You are on page 1of 7

International Journal of Scientific Research Engineering & Technology (IJSRET), ISSN 2278 – 0882,

36
Volume 3 Issue 1, April 2014

Design of a Hybrid Controller using Differential Evolution and MIT Rule for
Magnetic Levitation System
Priyank Jain*, M. J. Nigam**
*Department of Electronics and Communication Engineering, Indian Institute of Technology, Roorkee, India
**Department of Electronics and Communication Engineering, Indian Institute of Technology, Roorkee, India

ABSTRACT tuning results in non-linear behavior of the overall


In various industrial systems, parameters variation is one control system which results in good performance where
of the major problems faced by control engineers now nonlinearities and disturbances are inherent part of the
days. To overcome the problem of parameter variations, system [4].
this paper proposes the hybridization of MIT rule based Another approach to automatically tune the parameters
online tuning of classical PID controllers with of conventional controller is Differential Evolution Soft
Differential Evolution algorithm. The hybridization of Computing Algorithm widely known as DE algorithm
two techniques results in the offline as well as online [5]. This technique uses a population based search
tuning of PID controllers at the same time. The algorithm to estimate the controller parameters so as to
developed hybrid controller is then applied on magnetic minimize the integral square error (ISE) and this
levitation system using MATLAB and Simulink. The estimation is done automatically through writing a
paper also describes the basic steps involved in program using MATLAB [5-8]. DE based PID tuning
Differential Evolution (DE) algorithm and a comparison comes under offline tuning methods which sets the
between designed hybrid controller and the simple DE values of PID parameters based on a performance index
algorithm based controller has been carried out. The [6].Literature suggests that the performance of DE based
results depict that the performance of the hybrid controller is better than MIT based controller. This paper
controller is better than the offline tuned PID controller proposes the hybridization of above two techniques and
in terms of transient parameters such as peak overshoot the performance of the hybrid controller has been
and settling time. evaluated on magnetic levitation system. Section II-IV
explains strategies, simulations and results with
Keywords–Differential evolution, hybridization, MIT necessary graphs.
rule, PID controller, soft computing.
II. MIT RULE
I. INTRODUCTION
There are various approaches used to design the
Inherent disturbances and inaccuracies lead to adjustment mechanism for an adaptive controller and
parameter variations in any physical system which may gradient theory based MIT rule is one of them. MIT rule
result in degradation in the performance and sometimes uses the alteration of controller parameters in the
damage the system. To solve the problem of parameter negative direction of gradient of a cost function [4]. This
variation, one needs to design the control system with cost function is defined in terms of the error between the
more powerful and advanced techniques so as to actual behavior and ideal behavior of the plant.
maintain the satisfactory performance of the overall Defining the cost variable,
system. 3Adaptive Control is one of the widely used J(k) = e2 / 2 (1)
advanced control strategies, in which one needs to
design an adjustment mechanism to alter the adjustable In (1), e is the error between plant output and reference
parameters of controller [1]. Gradient theory based MIT model output, and k is the adjustable parameter of the
rule is one of them, which uses the concept of altering controller.
the adjustable parameters of conventional PID controller Applying gradient theory [4],
in the direction so that the error between plant output dk ∂J
∝− (2)
and reference input can be minimized [1-3]. This type of dt ∂k
control is also called online tuned PID control. Online

www.ijsret.org
International Journal of Scientific Research Engineering & Technology (IJSRET), ISSN 2278 – 0882,
37
Volume 3 Issue 1, April 2014

