of the International
Conference on Ocean, Offshore and Arctic Engineering, Shanghai, China, pp. 629–638
(2010)
42. Sakai, F., Takaeda, S., Tamaki, T.: Tuned liquid column damper-new type device for
suppression of building vibrations. In: Proceedings of the International Conference on
Highrise Buildings, Nanjing, China, pp. 926–931 (1989)
43. Chaiviriyawong, P., Webster, W.C., Pinkaew, T., et al.: Simulation of characteristics of tuned
liquid column damper using a potential-flow method. Eng. Struct. 29(1), 132–144 (2007)
44. Lee, H.H., Wong, S.-H., Lee, R.-S.: Response mitigation on the offshore floating platform
system with tuned liquid column damper. Ocean Eng. 33(8–9), 1118–1142 (2006)
45. Huo, L., Li, H.: Torsionally coupled response control of offshore platform structures using
Circular Tuned Liquid Column Dampers. China Ocean Eng. 18(2), 173–183 (2004)
46. Al-Saif, K.A., Aldakkan, K.A., Foda, M.A.: Modified liquid column damper for vibration
control of structures. Int. J. Mech. Sci. 53(7), 505–512 (2011)
47. Chatterjee, T., Chakraborty, S.: Vibration mitigation of structures subjected to random wave
forces by liquid column dampers. Ocean Eng. 87(1), 151–161 (2014)
48. Lee, H.H., Juang, H.H.: Experimental study on the vibration mitigation of offshore tension
leg platform system with UWTLCD. Smart Struct. Syst. 9(1), 71–104 (2012)
49. Mousavi, S.A., Zahrai, S.M., Bargi, K.: Optimum geometry of tuned liquid column-gas
damper for control of offshore jacket platform vibrations under seismic excitation. Earthq.
Eng. Eng. Vib. 11(4), 579–592 (2012)
50. Mousavi, S.A., Bargi, K., Zahrai, S.M. Optimum parameters of tuned liquid column-gas
damper for mitigation of seismic-induced vibrations of offshore jacket platforms. Struct.
Control. Health Monit. 20(3), 422–444 (2013)
51. Hochrainer, M.J., Ziegler, F.: Control of tall building vibrations by sealed tuned liquid column
dampers. Struct. Control. Health Monit. 13(6), 980–1002 (2006)
52. Ziegler, F.: Special design of tuned liquid column-gas dampers for the control of spatial
structural vibrations. Acta Mech. 201(1), 249–267 (2008)
53. Zeng, X., Yu, Y., Zhang, L., et al.: A new energy-absorbing device for motion suppression in
deep-sea floating platforms. Energies 8(1), 111–132 (2015)
54. Pinkaew, T., Fujino, Y.: Effectiveness of semi-active tuned mass dampers under harmonic
excitation. Eng. Struct. 23(7), 850–856 (2001)
55. Spencer, B.F. Jr, Dyke, S.J., Sain, M.K., et al.: Phenomenological model of a magnetorheo-
logical damper. J. Eng. Mech. 123(3), 230–238 (1997)
56. Karkoub, M., Lamont, L.A., Chaar, L.E.: Design of a test rig for vibration control of oil
platforms using Magneto-Rheological Dampers. J. Offshore Mech. Arct. Eng. 133(4), 041302
(2011). https://doi.org/10.1115/1.4003358
57. Sarrafan, A., Zareh, S.H., Khayyat, A.A., et al.: Performance of an offshore platform with
MR dampers subjected to wave. In: Proceedings of the IEEE International Conference on
Mechatronics, Istanbul, Turkey, pp. 242–247 (2011)
58. Ji, C., Yin, Q.: Study on a fuzzy MR damper vibration control strategy for offshore
platforms. In: Proceedings of the International Conference on Offshore Mechanics and Arctic
Engineering, San Diego, USA, pp. 363–368 (2007)
59. Ji, C., Chen, M., Li S.: Vibration control of jacekt platforms with magnetorheological damper
and experimental validation. High Technol. Lett. 16(2), 189–193 (2010)
60. Wu, B., Shi, P., Wang, Q., et al.: Performance of an offshore platform with MR dampers
subjected to ice and earthquake. Struct. Control Health Monit. 18(6), 682–697 (2011)
61. Wang, S.-Q., Li, N.: Semi-active vibration control for offshore platforms based on LQG
method. J. Mar. Sci. Technol. 21(5), 562–568 (2013)
62. Sarrafan, A., Zareh, S.H., Khayyat, A.A.A., et al.: Neuro-fuzzy control strategy for an
