See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/329314212

MRAC BasedApproachforStabilization of Single Link InvertedPendulum

Conference Paper · March 2018

DOI: 10.1109/ICCTCT.2018.8550846

CITATIONS READS

5 949

3 authors, including:

Mr. R Banerjee Ujjwal Mondal

University of Calcutta University of Calcutta
48 PUBLICATIONS 347 CITATIONS 41 PUBLICATIONS 146 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Mr. R Banerjee on 12 April 2019.

The user has requested enhancement of the downloaded file.

 Proceeding of 2018 IEEE International Conference on Current Trends toward Converging Technologies, Coimbatore, India

MRAC BasedApproachforStabilization of Single

Link InvertedPendulum
RamashisBanerjee, Naiwrita Dey UjjwalMondal
Department of AppliedPhysics, Department of AEIE, RCCIIT Department of AppliedPhysics,
CalcuttaUniversity [email protected] CalcuttaUniversity
[email protected] [email protected]

Abstract—The problem of balancing an inverted controller for second order system with MRAC using
pendulum has been a benchmark problem in MIT rule [5]. MRAC finds wide applications for
demonstrating various control design techniques. The nonlinear systems. Nonlinear actuator adds variation in
principal reasons for it’s popularity are its nonlinear process dynamics. MRAC can work for this system[6].
and unstable characteristics. In this paper a single link Miaoya Yu and Shigenox Okubo have shown the
inverted pendulum system has been chosen and observability, controllability and stability analysis of a
adaptive control algorithm has been implemented for double inverted pendulum model . Inverted pendulums
stabilization of the system so that pendulum is always are often used in nonlinear control education and research
maintained erected in its upright (inverted) position. due to challenging features like unstable and nonlinear
Model Reference Adaptive Control (MRAC) method is dynamics as well as non minimum–phase and non
implemented using MIT rule. Minimization of cost holonomic behavior. MD Shehu presents the dynamic
function is performed and system response has been behavior of a nonlinear single link inverted pendulum on
observed for different adaption gain. After system cart system based on Lagrange equation[7]. Claudio
modeling the whole process is done in MATLAB Malchiorri describes the dynamic model of two degree of
Simulink. freedom manipulator and dynamic equations are obtained
[8].This paper is organized as follows: The system
Keywords—Inverted Pendulum, Adaptive Control, dynamics is described very briefly and the transfer
MRAC, MATLAB, Stabilization function of it is developed. In Section III, a model
reference adaptive control scheme design is shown using
I. INTRODUCTION MIT rule . The proposed method of MRAC adaptive
controller is simulated for validation in Section IV, where
Stabilization of an inverted pendulum is considered as an the dynamic performance of the plant is shown followed
benchmark problem in the field of control system. Many by some concluding remarks in Section V.
research work has been carried out so far on various
effective control methods for inverted pendulum system. II. SYSTEM MODELLING
In the field of advanced control system adaptive control
has gained wide popularity among different control The physical system considered in this paper is an
strategies for better performance and accuracy [1].Model inverted pendulum which has been shown in Fig.1
Reference Adaptive Control (MRAC) which is a direct Inverted pendulum systems represent a significant class of
adaptive control strategy involves an adjustment nonlinear under actuated mechanical systems [9-11] .The
mechanism to change the controller parameters. In system is oriented is this way that a bar is placed
comparison with the conventional controllers, adaptive vertically . Its pivot point is attached to an angle sensor,
controllers perform well for online parameter updation encoder for the measurement of the angle and angular
for variation in process dynamics subjected to velocity .
environmental changes and varying characteristics of
disturbances. R.J. Pawar and B.J.Parvat implemented a
combination of MRAC and PID control scheme to an
inverted pendulum to extend the application of
conventional MRAC[2]. K.Suresh in his paper compared
the performance analysis of proposed fuzzy MRAC based
control for drying process is compared [3]. Suresh B.
Pingale compared the application of conventional PID
controller with MRAC to an inverted pendulum system.
He further implemented fuzzy MRAC which further gave
better simulation results [4]. To overcome these problems Fig.1. Single inverted pendulum system
Priyank Jain and M.J. Nigam deals with designing of

978-1-5386-3702-9/18/$31.00 © 2017 IEEE 1

 Proceeding of 2018 IEEE International Conference on Current Trends toward Converging Technologies, Coimbatore, India