Using the gradient theory, one will get the following (ii) Initialize parameter vectors over lower and upper
equation which depicts the relationship between the limits
change in parameter k with error e(t) [3]. xi,G = [x1,i,G , x2,i,G , . . . xD,i,G] ;
dk/dt = −γ′ e y (3) i = 1, 2, . . . ,N.
Hence the adjustment law developed by the equation
given above using Laplace transform will be, Step 2. Mutation:
γ′ (i) For a given parameter vector xi,G randomly select
k(s) = − ℒ {e(t). y (t)} (4)
three vectors xi,p, xi,q and xi,r such that the indices ‘i’, ‘p’,
s
In (4), ℒ represents the Laplace transformation and ‘s’ ‘q’ and ‘r’ are distinct.
represents the Laplace variable. The adjustment (ii) Add the weighted difference of two of the vectors
mechanism developed by the eq. (4) will be as shown in to the third,
Fig. 1 [4]. vi,G+1 = xp,G + F(xq,G – xr,G) (5)
The mutation factor F is a constant from [0, 2] and vi,G+1
is called the donor vector.
Step 3. Recombination:
(i) Recombination incorporates successful solutions
from the previous generation. The trial vector ui,G+1 is
developed from the elements of the target vector, xi,G,
and the elements of the donor vector, vi,G+1
(ii) Elements of the donor vector enter the trial vector
with probability CR
Fig. 1: Adjustment Mechanism of Model Reference v,, ; rand , <
Adaptive Controller using MIT rule ui,G+1 = (6)
x , , ; rand , >
i = 1, 2, . . . ,N; j = 1, 2, . . . ,D
III. DIFFERENTIAL EVOLUTION Step 4. Selection:
The target vectorx , is compared with the trial vector
Differential Evolution (DE) is a recently introduced
vi,G+1 and the one with the lowest function value is
population search based soft computing algorithm which
admitted to the next generation.
uses heuristic optimization. DE algorithm was first
u, ; f( u , ) < (x , )
introduced by Storn and Price in 1996 to solve the xi,G+1 = (7)
Chebychev Polynomial fitting problem used in filter x, ; otherwise
designing [13]. The decisive idea behind DE is an
arrangement for producing trial parameter vectors and IV. RESULTS AND DISCUSSIONS
the selection of these vectors is based on heuristic
optimization [14-15]. Linear model of Magnetic Levitation system has been
Typical parameters used in DE are listed below; used in the paper taken from [9] for simulations on
 D – problem dimension MATLAB and the transfer function of the system is
 N –No. of Population shown below, where Ex(s) is the position of the ball in
 CR – Crossover Probability terms of voltage and Ei(s) is the applied input voltage
[9].
 F – Scaling Factor
G(s) = Ex(s) ⁄ Ei(s)
 G – Number of generation/stopping condition
 L,H – boundary constraints
G(s) = 77.8421 (8)
(0.0311s 2 − 30.52)
Storn and Price have shown some rules in selecting the
control parameters in their first research paper published Table 1 shows the performance of magnetic levitation
in 1996 [5 & 8]. Theses rule are described with details in system using Differential Evolution based offline tuned
[13] and shown below with brief description; PID controller in terms of transient performance
Step 1. Initialization: parameters along with integral square error (ISE) for
(i) Define upper and lower boundaries [L, H] and various trials. The performance of the overall system is
initialize all DE parameters very satisfactory while considering linear approximated
model of the system in simulations as shown in Table 1.

www.ijsret.org
International Journal of Scientific Research Engineering & Technology (IJSRET), ISSN 2278 – 0882,
38
Volume 3 Issue 1, April 2014