offshore steel jacket platform subjected to wave-induced forces using magnetorheological
dampers. J. Mech. Sci. Technol. 26(4), 1179–1196 (2012)
186 References
63. Taghikhany, T., Ariana, Sh., Mohammadzadeh, R., et al.: The effect of semi-active controller
in Sirri jacket seismic vibration control under Kobe earthquake. Int. J. Mar. Sci. Eng. 3(2),
77–84 (2013)
64. Fischer, F.J., Liapis, S.I., Kallinderis, Y.: Mitigation of current-driven, vortex-induced vibra-
tions of a spar platform via “SMART” thrusters. J. Offshore Mech. Arct. Eng. 126(1), 96–104
(2004)
65. Zribi, M., Almutairi, N., Abdel-Rohman, M., et al.: Nonlinear and robust control schemes for
offshore steel jacket platforms. Nonlinear Dyn. 35(1), 61–80 (2004)
66. Zhang, B.-L., Hu, Y.-H., Tang, G.-Y.: Stabilization control for offshore steel jacket platforms
with actuator time-delays. Nonlinear Dyn. 70(2), 1593–1603 (2012)
67. Suhardjo, J., Kareem, A.: Structural control of offshore platforms. In: Proceedings of the
International Offshore and Polar Engineering Conference, Honolulu, USA, pp. 416–424
(1997)
68. Nakamura, M., Kajiwara, H., Koterayama, W., et al.: Control system design and model
experiments on thruster assisted mooring system. In: Proceedings of the International
Offshore and Polar Engineering Conference, Honolulu, USA, pp. 641–648 (1997)
69. Yamamoto, I., Matsuura, M., Yamaguchi, Y., et al.: Dynamic positioning system based on
nonlinear programming for offshore platforms. In: Proceedings of the International Offshore
and Polar Engineering Conference, Honolulu, USA, pp. 632–640 (1997)
70. Suhardjo, J., Kareem, A.: Feedback-feedforward control of offshore platforms under random
waves. Earthq. Eng. Struct. Dyn. 30, 213–235 (2001)
71. Kawano, K.: Active control effects on dynamic response of offshore structures. In: Proceed-
ings of the International Offshore and Polar Engineering Conference, Singapore, pp. 494–498
(1993)
72. Terro, M.J., Mahmoud, M.S., Abdel-Rohman, M.: Multi-loop feedback control of offshore
steel jacket platforms. Comput. Struct. 70(2), 185–202 (1999)
73. Mahadik, A.S., Jangid, R.S.: Active control of offshore jacket platforms. Int. Shipbuild. Progr.
50(4), 277–295 (2003)
74. Luo, M., Zhu, W.Q.: Nonlinear stochastic optimal control of offshore platforms under wave
loading. J. Sound Vib. 296(4–5), 734–745 (2006)
75. Suneja, B.P., Datta, T.K.: Active control of ALP with improved performance function. Ocean
Eng. 25(10), 817–835 (1998)
76. Yoshida, K., Suzuki, H., Nam, D.: Active control of coupled dynamic response of TLP hull
and tendon. In: Proceedings of the International Offshore and Polar Engineering Conference,
Osaka, Japan, pp. 98–104 (1994)
77. Ahmad, S.K., Ahmad, S.: Active control of non-linearly coupled TLP response under wind
and wave environments. Comput. Struct. 72(6), 735–747 (1999)
78. Alves, R.M., Battista, R.C., Albrecht, C.H.: Active control for enhancing fatigue life of
TLP platform and tethers. In: Proceedings of the International Congress of Mechanical
Engineering, Sao Paulo, Brazil (2003)
79. Li, H.-J., Hu, S.-L., Jakubiak, C.: H2 active vibration control for offshore platform subjected
to wave loading. J. Sound Vib. 263(4), 709–724 (2003)
80. Wang, W., Tang, G.-Y.: Feedback and feedforward optimal control for offshore jacket
platforms. China Ocean Eng. 18(4), 515–526 (2004)
81. Ma, H., Tang, G.-Y., Zhao, Y.-D.: Feedforward and feedback optimal control for offshore
structures subjected to irregular wave forces. Ocean Eng. 33(8–9), 1105–1117 (2006)
82. Ma, H., Tang, G.-Y., Hu, W.: Feedforward and feedback optimal control with memory for
offshore platforms under irregular wave forces. J. Sound Vib. 328(4–5), 369–381 (2009)
83. Zhang, B.-L., Liu, Y.-J., Han, Q.-L., et al.: Optimal tracking control with feedforward
compensation for offshore steel jacket platforms with active mass damper mechanisms. J.
