Professional Documents
Culture Documents
Electrical Engineering
a r t i c l e i n f o a b s t r a c t
Article history: This paper addresses the robust control problem of Twin Rotor Multi input multi output System (TRMS)
Received 30 May 2016 via H2 and H1 control techniques. An output error optimization technique is proposed to develop H2 and
Accepted 11 July 2019 H1 controllers for the well posed plant. Computer simulation results are presented for closed loop TRMS
Available online xxxx
in hovering positions which show marked improvements over previous works. The simulated plant exhi-
bits stable responses in hovering position at the desired pitch and yaw angles. Corrections are incorpo-
Keywords: rated in model formulation to compensate control and sensor singularities. The output error
Modeling
optimization technique proposed in the present paper can be essentially adopted in controlling of 2 by
Singularity
Output error optimization
2 plants exhibiting non-minimum phase dynamics.
Well-posedness Ó 2019 THE AUTHORS. Published by Elsevier BV on behalf of Faculty of Engineering, Ain Shams Uni-
TRMS versity. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/
by-nc-nd/4.0/).
https://doi.org/10.1016/j.asej.2019.07.001
2090-4479/Ó 2019 THE AUTHORS. Published by Elsevier BV on behalf of Faculty of Engineering, Ain Shams University.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
Please cite this article as: P. K. Paul and J. Jacob, H2 Vs H1 control of TRMS via output error optimization augmenting sensor and control singularities2 Vs
H1 control of TRMS via output error –>, Ain Shams Engineering Journal, https://doi.org/10.1016/j.asej.2019.07.001
2 P.K. Paul, J. Jacob / Ain Shams Engineering Journal xxx (xxxx) xxx
[3]. With these assumptions, the linearized system takes the form
of a classical master-slave cascade linear system, where the simpli-
fied linear system can be controlled with well-known linear strate-
gies for instance, H1, LQR, optimal, or switched control [2,6].
Contemporary to these results, several advances in the theory of
nonlinear control have been reported. One such control strategy
is the nonlinear extension of classical robust L2 control [3]. Lu
and Wen [7] have decoupled TRMS as two SISO plants and it is con-
trolled through one degree of freedom controller on each of the
SISO system.
Several robust control optimization techniques are discussed by
Skogestad and Postlethwaite in [8] from which the present work is
partially motivated. In this work, an output error optimization
technique is proposed by using generalized plant encapsulating
TRMS and other subsystems except robust controller. The H2 and
H1 optimization algorithm and the generalized plant are used to
design the robust controllers. Prior to this, the nonlinear dynamic
Fig. 1. Sketch of TRMS laboratory unit [1]. mathematical model of TRMS is linearized to obtain the nominal
model. The generalized plant is formulated by using the nominal
model of TRMS.
Determination of the nominal model of a physical plant is the
first and inevitable step in solving any control system problem in
general. In this case, the required nominal model is linear. This
can be determined by linearizing the mathematical relationship
of the system and obtaining the state space model. Several efforts
have been made in the past to derive a representative linear model
of TRMS that behaves in the same manner as the physical TRMS in
hover (translatory motion w.r.t. ground being zero). In a recent
work, Nejjari et al. [9] have presented system identification and
modeling of TRMS leading to Quasi-LPV modeling. In a former
work, the forward path Transfer Functions (T.F.s) were derived
by Lu and Wen [7] for the SISO sub-systems of the decoupled
TRMS. A dynamic model characterizing the TRMS in hovering posi-
tion was extracted using a black box system identification tech-
nique in the works of Ahmed et al. [10–13]. Identification of
system parameters of TRMS based on least squares parametric
Fig. 2. Working model of TRMS in hover. (Curtsey: Control Systems Laboratory, identification method is presented by Belkheiri et al. [14]. Tanaka
Department of Electrical Engineering, National Institute of Technology Calicut, et al. [15] considered linear parameter varying identification of
India.)