A DC Motor and a propeller were attached to the another MIT Rule:

end of the bar. As voltage is applied to the motor, the MIT rule works on the basis of gradient theory in which a
propeller starts to spin and the bar is pulled up. The cost function is considered which is a function of error.
control objective is to stabilize the position of the bar to a The difference between actual plant response and
specified angle . reference plant response is termed as error. In the negative
direction of the gradient of the considered cost function
The second order single inverted pendulum system can ,the controller parameters will be altered. This cost
be described by: function is defined in terms of the error between the actual
behaviour and ideal behaviour of the plant [12].
d2y dy
 a  by  bu The control parameter  is adapted with following cost
dt 2
dt (1)
function given by J
1
Let a=8 and b=600 Where y is the plant response. Plant Cost Function: J ( )  e 2 (3)
is a second order under damped system and u is the 2
controller output or manipulated variable can be seen in
Fig 2. Plant transfer function is given by: The adjustable parameter is to be adjusted to be
600 minimized to zero. This is the reason behind change in 
y 2 (2)
S  8S  600 is proportional to negative gradient of J .

III.MODEL REFERENCE ADAPTIVE CONTROL d J e

Update Rule:    e (4)
Model Reference Adaptive Controller is a direct adaptive
dt  
control approach .All the desired specifications are given
in the form of reference model . Design of the adaptive Relation between change in theta and cost function is
controller using MRAC is to be done in such a way that a known as the MIT rule. The partial derivative term in
reference model is chosen to generate a desired trajectory equation no (4) is known as sensitivity derivative of the
and an adjustment mechanism for online controller system. γ is a positive quantity which indicates the
parameter estimation so that the plant will follow the adaptation gain of the controller.
desired trajectory [12-13]. It consists of two loops inner
loop and outer loop. The reference model is in outer loop. Implementation of MRAC for Adaptive controller design:
Ordinary feedback is called inner loop containing the Transfer function of the actual plant is obtained from
plant and the controller. Parameters are adjusted using equation no 1.Reference model is given in terms of
the update law in adjustment mechanism depending upon transfer function obtained from the desired performance
the error which is the difference between produced output specifications (i.e. rise time, settling time, overshoot and
& reference model value . steady state error) .For this paper the reference model of
single inverted pendulum is designed such that settling
time 𝑇𝑠= .942 𝑠𝑒𝑐. and damping ratio ζ = .707.The
transfer function of the reference model is given below

36
ym  (5)
S  8.4S  36
2

The objective is to compare between the actual plant

response (y) and the idle (reference) response ( y m ) and,
by using MRAC method, the error needs to be minimized
to achieve the desired system rsponse . The controller
output is described by the following control law:

u(t )  1r (t )   2 y(t ) (6)

Fig.2 Basic Block diagram of MRAC system

2
 Proceeding of 2018 IEEE International Conference on Current Trends toward Converging Technologies, Coimbatore, India

Fig.4.Open Loop Response of the actual and reference

system

Fig.3 Mathematical model of proposed controller using

MRAC for single inverted pendulum

The adjustable controller parameters are chosen as:

bm
1 
b (7)

am  a
2  (8)
Fig. 5 .Closed Loop Response of the actual systemwithout
b controller

Controller parameters update rule using MIT rule is given

by the following two equations:

d1 e br a r
 e  e[ ]  e[ m ]
dt 1 p  am p  am (9)

d 2 a y
 e[ m ] (10)
dt ( p  am )

Where α is adaption gain and e is tracking error. Fig.6 .Closed Loop Response of the system
using MRAC based Adaptive controller
 *b
 (11)
am The time response characteristics for the plant and the
reference model, without using any control technique, are
shown in Fig.5. It can be observed that the characteristic
e  y  ym
(12) of the plant is oscillatory with high value of overshoot and
undershoot whereas the characteristic of the reference
IV. PERFORMANCE EVALUATION model is smooth without any oscillation. There is a large
dynamic error in system is reduced using control scheme
In this paper the model reference adaptive control model is is shown in Fig.6. It shows the output characteristic
simulated in MATLAB Simulink. The Fig.4 below shows (response curve) for the system with proposed control
the open loop response of single link inverted pendulum scheme. It can be seen that the characteristic of the plant
system without any controller. It can be clearly seen that has improved drastically with removal of oscillation in the
there are huge oscillations in the plant response whereas system response. Different value of adaptation gain is
the reference model response is stable and no oscillations chosen and system response is observed in Fig 7.a, Fig 7.b
are there in case of reference model. and Fig 7.c. With higher value of adaptation gain the
proposed controller results in better system response with

