Professional Documents
Culture Documents
BY
WG CDR D VISWANATH
OF
MAY 2008
4
CERTIFICATE
This is to certify that the Seminar paper titled SLIDING MODE CONTROL
– A SURVEY submitted by Wg Cdr D Viswanath, as Seminar II in Semester II in
partial fulfillment of the requirements for the award of the degree of Master of
Engineering (Aerospace) by the Defence Institute of Advanced Technology,
(Deemed University), Pune is a bonafide record of the student’s own work carried
out by him (except where indicated) under my supervision and guidance.
(Prof SB Phadke)
Sc ‘F’
Guide
Dept of Aerospace Engg
Defence Institute of
Advanced Technology
Date : May 2008 Girinagar, Pune-411025
COUNTERSIGNED
(Prof SB Phadke)
Scientist ‘F’
Chairman, Dept of Aerospace Engg
Defence Institute of Advanced Technology
5
ACKNOWLEDGEMENTS
(D Viswanath)
Wg Cdr
SWC-43
6
ABSTRACT
Variable structure control was first proposed and elaborated in the early
1950’s in the Soviet Union by Emelyanov and several co-researchers. In their
pioneer works, the plant considered was a linear second-order system modeled
in phase variable form. Since then, VSC has developed into a general design
method being examined for a wide spectrum of system types including nonlinear
systems, multi-input/multi-output systems, discrete-time models, large-scale and
infinite-dimensional systems, and stochastic systems. The most distinguished
feature of VSC is its ability to result in very robust control systems; the system is
completely insensitive to parametric uncertainty and external disturbances or
“invariant”. The sliding mode (SMC) is the major mode of operation in variable
structure systems.
However there are many problems faced in the attempt to employ the
properties of sliding modes for the design of automatic control systems. Various
publications on the matter show diverse viewpoints leading to diverse sliding
mode equations. This paper is an attempt to survey the current developments
vis-à-vis remedies to the problems in SMC.
7
TABLE OF CONTENTS
Page
ABSTRACT i
TABLE OF CONTENTS ii
CHAPTER
1. INTRODUCTION
2. SLIDING MODE CONTROL (SMC)
3. CHATTERING PROBLEM IN SMC AND REMEDIES
4. REACHING PHASE AND ITS ELIMINATION
5. NEED FOR BOUNDS OF UNCERTAINTY - ADAPTIVE CONTROL
6. NEED FOR FULL STATE VECTOR AND USE OF OBSERVER
7. INVARIANCE AND MATCHING CONDITIONS – BACKSTEPPING
8. TIME DELAY CONTROL AND INERTIAL DELAY CONTROL
9. CURRENT DEVELOPMENTS IN SMC APPLICATIONS
10. CONCLUSION
REFERENCES
8
LIST OF FIGURES
CHAPTER ONE
INTRODUCTION
1. Variable structure control originated in the early 1950’s in the Soviet Union.
In their pioneer works, Emelyanov and several co-researchers [43], [79],
considered the plant as a linear second-order system modeled in phase variable
form. The most important property of VSC is its ability to result in very robust
control systems; in many cases invariant control systems. Invariance means that
the system is completely insensitive to parametric uncertainty and external
disturbances. Successful results have been reported in terms of eliminating
disturbances, addressing nonlinearities, and achieving acceptable control in the
presence of modeling errors. During the last three decades since the publication
of the survey paper in the IEEE TRANSACTIONS ON AUTOMATIC CONTROL
in 1977 [1], significant work has been carried out on variable structure systems
(VSS) and sliding mode control (SMC). The sliding mode (SMC) is the major
mode of operation in variable structure systems. Today, research and
development continue to apply VSC to a wide variety of engineering systems.
Numerous VSC papers have been published in this area. The issues studied
include existence of the sliding mode, stability of the sliding mode, effects of
system parameter perturbations and outside disturbances and study of systems
having unmeasurable state variables. Two more survey papers [2],[3] have been
published in 1993 and 1999 respectively.
3. Sliding Mode Control (SMC) is best explained with a second order system
whose phase plane plot clearly shows the response of the system when in sliding
mode. Consider a system represented by
.
x1 x2
.