TRMS via grey box modeling. Rahideha et al. [16] and Ekbote
et al. [17] had used Lagrangian based formulation of equation of
the pitch and yaw angles are confined to slow dynamics. In addi- motion to derive a dynamic model of TRMS. In a recent attempt,
tion, the linear models obtained by traditional linearization meth- Vishnupriyan et al. [18] have presented uncertainty modeling
ods are nominal models and devoid of critical system information. and control using linearized plant model of TRMS. In another
It is also known that a robust controller is able to stabilize the sys- recent work, a model based on first-principle modeling was
tem when there is marked variation between the actual plant and derived and its parameters were refined by identification based
the linear model used to determine the robust controller. Thus, on real-time experiments as presented by Chapula et al. [19]. In
TRMS is a suitable platform for testing robust control strategies. the present work a nominal model of TRMS is obtained by lineariz-
The techniques and algorithms for the design of robust con- ing it using Taylor series expansion followed by defining Jacobians.
troller for TRMS have been improved gradually as reported in var- A 2 2 T.F. model is obtained from the state space representation.
ious national and international conferences and journals. In the But this T.F. model is devoid of the RHP zero dynamics of the plant.
present work, this class of systems have been extensively surveyed, The missing dynamics in the T.F. matrix are taken care by robust
identifying a number of control strategies that range from linear control technique developed in this work.
robust control techniques as stated by Smerlas et al. in [4] to more Subsequently, a generalized output error optimization tech-
recent nonlinear approaches as mentioned by Vilchis et al. in [5]. It nique is proposed in this paper. Two different robust controllers
can be proved that the dynamics of TRMS are non-minimum phase, are designed using this technique, viz. H2 and H1 controllers. A
exhibiting unstable zero dynamics, which make the system not feed forward gravity compensation block is used to compensate
suitable for classical feedback linearization techniques [2]. This acceleration due to gravity in the vertical plane. The control sys-
fact, jointly with important modeling uncertainties, especially in tems are simulated with unity feedback control of TRMS. Computer
the high frequency range, makes the system hard to control by simulation of TRMS demonstrates improved results over the past
standard techniques as mentioned by Mullhaupt et al. in [6]. contributions. The algorithm is superior over other robust control
Other classical approaches to sort out the difficulty associated techniques because the singularities are taken care by the design
with modeling uncertainties in the high frequency range, make itself. Following section gives the mathematical description of
use of simplifying assumptions to decouple rigid body and rotor TRMS. Table 1 defines the abbreviations and symbolic presenta-
dynamics, such that partial feedback linearization can be applied tions. All variables are measured in SI units.
Please cite this article as: P. K. Paul and J. Jacob, H2 Vs H1 control of TRMS via output error optimization augmenting sensor and control singularities2 Vs
H1 control of TRMS via output error –>, Ain Shams Engineering Journal, https://doi.org/10.1016/j.asej.2019.07.001
P.K. Paul, J. Jacob / Ain Shams Engineering Journal xxx (xxxx) xxx 3
Table 1 I2 u
€ ¼ M 2 M Bu M R ð7Þ
Nomenclature.
M G ¼ K gy M1 u
_ cos ðwÞ ð5Þ
T 11 s_ M ¼ T 10 sM þ k1 uM ð6Þ
Please cite this article as: P. K. Paul and J. Jacob, H2 Vs H1 control of TRMS via output error optimization augmenting sensor and control singularities2 Vs
H1 control of TRMS via output error –>, Ain Shams Engineering Journal, https://doi.org/10.1016/j.asej.2019.07.001
4 P.K. Paul, J. Jacob / Ain Shams Engineering Journal xxx (xxxx) xxx
Table 2
Parameters and values [1].
Parameter Value
I1 – moment of inertia of vertical rotor 6.8 102 kg m2
I2 – moment of inertia of horizontal rotor 2 102 kg m2
a1 – static characteristic parameter 0.0135
b1 - static characteristic parameter 0.0924
a2 – static characteristic parameter 0.01
Fig. 4. General control configuration.
b2 – static characteristic parameter 0.09
MG – gravity momentum 0.32 N-m
B1w – friction momentum function parameter 6 103N-m-s/rad.
B2w – friction momentum function parameter 1 103N-m-s/rad. This configuration is confined to the linear time invariant MIMO
B1u – friction momentum function parameter 1 101N-m-s/rad.
finite order systems tested for its well-posedness with finite H2/H1
B2u – friction momentum function parameter 1 103N-m-s/rad.