Vib. Control 22(3), 695–709 (2016)
84. Zhang, B.-L., Liu, Y.-J., Ma, H., et al.: Discrete feedforward and feedback optimal tracking
control for offshore steel jacket platforms. Ocean Eng. 91, 371–378 (2014)
References 187
85. Zhang, B.-L., Feng, A.-M., Li, J.: Observer-based optimal fault-tolerant control for offshore
platforms. Comput. Electr. Eng. 40(7), 2204–2215 (2014)
86. Li, H., Hu, S.-L.J.: Optimal active control of wave-induced vibration for offshore platform.
China Ocean Eng. 15(1), 1–14 (2001)
87. Ji, C., Li, H., Wang, S.: Optimal vibration control strategy for offshore platforms. In:
Proceedings of the International Offshore and Polar Engineering Conference, Kitakyushu,
Japan, pp. 91–96 (2002)
88. Yang, J.S.: Robust mixed H2 /H∞ active control for offshore steel jacket platform. Nonlinear
Dyn. 78(2), 1503–1514 (2014)
89. Zhang, B.-L., Tang, G.-Y.: Active vibration H∞ control of offshore steel jacket platforms
using delayed feedback. J. Sound Vib. 332(22), 5662–5677 (2013)
90. Zhang, B.-L., Ma, L., Han, Q.-L.: Sliding mode H∞ control for offshore steel jacket platforms
subject to nonlinear self-excited wave force and external disturbance. Nonlinear Anal. Real
World Appl. 14(1), 163–178 (2013)
91. Zhang, B.-L., Huang, Z.-W., Han, Q.-L.: Delayed non-fragile H∞ control for offshore steel
jacket platforms. J. Vib. Control 21(5), 959–974 (2015)
92. Zhou, Y.-J., Zhao, D.-Y.: Neural network-based active control for offshore platforms. China
Ocean Eng. 17(3), 461–468 (2003)
93. Chang, S., Kim, D., Chang, C., et al.: Active response control of an offshore structure under
wave loads using a modified probabilistic neural network. J. Mar. Sci. Technol. 14(2), 240–
247 (2009)
94. Kim, D.H.: Neuro-control of fixed offshore structures under earthquake. Eng. Struct. 31(2),
517–522 (2009)
95. Kim, D.H.: Application of lattice probabilistic neural network for active response control of
offshore structures. Struct. Eng. Mech. 31(2), 153–162 (2009)
96. Cui, H., Hong, M.: Adaptive inverse control of offshore jacket platform based on grey
prediction. In: Proceedings of the International Conference on Digital Manufacturing and
Automation, Zhangjiajie, China, pp. 150–154 (2011)
97. Li, X., Yu, X., Han, Q.-L.: Stability analysis of second-order sliding mode control systems
with input-delay using Poincare map. IEEE Trans. Autom. Control 58(9), 2410–2415 (2013)
98. Zhang, B.-L., Tang, G.-Y., Ma, H.: Optimal sliding mode control with specified decay rate for
offshore steel jacket platforms. China Ocean Eng. 24(3), 443–452 (2010)
99. Zhang, B.-L., Han, Q.-L., Zhang, X.-M., et al.: Integral sliding mode control for offshore steel
jacket platforms. J. Sound Vib. 331(14), 3271–3285 (2012)
100. Zhang, X.-M., Han, Q.-L., Han, D.-S.: Effects of small time-delays on dynamic output
feedback control of offshore steel jacket structures. J. Sound Vib. 330(16), 3883–3900 (2011)
101. Zhang, B.-L., Han, Q.-L., Zhang, X.-M., et al.: Sliding mode control with mixed current and
delayed states for offshore steel jacket platforms. IEEE Trans. Contr. Syst. Technol. 22(5),
1769–1783 (2014)
102. Nourisola and Ahmadi Nourisola, H., Ahmadi, B.: Robust adaptive sliding mode con-
trol based on wavelet kernel principal component for offshore steel jacket platforms
subject to nonlinear wave-induced force. J. Vib. Control (2014). https://doi.org/10.1177/
1077546314553319
103. Nourisola, H., Ahmadi, B., Tavakoli, S.: Delayed adaptive output feedback sliding mode
control for offshore platforms subject to nonlinear wave-induced force. Ocean Eng. 104, 1–9
(2015)