From table 1, one can observe that in every trial there algorithms.The performance of DE based controller with
is an overshoot of around 25-35% present in the appropriately chosen range for various parameters is
response. Although the settling time is much lesser for satisfactory as shown by the results. Selection of this
every trial but such a large overshoot in the system is range is very critical in DE algorithm and is carried out
undesirable and may cause the actuator breakdown. carefully. The limitation of DE based controller is that it
Integral square error (ISE) is well under the desirable does not have the ability to alter the controller
range and the overall performance is satisfactory. parameters during the run time. To overcome this
Fig. 3 shows the Simulink model of the proposed limitation, hybridization of DE algorithm with MIT rule
hybrid controller, in which a PID controller is generating has been proposed in the paper.
the control input for the system. The integral gain of PID Performance of the proposed hybrid controller is
controller is an adjustable gain which is being adjusted evaluated on MATLAB and the designed controller has
by MIT rule based adaptive mechanism as shown in Fig. been applied on magnetic levitation system. The
4. The performance of the designed controller for performance of hybrid controller very much depends
various trials, in terms of transient parameters and ISE, upon the value of adaptation gain and selection of this
is shown in table 2 and the corresponding responses are gain is very critical in designing the controller.
shown in Fig. 5. These trials have been performed to Effectiveness of the hybrid controller also depends upon
calculate the values of PID parameters in offline mode the efficient programming for DE algorithm carried out
using DE algorithm and fine tuning in the value of by an expert. Results carried out and shown in the paper
parameter KI has been carried out by MIT mechanism clearly shows that the performance of hybrid controller
with adaptation gain of 23.5 in online mode. is much better than the DE based controller. Results also
For a particular trial, different responses and depict the reduction in peak overshoot from 30% to 15%
corresponding adjustable parameter for various values of and in settling time from 0.4 sec to 0.3 sec, which proves
adaptation gain have been obtained and are shown in the better performance of proposed hybrid technique.
Fig. 6 and Fig. 7. From Fig. 6, one can observe that for
smaller values of adaptation gain, response of the system REFERENCES
is sluggish with large settling time. As one increases the [1] K. J. Astrom and B. Wittenmark, Adaptive control
value of adaptation gain, the response of the system (2nd edition), Dover Publications Inc., New York,
becomes better. But after a certain limit of adaptation 1995, ch. 5, pp. 185–234.
gain the response of the system starts deteriorating. [2] P. Swarnkar, S. K. Jain and R. K. Nema, “Effect of
Hence, selection of a range for adaptation gain is very adaptation gain on system performance for model
critical factor. reference adaptive control scheme using MIT rule”,
For a particular value of adaptation gain, responses presented at International Conference of World
obtained for various trials are shown in Fig. 5 and the Academy of Science, Engineering and Technology,
performance of the system has been tabulated as shown Paris, pp. 70-75, 2010.
in table 2. On comparing the performances of the system [3] P. Swarnkar, S. Jain, R. K. Nema,“Application of
with DE controller and DE+MIT hybrid controller, one Model Reference Adaptive Control Scheme To
can observe that the performance of the hybrid controller Second Order System Using MIT Rule”, presented
is better in terms of peak overshoot and settling time. at International Conference on Electrical Power
The overshoot and settling time has been reduced to and Energy Systems (ICEPES-2010), MANIT,
15% and 0.30 sec respectively. Results given in the Bhopal, India, 2010.
paper verify that the performance of the system has been [4] Priyank Jain and M. J. Nigam, “Real Time Control
improved to a large extent with the proposed of Ball and Beam System with Model Reference
hybridization of DE algorithm and MIT rule. Adaptive Control Strategy using MIT Rule”,
presented at 2013 IEEE International Conference
V. CONCLUSION on Computational Intelligence and Computing
Research (ICCIC), Madurai, pp. 305-308, 2013.
A brief overview on MIT rule and Differential [5] R. Storn, K. Price, “Differential evolution-a simple
Evolution soft computing algorithm has been carried out and efficient adaptive scheme for global
in this paper along with the hybridization of two optimization over continuous spaces," Journal of
techniques.DE algorithm is very simple and effective Global Optimization, Vol.11, pp. 341-359, 1997.
population based generation algorithm which takes very [6] Rainer Storn, “Differential Evolution for
less computation time than other soft computing Continuous Function Optimization,

www.ijsret.org
International Journal of Scientific Research Engineering & Technology (IJSRET), ISSN 2278 – 0882,
39
Volume 3 Issue 1, April 2014

"http://www.icsi.berkeley.edu/storn/code.html, Conference on Vehicle, Power and Propulsion,