Kgy – gyroscopic momentum parameter 0.05 s/rad norms. Commonly available optimization methods for RHP zero
k1 – motor 1 gain 1.1 plants frequently fail prematurely solely because of the divide by
k2 – motor 2 gain 0.8 zero encounters while iteration, if not taken care at the design
T11 – motor 1 denominator parameter 1.2 stage. Manipulating programming code or resetting word length
T10 – motor 1 denominator parameter 1
of variable types in simulation software is rather an uneducated
T21 – motor 2 denominator parameter 1
T20 – motor 2 denominator parameter 1 approach in solving such hurdles. A sophisticated output error
Tp – cross reaction momentum parameter 2 optimization approach is proposed in this paper to overcome such
T0 – cross reaction momentum parameter 3.5 hurdles.
kc – cross reaction momentum gain 0.2
State space representation of P(s) is considered as
_
XðsÞ ¼ AXðsÞ þ B1 WðsÞ þ B2 UðsÞ ð23Þ
2 3
0 0 1 0 0 0 0
6 0 0 0 1 0 0 0 7 ZðsÞ ¼ C 1 XðsÞ þ D11 WðsÞ þ D12 UðsÞ ð24Þ
6 7
6 7
6 4:348 0 0:088 0 1:246 0 0 7
6 7 YðsÞ ¼ C 2 XðsÞ þ D21 WðsÞ þ D22 UðsÞ ð25Þ
A¼6
6 0 0 0 5 1:482 3:6 18:75 7
7
6 7 Thus, the generalized plant acquires the form in state space as:
6 0 0 0 0 0:833 0 0 7
6 7 2 3
4 0 0 0 0 0 1 0 5 A ½ B1 B2
6 D 7
0 0 0 0 0:017 0 0:5 PðsÞ ¼ 4 C 1 11 D12 5 ð26Þ
ð18Þ C2 D21 D22
T with inputs W(s) and U(s), outputs Z(s) and Y(s). The robust con-
0 0 0 0 1 0 0 troller K(s), to be designed by H2/H1 optimization algorithm, is pre-
B¼ ð19Þ
0 0 0 0 0 1 0 sented in state space as:
Please cite this article as: P. K. Paul and J. Jacob, H2 Vs H1 control of TRMS via output error optimization augmenting sensor and control singularities2 Vs
H1 control of TRMS via output error –>, Ain Shams Engineering Journal, https://doi.org/10.1016/j.asej.2019.07.001
P.K. Paul, J. Jacob / Ain Shams Engineering Journal xxx (xxxx) xxx 5
Otherwise, if iteration fails, then early independent and the rows of the matrix C2(sI-A)1B1 + D21
are linearly independent.
cnþ1 ¼ 0:5 cnhigh þ cn ð31Þ
4.2.3. Condition 3
In case of H2 optimization, the condition, D11 ¼ 0 and D22 ¼ 0
are treated separately.
cnþ1
low ¼ c
n
ð32Þ
The iteration repeats until we get, 4.2.3.1. Condition D11 ¼ 0. The software used in many cases do not
support optimization with this condition. So, with such occurrence
cnhigh cnlow < tolerance ð33Þ the existing algorithms may be modulated to acquire D11 –0.
Stable H2/H1 controllers can be designed by directly computing 4.2.3.2. Condition D22 ¼ 0. This case resembles to direct transfer of
the norm constrained stable transfer matrices Q in the H2/H1 sub- input to output. It is compensated manually.
optimal controller parameterization. First the problem is trans-
formed into nonlinear unconstrained optimization problem.