3
 Proceeding of 2018 IEEE International Conference on Current Trends toward Converging Technologies, Coimbatore, India

lesser oscillation. future this work can be extended for optimization of

adaption gain and also experiment can be carried out.
MRAC controller can be designed using Liapunov's
stability method.

REFERENCES

[1]. G. Feng, R. Lozano, " Adaptive Control Systems", Read

Educational and Professional Publishing, 1999.
[2]. R.J.Pawar, B.J.Parvat, “Design and Implementation of
MRAC and Modified MRAC Technique For inverted
Pendulum”,International Conference on Pervasive
Computing (ICPC),Pune, India,8-10 January,2015
[3]. K.Suresh, “Performance Analysis on MRAC and Fuzzy
Fig.7.a .System response for   0.1 MRAC Control For Dryer In Paper Making
Process”,IEEE International Conference on Advanced
Communications, Control and Computing
Technologies,Ramanathapuram, India 8-10 May 2014.
[4]. S. B. Pingale, S. P. Jadhav, V. P. Khalane, “Design of
fuzzy model reference adaptive controller for inverted
pendulum”,International Conference on Information
Processing (ICIP),Pune, India,16-19 Dec. 2015.
[5]. P. Jain and M.J. Nigam, “Designing On Controller for
Second Order System",Advance in Electronic and
Electric Engineering,ISSN 2231-1297, Volume 3,
Number 4 , pp. 477-484,2013
[6]. I. Landau, et al.," Adaptive Control:Algorithms, Analysis
and Applications", London. Springer-Verlag, 2011.
[7]. M. Shehu , M. R. Ahmad , A Shehu and Ahmad
Fig.7.b .System response for  2 Alhassan, "LQR, Double –PID and Pole Placement
Stabilizatiion and Tracking Control of Single Link
Inverted Pendulum",IEEE International Conference on
Control System, Computing and Engineering
(ICCSCE),George Town, Malaysia 27-29 Nov. 2015.
[8]. Claudio Melchiorri, "Dynamic model of robot
manipulators".http://wwwlar.deis.unibo.it/people/cmelch
iorri/Files_Robotica/FIR_05_Dynamics.pdf
[9]. S. Coman and Cr. Boldişor, "Model Reference Adaptive
Control for a DC Electrical Drive", Bulletin of the
Transilvania University of Braşov ,Series I: Engineering
Sciences, Vol. 6 (55) No. 2 , 2013 .
[10]. P. Swarnkar ,S Jain and R. K. Nema, "Effect of
Adaptation Gain in Model Reference Adaptive
Fig.7.c .System response for   4.5 Controlled Second Order System",Engineering,
Technology & Applied Science Research Vol. 1, No.3,
V. CONCLUSION 2011, 70-75.
[11]. A. Benaskeur and A. Desbiens,"Application of adaptive
The experiment is performed in simulation. Feedback backstepping to the „Stabilization of the inverted
response without any controller is shown above and it is pendulum",IEEE, 0-7803-4314-X,1998
[12]. S. Pankaj and J. S.Kumar ,"Comparative Analysis of
observed that the system is not stable and contains huge
MIT Rule and Lyapunov Rule in Model Reference
oscillations. In such case where frequent parameter Adaptive Control Scheme", Vol 2, No 4, 2011.
variation is present in the system, adaptive controller [13]. P. Agrawal, R. Singh, R.Yadav," Model Reference
gives effective performance. Hence MRAC controller is Adaptive Control (MRAC) Techniques For Control Of
used where the controller parameters are adjusted to give Dc Motor," Volume 02, Issue 05, Paper id - IJIERM-II-
a desired closed-loop performance throughout the V-1430, October 2015.
working of the system. The adaptation gain (α) is varied
in a range from 0.1 to 5 .The best result is obtained for α
equal to 4.5, for the selected range of adaptation gain.In

View publication stats