104. Robinett, R.D., Petterson, B.J., Fahrenholtz, J.C.: Lag-stabilized force feedback damping. J.
Intell. Robot. Syst. 21(3), 277–285 (1998)
105. Zhao, Y.-Y., Xu, J.: Effects of delayed feedback control on nonlinear vibration absorber
system. J. Sound Vib. 308(1–2), 212–230 (2007)
106. Zhang, D., Han, Q.-L., Jia, X.-C.: Network-based output tracking control for a class of T-S
fuzzy systems that can not be stabilized by non-delayed output feedback controllers. IEEE
Trans. Cybern. 45(8), 1511–1524 (2015)
188 References
107. Zhang, B.-L., Han, Q.-L.: Robust sliding mode H∞ control using time-varying delayed states
for offshore steel jacket platforms. In: Proceedings of the IEEE International Symposium on
Industrial Electronics, Taipei, Taiwan, pp. 1–6 (2013)
108. Zhang, B.-L., Han, Q.-L., Huang, Z.-W.: Pure delayed non-fragile control for offshore steel
jacket platforms subject to non-linear self-excited wave force. Nonlinear Dyn. 77(3), 491–502
(2014)
109. Sakthivel, R., Selvaraj, P., Mathiyalagan, K., et al.: Robust fault-tolerant H∞ control for
offshore steel jacket platforms via sampled-data approach. J. Franklin Ins. 352(6), 2259–2279
(2015)
110. Sakthivel, R., Santra, S., Mathiyalagan, K., et al.: Robust reliable sampled-data control for
offshore steel jacket platforms with nonlinear perturbations. Nonlinear Dyn. 78(2), 1109–
1123 (2014)
111. Sivaranjani, K., Rakkiyappan, R., Lakshmanan, S., et al.: Robust stochastic sampled-data
control for offshore steel jacket platforms with non-linear perturbations. IMA J. Math. Control
Info. (2015). https://doi.org/10.1093/imamci/dnv046
112. Zhang, B.-L., Meng, M.-M., Han, Q.-L., et al.: Robust non-fragile sampled-data control for
offshore steel jacket platforms. Nonlinear Dyn. 83(4), 1939–1954 (2016)
113. Huang, S., Cai, M., Xiang, Z.: Robust sampled-data H∞ control for offshore platforms subject
to irregular wave forces and actuator saturation. Nonlinear Dyn. (2017). https://doi.org/10.
1007/s11071-017-3404-6
114. Peng, C., Han, Q.-L., Yue, D.: To transmit or not to transmit: a discrete event-triggered
communication scheme for networked Takagi-Sugeno fuzzy systems. IEEE Trans. Fuzzy
Syst. 21(1), 164–170 (2013)
115. Zhang, X.-M., Han, Q.-L.: Event-triggered dynamic output feedback control for networked
control systems. IET Control Theory & Appl. 8, 226–234 (2014)
116. Zhang, X.-M., Han, Q.-L.: Event-based H∞ filtering for sampled-data systems. Automatica
51, 55–69 (2015)
117. Ge, X., Yang, F., Han, Q.-L.: Distributed networked control systems: a brief overview. Inf.
Sci. 380, 117–131 (2017)
118. Zhang, X.-M., Han, Q.-L., Yu, X.: Survey on recent advances in networked control systems.
IEEE Trans. Ind. Informat. 12(5), 1740–1752 (2016)
119. Zhang, X.-M., Han, Q.-L., Zhang, B.-L.: An overview and deep investigation on sampled-
data-based event-triggered control and filtering for networked systems. IEEE Trans. Ind.
Informat. 13(1), 4–16 (2017)
120. Zhang, B.-L., Han, Q.-L.: Network-based modelling and active control for offshore steel
jacket platforms with TMD mechanisms. J. Sound Vib. 333(25), 6796–6814 (2014)
121. Jiang, X., Han, Q.-L.: On H∞ control for linear systems with interval time-varying delay.
Automatica 41(12), 2099–2106 (2005)
122. Jiang, X., Han, Q.-L.: Delay-dependent robust stability for uncertain linear systems with
interval time-varying delay. Automatica 42(6), 1059–1065 (2006)
123. Zhang, B.-L., Han, Q.-L., Zhang, X.-M.: Event-triggered H∞ reliable control for offshore
structures in network environments. J. Sound Vib. 368, 1–21 (2016)
124. Yue, D., Tian, E., Han, Q.-L.: A delay system method for designing event-triggered controllers
of networked control systems. IEEE Trans. Automa. Control 58(2), 475–481 (2013)
125. Peng, C., Han, Q.-L.: A novel event-triggered transmission scheme and L2 control co-design
for sampled-data control systems. IEEE Trans. Automa. Control 58(10), 2620–2626 (2013)
126. Sarpkaya, T., Isaacson, M. (eds.): Mechanics of Wave Forces on Offshore Structures. Van
Nostrand Reihhold, New York (1981)
127. Chakrabarti, S.K. (ed.): Hydrodynamics of Offshore Structures. Springer, Berlin (1987)
128. Xie, L.: Output feedback H∞ control of systems with parameter uncertainty. Int. J. Control
63(4), 741–750 (1996)
129. Lancaster, P., Lerer, L., Tismenetsky, M.: Factored forms for solutions of AX − XB = C and
X − AXB = C in companion matrices. Linear Algebra Appl. 62, 19–49 (1984)
References 189
130. Gahinet, P., Apkarian, P.: A linear matrix inequality approach to H∞ control. Int. J. Robust
Nonlinear Control 4, 421–448 (1994)