____________________________(1.1)
x 2 a1 x1 a2 x2 bu
Where a1 and a2 and b are parameters that are not exactly known but some
knowledge about their range is known i.e.,
_ _
ai ai a i i=1,2,… and 0 b b b 1 .
_ _
Our objective is to design a control ‘u’ in such a way that the response of the
system represented by equation (1.1) is independent of a1, a2 and b after some
time t=t1.
4. Consider a line drawn through the origin of the phase plane plot across
the second and fourth quadrants as shown in figure(Fig (1.1)) below. It can be
shown that all trajectories which lie in these two quadrants are stable i.e., they
asymptotically tend towards origin with time. This line is designated as the sigma
line and is represented by the equation,
c1 x1 x2 0 ____________________________(1.2)
where c1 is a user chosen constant. The control is to be designed such that the
system trajectory is attracted to the sigma line and once the trajectory intersects
it, keep switching the control to keep the trajectory on the sigma line so that it
does not leave it. As a result after a certain finite time, the system is governed by
11
the equation of the sigma line only and not on the system equation. Thus the
dynamics will be independent of the system parameters.
5. One such trajectory is shown in the phase plane plot of Fig (1.2). Along
the curve PQ the system is affected by equation (1.1). This is called the reaching
phase. Along the line QO the system is affected by equation (1.2). This is called
the sliding mode (SM) phase. The control initially applied in equation (1.1) helps
the system trajectory to move towards the sigma line. Thereafter the control is
switched so that the trajectory does not leave the sigma line. The switching
action can be explained as follows: -
(a) When the trajectory (consider curve PQ) crosses the sigma line,
sigma becomes greater than zero. We now use a control that makes the
rate of change of sigma become less than zero so that the trajectory is
brought back to the sigma line.
12
(b) If now the trajectory crosses the sigma line again but from the
opposite direction, sigma becomes less than zero. We use a control
that makes the rate of change of sigma to become greater than
zero so that the trajectory is again brought back to the sigma line.
6. It is obvious that for the trajectory to follow the sigma line very smoothly
the switching action has to be very fast. In practice, switching delays exist and
this causes the phenomenon called chattering which is the major and most
important disadvantage of SMC systems. The chattering phenomenon [3] is
generally perceived as motion which oscillates about the sliding manifold. There
are two possible mechanisms which produce such a motion.
(a) First, when the switching device is not switching ideally at an infinite
frequency, the presence of parasitic dynamics in series with the plant
causes a small amplitude high-frequency oscillation to appear in the
neighborhood of the sliding manifold. These parasitic dynamics represent
13
the fast actuator and sensor dynamics which, are often neglected in the
open-loop model used for control design if the associated poles are well
damped, and outside the desired bandwidth of the feedback control
system. Generally, the motion of the real system is close to that of an ideal
system in which the parasitic dynamics are neglected, and the difference
between the ideal and the real motion, which is on the order of the
neglected time constants, decays rapidly. However, the theory is not
applicable for VSS since they are governed by differential equations with
discontinuous right hand sides. The interactions between the parasitic
dynamics and VSC generate a non-decaying oscillatory component of
finite amplitude and frequency, and this is generically referred to as
chattering.
(b) Second, the switching non-idealities alone can cause such high-
frequency oscillations.
(c) SMC requires that the full state vector be available for the control to
be applied effectively. But states may not be available always.
(d) In the reaching phase the system parameters affect the trajectory
which is undesirable.
CHAPTER TWO
CLASSIFICATION OF SMC
1. Classification
Sliding mode control can be classified into the following modes [3]: -
(a) Continuous-time Sliding Mode
(b) Discrete Time Sliding Mode and
(c) Sampled Data Sliding Mode
problems in SMC. Observers are most likely constructed in discrete time for any
real life control implementations. However, in order for these observer based
design to work, the sampling rate has to be relatively high since the notion of
continuous-time sliding mode is still applied.
5. Classification by methods
Sliding mode control can be classified into the following methods based on
sliding surface used: -
17
CHAPTER THREE
3. Chattering Refreshed
(a) The ‘sat’ or saturation function: Thus the control takes the form
20
un sgn( SB); SB
SB _____________________ (3.7)
un ; SB
(b) The sigmoid function: Here the control takes the form
SB
un ; SB
SB
_____________________ (3.8)
SB
un ; SB
(c) The third method involves the use of a conical boundary layer
centred symmetrically at the origin instead of the cylindrical boundary
layer dictated by sat function.