2005. Arlington, TX, pp. 78-83, 2007.
[7] K. Price, “An introduction to differential [12] K. S. Narendra and A. M. Annaswamy, Stable
evolution," New Ideas in Optimization, eds. D. Adaptive Systems, Prentice-Hall, Englewood Cliffs,
Corne, M. Dorigo, and F. Glover, McGraw-Hill, New Jersey, 1989.
London (UK), pp. 79-108, 1999. [13] Mohd S. Saad, HishamuddinJamaluddin, Intan Z.
[8] R. Storn, “On the Usage of Differential Evolution M. Darus “PID Controller Tuning Using
for Function Optimization”. in Proceedings of the Evolutionary Algorithms, WSEAS Transactions on
Fuzzy Information Processing Society, Berkeley, System and Control, Vol. 7, 4, pp. 139-149, 2012.
CA, USA, 1996. [14] K. J. Astrom and T. Hagglund, “Automatic Tuning
[9] AbhishekRawat and M. J. Nigam, “Comparison of PID Controllers”,Instrument Society of America,
between adaptive linear controller and radial basis 1988.
function neural neurocontroller with real time [15] J. G. Ziegler and N.B. Nichols, “Optimum Settings
implementation on magnetic levitation system”, for Automatic Controllers”, Trans. ASME. Vol. 64,
presented at 2013 IEEE International Conference 8: pp. 759–768, 1942.
on Computational Intelligence and Computing [16] B. C. kuo and F. Golnaraghi, Automatic Control
Research (ICCIC), Madurai, pp. 515-518, 2013. Systems (9thedition), John Wiley & Sons, Inc.,
[10] Adrian-VasileDuka, StelianEmilianOltean, and 2010, ch. 4, pp. 147-252.
MirceaDulău, “Model Reference Adaptive vs. [17] G. Saravanakumar, R. S. D. Wahidhabanu, and V.
Learning Control for the Inverted Pendulum”, I. George, “Robustness and Performance of
presented at the International Conference on Modified Smith Predictors for Processes with
Control Engineering and Applied Informatics Longer Dead-Times”, International Journal on
(CEAI), vol. 9, pp. 67-75, 2007. Automatic Control and System Engineering
[11] M. S. Ehsani, “Adaptive Control of Servo Motor by
MRAC Method”, presented at IEEE International

Fig. 2 Simulink diagram of magnetic levitation system with DE based control

TABLE 1: Various values of PID parameters, transient parameters and respective value of ISE on different trials of
DE based controller for Magnetic Levitation system

PID Parameters Settling Integral


Overshoot
Trial Time Square Error
KP KI KD (%)
(sec) (ISE)
I 2.9900 0.0443 15.7818 25.5 0.478 0.0498
II 2.8700 0.0685 18.0636 29 0.32 0.0227

www.ijsret.org
International Journal of Scientific Research Engineering & Technology (IJSRET), ISSN 2278 – 0882,
40
Volume 3 Issue 1, April 2014

III 1.9500 0.0395 19.7300 32 0.37 0.0314


IV 2.7650 0.0673 17.8495 30 0.31 0.0282
V 2.1150 0.0422 19.0800 33 0.35 0.0337

Fig. 3 Simulink diagram of magnetic levitation system with DE and MIT rule based hybrid controller

Fig. 4 Simulink diagram of adjustable PID controller

TABLE 2: Various values of PID parameters, transient parameters and respective value of ISE on different trials of
hybrid controller for Magnetic Levitation system
PID Parameters Settling Integral
Overshoot
Trial Time Square Error
KP KI KD (%)
(sec) (ISE)
I 2.9732 0.0899 15.3613 16.6 0.23 0.0087
II 2.8635 0.2234 18.6812 15.49 0.30 0.0095
III 2.9989 0.4981 16.2816 14.74 0.34 0.0082
IV 2.9006 0.3649 16.2532 15.33 0.26 0.0086

www.ijsret.org
International Journal of Scientific Research Engineering & Technology (IJSRET), ISSN 2278 – 0882,
41
Volume 3 Issue 1, April 2014

Fig. 5 Response of Magnetic Levitation system with DE-MIT hybrid controller for different trials with adjustable
parameter k=23.5

Fig. 6 Response of Magnetic Levitation system with DE-MIT hybrid controller for different values of adaptation gain:
Position of the ball

www.ijsret.org
International Journal of Scientific Research Engineering & Technology (IJSRET), ISSN 2278 – 0882,
42
Volume 3 Issue 1, April 2014

Fig. 7 Response of Magnetic Levitation system with DE-MIT hybrid controller for different values of adaptation gain:
Adjustable parameter

www.ijsret.org

You might also like