5. Design of H2/H‘ controllers by output error optimization
Then, a two stage numerical search is carried out. Consider Gs(s)
as a stable transfer matrix with state space realization,
The desired performance of TRMS is specified in terms of a ref-
erence model G0(s) having the form as:
A B
Gs ðsÞ ¼ ð34Þ g 01 ðsÞ 0
C 0 G 0 ð sÞ ¼ ð40Þ
0 g 02 ðsÞ
Then, g 01 ðsÞ and g 02 ðsÞ are chosen as standard second order systems satis-
fying the implied stability conditions.
jjGs jj22 ¼ TraceðB QBÞ ¼ TraceðCRC Þ ð35Þ
In the present technique, outputs of the generalized plant P(s)
Then, R and Q are the controllability and observability grami- are penalized by comparing outputs of TRMS with that of G0(s).
ans, which can be obtained by solving the Lyapunov equations, More specifically, output errors are penalized in this case. The ref-
erence plant, G0(s) generates desired outputs. Any deviation of
AR þ RA þ BB ¼ 0 ð36Þ actual outputs (pitch and yaw displacements of TRMS) is consid-
ered as controlled error outputs of the plant P(s) to be tackled by
A Q þ QA þ C C ¼ 0 ð37Þ H2/H1 control algorithms. Further, solving H2 and H1 problems
Consider the generalized plant P(s) is Eq. (26). Rewriting the using mathematical tools are often limited by sensor and control
same for the purpose of lower Linear Fractional Transformation singularities at zero and infinite frequencies, respectively, in addi-
(LFT), tion to the deficit observable and/or controllable variables. To over-
2 3 come this hurdle, control singularity at infinite frequency is
A ½ B1 B2 compensated by adding an additional plant output as euV(s). Sim-
6C D 7 P 11 P12
PðsÞ ¼ 4 1 11 D12 5 ¼ ð38Þ ilarly, the sensor singularity at steady state is compensated by add-
P 21 P22 ing an additional disturbance term eyW(s). Unlike previous efforts
C2 D21 D22
to augment physical input/output variables to overcome singular-
Recalling Fig. 4, ity as well as controllability and/or observability issues, mathemat-
ical augmentation of such variables in generalized plant
T zw ¼ F l ðP; K Þ ¼ P11 þ P21 KðI P22 KÞ1 P 21 ð39Þ
formulation is the essence of the approach proposed in this paper.
The lower LFT presented in Eq. (39) is optimized to iterate opti- Thus the generalized plant will have four controlled outputs, two
mal H2 controller and suboptimal H1 controllers. sensor outputs, two reference inputs and two disturbance inputs.
Fig. 5 is a line diagram presenting interconnection of subsystems.
4.2. Conditions for well-posedness TRMS being a 2 2 plant, Fig. 5 can be illustrated more elabo-
rately from physical perspective as shown in Fig. 6:
A well-posed feedback loop establishes all transfer functions in Input vector; ½ W
T
U ¼ ½ r1 r2 w1 w2 v 1 v 2 T ð41Þ
the feedback loop as well defined and proper, failing which the
optimization algorithm is prone to collapse [22–25]. The three con-
Output vector; ½ Z Y T ¼ ½ Z 1 Z2 y1 y 2 T
ditions to be tested to assure well-posedness of the system are
considered in the rest of this section. Testing for well-posedness ¼ ½ q1 z1 q2 z2 y1 y 2 T ð42Þ
is equivalent to validating the controllability and observability of
the generalized plant formulated in the context. State vector; X ¼ ½ x1 x2 x3 x4 x5 x6 x7 x8 x9 x10 x11 T
ð43Þ
4.2.1. Condition 1
This constraint pertains to the controllability and observability
tests for the pairs (A, B2) and (C2, A), respectively. This is to ensure a
stable feedback control system. Failing these tests, the system
needs to be amended with additional control inputs and measure-
ments for stability.
4.2.2. Condition 2
The second condition ensures that any exogenous (disturbance)
input is measured at output and considered for generating control
signal. Thus, the columns of the matrix C1(sI-A)1B2 + D12 are lin- Fig. 5. Control scheme of TRMS as output error optimization problem.