To make the system behavior to be as close to that of the ideal sliding mode,
particularly when an unknown disturbance is to be rejected, sufficiently high
magnitude of control signal gain is needed. In the absence of disturbance the
boundary layer thickness can be enlarged further while magnitude of control is
decreased, to reduce the oscillatory behavior or chattering about the sliding
manifold. However, this reduces the system to one with no sliding mode inside
the boundary layer and hence the system is no longer robust to uncertainties
21
inside the boundary layer. Thus though the proposed method has shortcomings.
The effectiveness of boundary layer control is immediately challenged when
realistic parasitic dynamics are considered. [4]. Parasitic dynamics must be
carefully modeled and considered in the feedback design in order to avoid
instability inside the boundary layer which leads to chattering.
Since it known that the essential triggering mechanism for chattering is the
interactions of the switching action with the parasitic dynamics, an approach
which utilizes asymptotic observers to construct a high-frequency by pass loop
has been proposed [5]. This design exploits a localization of the high frequency
phenomenon in the feedback loop by introducing a discontinuous feedback
control loop which is closed through an asymptotic observer of the plant [6].
Since the model imperfections of the observer are supposedly smaller than those
in the plant, and the control is discontinuous only with respect to the observer
variables, chattering is localized inside a high-frequency loop which bypasses the
plant. However, this approach assumes that an asymptotic observer can indeed
be designed such that the observation error converges to zero asymptotically.
6. Disturbance Compensation
In SMC, the main purpose of sliding mode is to make the system robust to
disturbances unknown parametric perturbations. Building on the observer based
SMC, a sliding mode disturbance estimator which uses sliding mode to estimate
the unknown disturbances and parametric uncertainties was introduced [7]. In
this approach, the control law consists of a conventional continuous feedback
control component, and a component derived from the SM disturbance estimator
for disturbance compensation. If the disturbance is sufficiently compensated,
there is no need to evoke a discontinuous feedback control to achieve sliding
22
design such that the matching condition is indeed satisfied in the extended space.
Another alternative approach is to utilize sliding mode to estimate the disturbance
for compensation as discussed earlier. Since sliding mode is not introduced
primarily to reject disturbances, the matching conditions are of no significance in
this design. This approach may also resolve the limitations imposed by actuator
bandwidth constraints on the design of sliding mode based controllers.
1. The motion of a VSC system includes two phases: the reaching phase
and the sliding phase. During the reaching phase, the system state is pushed
toward the switching surfaces. During this period, however, the tracking error
cannot be controlled directly and the system response is sensitive to parameter
uncertainties and noise. Thus, one ideally would like to shorten the duration or
even eliminate the reaching phase. Several methods have been proposed to
completely eliminate the reaching phase. Such works include Chang and
Hurmuzlu [11], Choi et. al. [12], Bartoszewicz [13], and Roy and Olgac [14]. All of
the past methods consider (except the one presented in Roy and Olgac [14])
second order systems. These approaches reformulate the sliding domain
equations such that the initial state of the system resides on the modified sliding
domain. When this is achieved, and since the system state never leaves the
sliding domain, the reaching phase is eliminated from the motion of the resulting
system. Naturally, one has to take additional steps to ensure that the modified
sliding domain converges to the original one as the state evolves in time. This is
realized by replacing the static sliding domain equations in the classical case with
dynamic ones. This can be done by formulating a modified sliding domain
equation that exponentially decays to the original one (Chang and Hurmuzlu [11]).
Alternatively, one may achieve the same goal by replacing the exponential decay
by a series of rotations and translations sequentially applied to the sliding surface
(Roy and Olgac [14]).