131. Han, Q.-L.: Absolute stability of time-delay systems with sector-bounded nonlinearity.
Automatica 41, 2171–2176 (2005)
132. Zhang, X.M., Wu, M., She, J.H., et al.: Delay-dependent stabilization of linear systems with
time-varying state and input delays. Automatic 41(8), 1405–1412 (2005)
133. Peng, C., Fei, M.-R.: An improved result on the stability of uncertain T-S fuzzy systems with
interval time-varing delay. Fuzzy Sets Syst. 212, 97–109 (2012)
134. Zhang, X.M., Han, Q.-L.: Novel delay-derivative-dependent stability criteria using new
bounding techniques. Int. J. Robust Nonlinear control 23(13), 1419–1432 (2013)
135. Spurgeon, S., Edwards, C. (eds.): Sliding Mode Control: Theory and Applications. Taylor and
Francis, London (1998)
136. Yu, X., Kaynak, O.: Sliding mode control with soft computing: a survey. IEEE Trans. Ind.
Electron. 56(9), 3275–3285 (2009)
137. Ghaoui, L.E., Oustry, F., AitRami, M.: A cone complementarity linearization algorithms for
static output feedback and related problems. IEEE Trans. Automa. Control 42(8), 1171–1176
(1997)
138. Robinett, R.D., Petterson, B.J., Fahrenholtz, J.C.: Lag-stabilized force feedback damping. J.
Intell. Robot. Syst. 21(3), 277–285 (1998)
Index
N Seismic response, 2
Networked control systems (NCSs), 152 Semi-active control, 1, 6–7, 181
Networked-induced delays, 14, 133, 140–145, Sensor, 7, 13, 115, 132, 133, 181
150–153, 156, 170, 174, 176, 177, 180, Sliding mode control (SMC), 2, 10–11, 15, 49,
181 54–56, 58, 59, 62, 63, 67, 68, 71, 76,
Neural network, 7, 10, 68 79–82, 87–89, 107, 128, 146, 148–150
Non-fragile control, 13, 92, 93, 101, 107 Stability, 2, 7, 8, 10, 14, 51–53, 74, 94, 112,
119, 131, 134, 136, 155, 159, 164, 167,
180, 181
O Stability analysis, 49–52, 134–140, 153
Ocean engineering, 181 Stabilization, 107, 177, 179
Offshore platform, 1–15, 17–31, 33, 36, 38–41, Stochastic control, 48
43–49, 56–58, 63, 67–69, 71, 78, 79, 82,
83, 86–89, 91, 92, 100, 103, 107–110,
116, 123, 126–129, 131, 132, 134, 138,
T
140–142, 144–148, 150–153, 156, 169,
Tension leg platform (TLP), 1, 4–6, 8, 10
171, 172, 179–181
Time-delay, 2, 9, 12, 13, 15, 71, 79–83, 86–89,
Offshore structure, 1, 18–20, 132, 153, 155,
91, 92, 102–104, 106–109, 115, 116,
156, 167, 169–171, 173, 174, 176, 177,
124, 126–128, 133, 157
179, 180
TLP, see Tension leg platform
Optimal control, 2, 7–10, 33, 36–38, 44–48,
Tracking control, 9, 15, 21, 31–48, 173
68, 173
Tuned mass damper (TMD), 4–5, 8, 9, 11–15,
Output feedback control, 9, 11–13, 91, 109
17, 22–29, 56, 57, 78, 107, 109, 113,
131, 132, 140, 153
P
Passive control, 1–7
V
Vibration control, 1, 2, 5–7, 10, 15, 181
R Vibration mitigation, 5
Robust controller design, 89 Vibration reduction, 4, 11
Robust H∞ control, 10, 49, 54–56, 62, 89,
91–108
Robust stability analysis, 94
W
Wave load, 5
S
Sampled-data control, 2, 13, 129
Schur complements, 29, 51, 74, 95, 113, 120, Z
138, 163, 164 Zero order hold (ZOH), 156, 176