Please cite this article as: P. K. Paul and J. Jacob, H2 Vs H1 control of TRMS via output error optimization augmenting sensor and control singularities2 Vs
H1 control of TRMS via output error –>, Ain Shams Engineering Journal, https://doi.org/10.1016/j.asej.2019.07.001
6 P.K. Paul, J. Jacob / Ain Shams Engineering Journal xxx (xxxx) xxx
T
0 0 e1u 0
D12 ¼ ð53Þ
0 0 0 e2u
1 0 0 0
D21 ¼ ð54Þ
0 1 0 0
0 0
D22 ¼ ð55Þ
0 0
Thus,
2 3
g 01 0 g 11 e1y 0 g 11 0
6 0 g 0 g 22 e2y 0 g 22 7
6 02 7
6 7
6 0 0 0 0 e1u 0 7
Fig. 6. Elaborate control structure for robust output error optimization technique P ð sÞ ¼ 6
6 0 0
7 ð56Þ
with TRMS. 6 0 0 0 e2u 77
6 7
4 1 0 g 11 e1y 0 g 11 0 5
T 0 1 0 g 22 e2y 0 g 22
i:e:; X ¼ w u w_ u
_ sM sT MR y10 y_ 10 y20 y_ 20
ð44Þ The generalized model of P(s) as obtained in Eqs. (47)–(56) is
used to iterate c using H1 and H2 optimization algorithms. An
g 01 ðsÞ ¼ xn1 2 = s2 þ 2n1 xn1 s þ xn1 2 ð45Þ optimal H2 controller and a suboptimal H1 controller are thus
obtained by iteration. The controllers are tested by simulating
TRMS with main rotor feed-forward gravity compensation function
g 02 ðsÞ ¼ xn2 2 = s2 þ 2n2 xn2 s þ xn2 2 ð46Þ
as 0.3(w + 1). Simulation results are presented and discussed in
The matrices in state space representation can be expressed as: the following section.
2 3
0 0 1 0 0 0 0 0 0 0 0
6 0 0 0 1 0 0 0 0 0 0 0 7
6 7
6 7
6 4:348 0 0 0 1:246 0 0 0 0 0 0 7
6 7
6 0 0 0 5 1:482 3:6 18:75 0 0 0 0 7
6 7
6 7
6 0 0 0 0 0:833 0 0 0 0 0 0 7
6 7
A¼6 0 0 0 0 0 1 0 0 0 0 0 7 ð47Þ
6 7
6 0 0 0 0 0:017 0 0:5 0 0 0 0 7
6 7
6 0 0 0 0 0 0 0 0 1 0 0 7
6 7
6 7
6 0 0 0 0 0 0 0 xn1 2 2n1 xn1 0 0 7
6 7
4 0 0 0 0 0 0 0 0 0 0 0 5
0 0 0 0 0 0 0 0 0 xn2 2 2n2 xn2
2 3T
0 0 0 0 0 0 0 0 g a1 xn1 2 0 0 6. Results and discussion
6 0 0 0 0 0 0 0 0 0 0 g a2 xn2 2 7
6 7 ð48Þ
B1 ¼ 6 7 All the conditions given in Section 4.2 are tested for well-
4 0 0 0 0 e1y 0 0 0 0 0 0 5
posedness of the generalized plant. Both the H2 and H1 controllers
0 0 0 0 0 e2y 0 0 0 0 0 designed are of type zero and 11th order. A decoupler is suitable
T chosen to minimize the cross path interactions between yaw and
0 0 0 0 1 0 0 0 0 0 0 pitch. Simulation results of pitch and yaw displacements are illus-
B2 ¼ ð49Þ
0 0 0 0 0 1 0 0 0 0 0 trated and discussed in rest of this section. Rotor inputs in simula-
tion interface of MATLAB are limited to ±2.5 V.
2 3
1 0 0 0 0 0 0 1 0 0 0
6 0 1 0 0 0 0 0 0 0 1 0 7 6.1. H2 control simulation results
6 7
C1 ¼ 6 7 ð50Þ
4 0 0 0 0 0 0 0 0 0 0 0 5
Numerical values assigned to different parameters and con-
0 0 0 0 0 0 0 0 0 0 0
stants for H2 optimization are as follows:
1 0 0 0 0 0 0 0 0 0 0 xn1 ¼ 0:5; xn2 ¼ 0:5; g a1 ¼ 1:02; g a2 ¼ 1:02; e1y ¼ 0:01;
C2 ¼ ð51Þ
0 1 0 0 0 0 0 0 0 0 0 e2y ¼ 0:01; e1u ¼ 0:01; e2y ¼ 0:01; n1 ¼ 0:9; n2 ¼ 0:9;
2 3 Nominal model of linearized TRMS obtained in Eq. (22) are
0 0
6 . . .