4. The main difference can be observed in the control effort that is required
to achieve tracking using the individual approaches. The rotation/translation
approach seems to have the most problematic profile when it comes to practical
implementation. For this case, the control signal includes sudden jumps due to
the discontinuous translations in the sliding surface. Compared to the other two,
the continuously sliding approach requires significantly less control action to
achieve the required tracking performance. We should note that although some
of the conclusions are based only on a specific example, they should also remain
valid for more general cases. The modified surface exponentially converges to
the original one. Once the system state reaches the sliding surface it never
26
leaves it. Thus, this feature may be used to accurately predict the system
response without solving the differential equations. Using this advantage, we
have developed a standard optimization procedure to optimize convergence
parameters. In conclusion, compared to existing approaches, the main
advantage of the proposed controller generates a continuous and an optimized
control signal.
27
CHAPTER FIVE
The handle uS which ensures equation (5.4) is satisfied under any condition is
given by
uS ( SB)1 K _______________________________________ (5.8)
Thus the sliding condition of equation (5.4) is satisfied if the control is chosen as
per the control law of equation (5.5) under substitution from equations (5.6) and
(5.7).
3. In the above design, the continuous positive scalar valued function rho
may not be easily obtained due to complex structure of the uncertainties and the
magnitude of the external disturbance in particular cannot be simply estimated.
Yoo and Chung [20] modified assumption 5.2 which is stated below:
Assumption 5.3. There are positive constants c0 and c1 such that
29
e( x, u, t ) c0 c1 x ________________________________ (5.9)
Where c0^ and c1^ are the adaptive parameters about c0 and c1,
respectively.
Now consider the simple adaptation laws for the upper bound of the norm
e( x, u, t ) such that
c0 ( x, u , t ) q0 BT S T
______________________________ (5.11)
c1 ( x, u, t ) q1 BT S T x
Where q0 and q1 are adaptation gains with positive values. Since c0 and
c1 are constant the adaptive parameters can be obtained by integrating equation
(5.10),
t
c0 ( x, u, t ) c0i q0 BT S T dt
t0
_________________________ (5.12)
t
c1 ( x, u, t ) c1i q1 BT S T x dt
t0
By choosing appropriate q0 , q1 and c0i , c1i , we can adjust the rate of
5. The use of boundary layer technique may result in the estimated gain to
grow unboundedly inside the boundary layer since the restriction on the sliding
surface cannot always be achieved [21]. Yuri, Su and Stepanenko propose an
improved adaptation law, the stability analysis of which shows that this method
guarantees boundedness of both the state of the plant and the adaptive gain
when boundary layer technique is applied.
31
CHAPTER SIX
1. Most modern control laws need the full state vector for their
implementation. In general, the full state is not accessible making these laws
unimplementable. Even when the states are accessible the cost of the sensors
could be very high. This makes the use of state estimators mandatory. Another
problem that real plants present is that their models are not known accurately.
For such plants the classical Luenberger observer does not work, thus further
compounding the problem of state estimation.
3. The concept of sliding mode control has been extended to the problem of
state estimation by an observer, for linear systems, uncertain linear systems and
nonlinear systems. Using the same design principles as for variable structure
control, the observer trajectories are constrained to evolve after a finite time on a
suitable sliding manifold by the use of a discontinuous output injection signal (the
sliding manifold is usually given by the difference between the observer and the
system output). Subsequently the sliding motion provides an estimate
(asymptotically or in finite time) of the system states. Sliding mode observers
32
4. The Lyapunov based approach of Walcott and Zak [22, 23, 24] considered
the problem of state observation in presence of bounded uncertainties/UI. Slotine
[25] examined the potential use of sliding surfaces for observer design for
systems in companion form with extension to non-linear systems. Lopez in [26]
formulated an alternative form of sliding mode observer wherein the output
disturbances are transformed into state disturbances, avoiding noise
amplification. Similar work was seen in [27] wherein the UI and model
uncertainties were considered as a fictitious state added to the slotine like
structure [25]. Chen [29] proposed a disturbance observer which required the
states of the system. There are other disturbance observers also in literature [28]
which also require the states in order to estimate the disturbances. In general,
the states are not available. The state estimator is possible using Luenberger
type [30] of observers provided the system does not have uncertainties. Thus, we
have a situation that the state estimation can be done in the absence of
disturbances and disturbance estimation is possible if the states are available.
5. Luenberger Observer
The classical Luenberger observer is the fundamental state observer and all
other observers are extensions of this basic structure. Consider a plant defined
by,
x = Ax + Bu ___________________________________ (6.1)
y = Cx
Assuming the state x to be approximated by the state ˆx of the dynamic model,
x A x Bu L( y x) _________________________________ (6.2)
and denoting e = x − ˆx, the state error dynamics is deduced as,
33
e ( A LC )e _____________________________ (6.3)
Thus, the dynamic behavior of the error vector is determined by the eigen values
of (A − LC). Thus, by choosing L such that (A − LC) is stable, the error dynamics
will reduce to zero asymptotically.
The problem with this observer is that it fails when the output is sensed in
presence of model uncertainties and/or sensor noise. To examine the reason
consider the plant with lumped uncertainties d as,
.
x Ax Bu Bd __________________________ (6.4)
y Cx n0
and the observer modeled as eqn. 2.3, the error dynamics is thus deduced as,
e ( A LC )e Ln0 Bd ____________________ (6.5)
The last terms shows that the proportional gain observer tends to amplify the
measurement noise by L and that the lumped uncertainty d affects the
convergence.
Where, z1 and are the estimates of the unknown disturbance and the internal
state of the nonlinear observer respectively and p(x) is a nonlinear function to be
designed. The nonlinear observer gain l(x) is defined as,
p(x)
l(x) = ______________________________ (6.9)
x
Let, the disturbance estimation error be defined as,
d z1 ___________________________ (6.10)
Thus, it can be shown that, under the assumption that the disturbances are slow
varying, z1 approaches d exponentially if p(x) is chosen such that,
p(x)
+ g2(x) =0 _______________________ (6.11)
x
is globally exponentially stable for all x 2 <n. As far as the stability of the
p(x)
estimation error is concerned, any nonlinear vector-valued function l(x) =
x
such that equation (7.11) is asymptotically stable and can be chosen. After l(x)
has been chosen, p(x) is found by integration.
8. Limitations of Chen’s DO
.
The problem with this observer is that it fails when the assumption of d 0 is
.
violated. Thus, under the assumption that d 0 we can write,
d z1
_____________________________________________ (6.12)
d C1
or
Many controllers need information of the full state vector. One more requirement
is the estimation of uncertainties as well. This requirement of obtaining the
estimate of uncertainty as well as states in an integrated manner is met by the
Extended State Observer (ESO) [31], [32]. The ESO is an observer which can
estimate the uncertainties along with the states of the system enabling
disturbance rejection or compensation. Unlike traditional (linear or nonlinear)
observers, the ESO estimates the effect of uncertainties, unmodeled dynamics
and external disturbances acting on the system as an extended state of the
original system. Hence, it can be viewed as an unknown input observer or
disturbance observer. Since the observer estimates the uncertainty as an
extended state of the original system, it is designated as Extended State
Observer. Its merit is that it is relatively independent of mathematical model of
the plant, performs better and is simpler to implement. In Ref. [31], a comparison
study of performances and characteristics of three advanced state observers
namely high gain observer, ESO and Sliding Mode Observer is presented and it
is shown that over all the ESO is much superior in dealing with uncertainties,
disturbances and sensor noise. Several diverse applications of ESO based
control strategies have appeared in literature. Control of induction motor drive
[33], aircraft attitude control [34], hydraulic position servo system [35], and
torsional vibration control of main drive system of a rolling mill [36] are some
examples to mention.
36
CHAPTER SEVEN
MATCHING CONDITIONS
2. The Model Of The System. The multi input systems described with
the set of n differential first order equations will be discussed.
.
x Ax Bu DF (7.1)
X n state vector-column. It is understood that all its elements are
available for forming a control function.
B (n×m) constant matrix, with linearly independent columns
U m control vector-column. Its elements, called control functions, are
linearly independent.
D (n ×L) matrix with linearly independent columns
F L vector-column of disturbances, with linearly independent
elements
A (n × m) matrix
is given by substituting equation (7.1) as
.
S x SAx SBu SDF (7.4)
If the matrix SB is nonsingular, u, is determinable in a unique manner as:
u (SB)1 (SAx SDF ) (7.5)
Substituting this value of u in (7.1), the sliding mode equations are obtained as:
.
x ( I B(SB)1 S ) Ax ( I B(SB)1 S ) DF (7.6)
where I is the unity matrix.