. . .. 7
quoted below for reference (see Figs. 7a and 7b, Figs. 8a and 8b).
D11 ¼ 4 .. 5 ð52Þ
0 0 g 11 ðsÞ ¼ 1:359= s3 þ 0:9091s2 þ 4:706s þ 4:278 ;g 22 ðsÞ ¼ 3:6= s3 þ s2
44
Please cite this article as: P. K. Paul and J. Jacob, H2 Vs H1 control of TRMS via output error optimization augmenting sensor and control singularities2 Vs
H1 control of TRMS via output error –>, Ain Shams Engineering Journal, https://doi.org/10.1016/j.asej.2019.07.001
P.K. Paul, J. Jacob / Ain Shams Engineering Journal xxx (xxxx) xxx 7
g 11 ðsÞ and g 22 ðsÞ are taken as given in Section 6.1 (see Figs. 9a and
9b, Figs. 10a and 10b).
Yaw control is achieved within specifications with H1 con-
troller as illustrated in the following Figs. 11a and 11b, Figs. 12a
and 12b.
Fig. 7a. Pitch output against set point of 28° (without gravity compensation) – H2
control.
Fig. 9a. Pitch output against set point of 28° (without gravity compensation) – H1
control.
Fig. 7b. Main rotor input for pitch set point of 28° (without gravity compensation)
H2 control.
Fig. 9b. Main rotor input for pitch set point of 28° (without gravity compensation)
H1 control.
Fig. 8a. Pitch output against set point of 28° (with gravity compensation) – H2
control.
Fig. 8b. Main rotor input for pitch set point of 28° (with gravity compensation) H2 Fig. 10a. Pitch output against set point of 28° (with gravity compensation) – H1
control. control.
Please cite this article as: P. K. Paul and J. Jacob, H2 Vs H1 control of TRMS via output error optimization augmenting sensor and control singularities2 Vs
H1 control of TRMS via output error –>, Ain Shams Engineering Journal, https://doi.org/10.1016/j.asej.2019.07.001
8 P.K. Paul, J. Jacob / Ain Shams Engineering Journal xxx (xxxx) xxx
Fig. 10b. Main rotor input for pitch set point of 28° (with gravity compensation) -
H1 control. Fig. 12a. Yaw output against set point of 60° – H1 control.
Table 3
Relative and absolute stavility parameters – H2 Vs H1 pitch control.
Table 3 contains the list of relative and absolute stability the sensitivity function shaping methods in which the controlled
parameters for (1) H2 controller and (2) H1 controller: outputs are penalized by using a weighting function which is again
Table 3 summarizes a range of stability parameters of TRMS optimized by trial and error method. Hence, output error optimiza-
with H2 and H1 controllers for pitch. It can be concluded that tion technique is a straight forward method and minimizes con-
H1 controller exhibits superior results in aggregate. However, per- troller design time and other requirements.
formance of H1 controller without gravity term is not upto the
mark while it performs well in the presence of gravity compensa- 7. Conclusion
tion. In addition, H2 optimization did not yield satisfactory results
for yaw displacement whereas with H1 controller desirable yaw The present work establishes the superiority of (1) H1 con-
movements are observed. troller over H2 controller when optimized via the proposed output
Further, the approach proposed in this work imbibes the rela- error optimization technique tested on TRMS and (2) output error
tive and absolute stability requirements in the form of second optimization technique over sensitivity function shaping method
order transfer functions (G0(s)). Thus, this method is superior to using weighting functions. Thus, output error optimization
Please cite this article as: P. K. Paul and J. Jacob, H2 Vs H1 control of TRMS via output error optimization augmenting sensor and control singularities2 Vs
H1 control of TRMS via output error –>, Ain Shams Engineering Journal, https://doi.org/10.1016/j.asej.2019.07.001
P.K. Paul, J. Jacob / Ain Shams Engineering Journal xxx (xxxx) xxx 9
technique proposed in this paper is time efficient and guaranteed system. In: Proceedings of Instn. Mech. Engrs, Journal of Systems and
Control Engineering, 218 Part I; 2004. p. 451–463.