It is seen from (7.6) that disturbances F, in general, act in equations of the sliding
mode motion. There are two sorts of disturbance influences on that kind of
motion:
(b) The values of the disturbances after the sliding mode begins, but
only if F acts in (7.6).
Equation (7.7) can be satisfied for all the possible values of F, if all the columns
of D are linear combinations of the columns of B i.e.,
DF B (7.8)
This requirement is represented by the equation
rank [B, D] = rank B (7.9)
where [B, D] is a matrix composed of all the columns of D and B.
There are many systems with nonlinearities known from physical laws, such as
kinematic nonlinearities, or energy, flow and mass balance nonlinearities. Some
of these nonlinearities may appear multiplied with unknown parameters and give
rise to the problem of controlling nonlinear systems with parametric uncertainty.
For a broader class of systems, the nonlinearities themselves may be unknown.
Such difficult problems may still be tractable if the uncertainties are within some
known nonlinear bounds, the so called nonlinear interval uncertainties.
CHAPTER EIGHT
1. Time Delay Control was proposed as a robust control method for systems
with unknown dynamics by Yousuf-Toumi and Ito [38,39]. In recent years many
methods have been reported for designing control for time-delay systems [40],
[41] and criteria for the stability of time-delay systems [42], [43] have been
developed. Time-delays may appear in many ways: Delays in measurement of
system variables including physical properties of equipment used in the system
or signal transmission, delays in control which arise in many chemical processes
and radiation problems in physics.
.
x(t L) Ax(t L) Bu(t L) Be(t L) ___________________ (8.2)
Thus
.
e(t L) B [ x(t L) Ax(t L) Bu (t L)] ______________ (8.3)
Assumption 2. The interval L is very small so that system does not change.
If we define e e(t L) , then e( x, u, t ) e( x, u, t L) . Hence
e e(t L) _____________________________________ (8.4)
Design of Control. A conventional sliding mode behaviour would have a sliding
surface dynamics of the form
Sx _________________________________________ (8.5)
The sliding condition is given by
0 _________________________________________ (8.6)
Choosing the control as
u ueq un ____________________________________________ (8.7)
Thus the control of equation of (8.7) using (8.8) and (8.9), when substituted in
equation (8.1), nullifies the effect of uncertainties and disturbances. Also, the
control does not have any discontinuity.
(a) TDC inherently requires that all the states and their derivatives be
available for feedback.
Defining u as in equation (8.7) , the first part for compensating known terms can
be given as
ueq (SB)1 SAx K ________________________ (8.12)
Thus e is given by
e ( SB)1 ( K ) un ________________________ (8.14)
7. TDC adopts an estimation of this signal by using a time delay in the time
domain. However, in IDC, the control signal and states are used to observe the
uncertainties and disturbances.
Assumption. Let Gf(s) be a strictly proper low-pass filter with unity steady-
state gain and broad enough bandwidth.
The signal of equation (8.14) is passed through this filter. Thus
mathematically
e( x, u, t ) [( SB) 1 ( K ) un ]G f ( s) _______________ (8.15)
Assuming that the frequency range of the system dynamics and the external
disturbances is limited by wf, then the ideal low pass filter Gf has a low frequency
gain (w<wf) of 1 (making 1-Gf(s) zero) and a high frequency gain (w>wf) of zero
44
(making Gf zero). Hence the estimation error of the uncertainty and disturbance
will be approximately zero at both high and low frequency ranges.
1
8. Let G f ( s) ___________________________________(8.18)
1 s
Then equation (8.17) becomes
un [( SB) 1 ( / s)] [( SB) 1 K / s)]
________________ (8.19)
un [( SB) ( / )] [( SB) ( K dt ) / ]
1 1
Stability analysis [47] shows that the error feedback gain K can be used to
guarantee the robust stability of the closed-loop system. Intuitively, if T is small
enough the system is always stable. However, this might be difficult to implement
due to larger magnitude of the control and hence the value of T has to be a
compromise. The accuracy of estimation can be improved by an appropriate
choice of Gf(s).
9. Thus the TDC and its improvement IDC can be implemented to give a far
better performance and without the inherent drawback of chattering since the
control is no longer discontinuous.
45
CHAPTER NINE
and angular rates where nonlinearities are dominant. For example, the missiles
having wingless configurations will need to execute high angle of attack
maneuvers to generate the desired acceleration levels. Under these
circumstances, the assumption of small motion about operating point do not hold
true and so one need to employ nonlinear approaches for designing of missile
autopilots. Various theories have been developed for design of controllers for
nonlinear dynamic systems and applications of the same for missile autopilot
designs. Owing to the highly nonlinear dynamics, various nonlinear approaches
such as the Feedback Linearization (FL) [54], Predictive Control [55], Sliding
Mode Control [56] have been proposed for the designing of the missile autopilot.
The issue of robustness in nonlinear controllers has also been addressed widely.
3. Apart from the highly nonlinear dynamics, the problem areas in the design
of missile autopilots broadly include the following: -
Thus sliding mode control is found to address all the above issues and can
provide a robust controller by proper design. Some methods based on research
papers are discussed below.
model developed by the authors, where the coupling term between the lateral
and longitudinal motion is also considered. Using the wind angle estimate, we
can design the observer-based controller to show the desired acceleration
tracking performance.
6. Future missile systems will be required to possess higher turn rates and
larger maneuverability envelopes, while simultaneously requiring reduced
storage and signature volumes. In this respect, efforts are under way to evaluate
alternate means of missile control as opposed to purely aerodynamic control [59],
[60]. Several technology payoffs can be envisioned if alternate control strategies
are implemented. Some of them are: -
(a) Smaller fin sizes that decrease stowage volume for internal
carriage, especially important for the type of fighters currently being
developed.
10. Variable structure control (VSC) methods are suggested for the control
design for several reasons. First of all, the maneuver characteristics present
nonlinear dynamics with large parameter variations. Second, the autopilot will
require high gain capabilities, due to the maneuver requirements in terms of time
of steady state acquisition. Third, the use of reaction jets, as the main component
of the actuation system, leads naturally to the implementation of nonlinear, relay-
type logic. VSC, with due care, is one of the methodologies capable of
addressing all these issues in a structured framework.
11. From the simulation results shown in the paper [61] the advantage of
using RCS in conjunction with fins is evident, as shown by figures below. The
turn radius for the missile autopilot using RCS in conjunction with fins is of the
order of 150 feet and the time necessary for a 180 degree maneuver is 2
seconds while the turn radius for a standard g-command autopilot using the
51
same configuration is of the order of 15 000 feet and the time necessary for the
maneuver is about 50 s. Dynamic pressure loss is recovered by the main engine
according to the attitude of the missile, or by performing the maneuver during
boost phase. Autopilot performances are validated via nonlinear simulation,
showing the robustness with respect to aerodynamic uncertainties and speed
variations. Current work [63] indicates that the autopilot gains do not require
scheduling even in the presence of varying flight conditions (initial reference
speed), reaction jet thrust, and main engine main thrust values. Future work is
directed toward the extension of the design to the three-dimensional case, to the
selection of the best VSC structure, and to the problem of output feedback, since
the autopilot presented in this work assumes full state availability.
Fig. 9.4 Simulation Results for Missile Autopilot Using RCS (Time
Response for angles)
CHAPTER TEN
CONCLUSION
REFERENCES
[2] John Y. Hung, Weibing Gao, James C. Hung, “Variable Structure Control:
A Survey” IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 40,
NO. 1, FEBRUARY 1993.
[11] Chang, T.H., and Hurmuzlu, Y., \Sliding Control Without Reaching Phase
and Its Application to Bipedal Locomotion," Journal of Dynamics Systems,
Measurement and Control, Vol.115, pp. 447-455, 1993.
[12] Choi, S. , Park, D., and Jayasuriya, S., \A Time-Varying Sliding Surface for
Fast and Robust Tracking Control of Second-Order Uncertain Systems,"
Automatica, 30, pp. 899-904, 1994.
55
[14] Roy, R.G., and Olgac, N., \Robust Nonlinear Control via Moving Sliding
Surfaces - n-th order case," Proceedings of the 36th Conference on Decision and
Control, San Diego, California USA, pp. 943-948, 1997.