approach. As order of the H1 controller K(s) and that of the plant
[14] Belkheiri M, Rabhi A, Boudjema F, Elhajjaji A, Bosche J. Model parameter
P(s) are the same, the controller is strictly proper (numerator – Identification and nonlinear control of a twin rotor mimo system – trms. In:
10th order and denominator – 11th order) guaranteeing strong sta- Proceedings of the 15th IFAC Symposium on System Identification, Saint-Malo,
bility of control loop. The work contributes to the class of 2 by 2 France; 2009. p. 1487–1492.
[15] Tanaka H, Ohta Y, Okimura Y. A local approach to lpv-identification of a twin
plants with non-minimum phase dynamics like TRMS that cannot rotor mimo system. In: Proceedings of the 18th World Congress - The
be tackled using traditional approaches. This approach clearly International Federation of Automatic Control, Milano (Italy); 2011. p. 7749–
explains the compensation of singularities at zero and infinite fre- 7754.
[16] Rahideha A, Shaheed MH, Huijberts HJC. Dynamic modelling of a TRMS using
quencies which are common hurdles in computer simulations. It analytical and empirical approaches. Control Eng Pract 2008;16:241–59.
further guarantees controllability and observability via well- [17] Ekbote AK, Srinivasan NS, Mahindrakar AD. Terminal sliding mode control of a
posedness tests. Simulation results have illustrated the robust twin rotor multiple-input multiple-output system. In: Proceedings of the 18th
World Congress - The International Federation of Automatic Control, Milano
behavior of the controlled system as it is clearly shown that the (Italy); 2011. p. 10952–10957.
controllers are designed using nominal model of TRMS, reference [18] Vishnupriyan J, Manoharan PS, Ramalakshmi APS. Uncertainty modeling of
model and generalized plant model. Based on the simulation nonlinear 2-dof helicopter model. In: International Conference on Computer
Communication and Informatics (ICCCI -2014), Coimbatore, India; 2014.
results, H1 controller is established as the superior one in this case. [19] Chalupa P, Přikryl J, Novák J. Modelling of twin rotor MIMO system. Procedia
The present method can be extended to control the class of sys- Eng 2015;100:249–58.
tems where number of measurements is deficit for full control of [20] Megretski A. On the order of optimal controllers in the mixed H-2/H-infinity
control. In: Proceedings of the 33rd conference on Design and Control, Lake
the plant. Results of the real time implementation of this controller
Buena Vista (Florida); 1994. p. 3173–3174.
will be reported in an ensuing paper. [21] Delgado DUC, Zhou K. A parametric optimization approach to H1 and H2
strong stabilization. Automatica 2003;39:1205–11.
Acknowledgements [22] Zhou K, Doyle JC, Glover K. Robust and optimal control. New Jersey: Prentice
Hall; 1996.
[23] Iwasaki T, Hara S. Well-posedness of feedback systems: insights into exact
This research work was performed at the Control Systems lab- robustness analysis and approximate computations. IEEE Trans Autom Control
oratory of the Electrical Engineering Department, National Insti- 1998;43(5):619–30.
[24] Francis BA. A course in H-infinity control theory, lecture notes in control and
tute of Technology Calicut, India. The authors are thankful to the information science. Berlin: Springer-Verlag; 1987.
technical staff of the Institute for providing the experimental facil- [25] Doyle JC, Francis BA, Tannenbaum AR. Feedback control theory. New
ities and arrangement. The authors appreciate the valuable com- York: Macmillan; 1992.
ments from the reviewers.
Please cite this article as: P. K. Paul and J. Jacob, H2 Vs H1 control of TRMS via output error optimization augmenting sensor and control singularities2 Vs
H1 control of TRMS via output error –>, Ain Shams Engineering Journal, https://doi.org/10.1016/j.asej.2019.07.001