[15] Sastry, S, and Bodson, M., Adaptive Control - Stability, Convergence, and
Robustness, Prentice Hall, New Jersey, 1989.
[24] Chen, W.and Saif, M., “Novel Sliding Mode Observers for a Class of
Uncertain Systems,” IEEE Transactions on Automatic Control , Vol. 51, No. 5,
2006, pp. 814–818.
56
[25] Slotine, J. J. E. and Misawa, E. A., “On Sliding Mode Observers for
Nonlinear Systems,” Journal of Dynamic Systems, Measurement and Control ,
Vol. 109, 1987,pp. 245–252.
[26] Lopez, R. and Yescas, R., “State Estimation for Nonlinear Systems under
Model Uncertainties: A Class of Sliding-mode Observers,” Journal of Process
Control , Vol. 15, 2005, pp. 363–370.
[27] Jiang, L.and Wu, Q., “Nonlinear Adaptive Control via Sliding-mode State and
Perturbation Observer,” IEE Proc.- Control Theory Applications, Vol. 149, No. 4,
2002, pp. 269–277.71
[28] Liu, C. and Peng, H., “Disturbance Observer Based Tracking Control,”
Journal of Dynamic Systems, Measurement and Control, Transactions of ASME,
Vol. 122, 2000, pp. 332–335.
[32] D. Yoo, S. S.-T. Yau, and Z. Gao, “Optimal fast tracking observer
bandwidth of the linear extended state observer,” International Journal of Control,
vol. 80, no. 1, pp. 102–111, 2007.
[34] Y. Huang, K. Xu, J. Han, and J. Lam, “Flight control design using
extended state observer and non-smooth feedback,” in Proc. of the 40th IEEE
Conference on Decision and Control, (Orlando, Florida, USA), pp. 223–228,
December 2001.
[36] R. Zhang and C. Tong, “Torsional vibration control of the main drive
system of a rolling mill based on an extended state observer and linear quadratic
control,” Journal of Vibration and Control, vol. 12, no. 3, pp. 313–327, 2006.
57
[37] Chi-Man Kwan, “Sliding Mode Control of Linear Systems with Mismatched
Uncertainties*”, Automatica, vol.31, no.2, pp. 303-307, 1995.
[38] Youcef-Toumi K., and Ito, O., 1987, “Controller Design for Systems With
Unknown Nonlinear Dynamics”’ Proc. Of ACC, Minnesota, pp. 836-844.
[39] Youcef-Toumi K., and Iti, O., 1990, “A Time Delay Controller for Systems
with Unknown Dynamics,” ASME J. Dyn. Syst., Meas., Control, 112(1), pp. 133-
142.
[46] Zhong, Q,-C., and Rees, D., 2004, “Control of Uncertain LTI Systems
Based on an Uncertainty and Disturbance Estimator”, ASME, J. Dyn. Syst. Meas.,
Control, 126(4), pp. 905-910.
[50] P. Zarchan, Tactical and Strategic Missile Guidance, vol. 199 of Progress
in Astronautics and Aeronautics. Reston, VA: American Institute of Aeronautics
and Astronautics, Inc., 4th ed., 2002.
[56] D.-G. Choe and J.-H. Kim, “Pitch autopilot design using model following
adaptive sliding mode control,” Journal of Guidance, Control, and Dynamics, vol.
25, no. 4, pp. 826–829, 2004.
[58] Dongkyoung Chwa and Jin Young Choi, “Observer-Based Control for Tail-
Controlled Skid-to-Turn Missiles Using a Parametric Affine Model”, IEEE
TRANSACTIONS ON CONTROL SYSTEMS TECHNOLOGY, VOL. 12, NO. 1,
JANUARY 2004 pp 167-175.
[61] Ajay Thukral and Mario Innocenti, “A Sliding Mode Missile Pitch Autopilot
Synthesis for High Angle of Attack Maneuvering”, IEEE TRANSACTIONS ON
CONTROL SYSTEMS TECHNOLOGY, VOL. 6, NO. 3, MAY 1998 pp. 359-371
[63] M. Innocenti and A. Thukral, “Variable structure autopilot for high angle of
attack maneuvers, using on–off thrusters,” in 33rd Contr. Decision Conf., Lake
Buena Vista, FL, Dec. 1994.