Sei sulla pagina 1di 34

BACHELOR THESIS PROJECT

PID CONTROLLER TUNING USING SOFT


COMPUTING TECHNIQUES
PULAK MALIK SARTHAK GARG SHASHWAT BHAGERIA SHUBHAM SINGH
513/IC/14 526/IC/14 531/IC/14 536/IC/14
ICE-2 ICE-3 ICE-3 ICE-3
NSIT NSIT NSIT NSIT
CONTENTS
ABSTRACT

NEED OF COMBINING META-HEURISTIC ALGO

PROPOSED SOLUTION

ADVANTAGES OF PROPOSED SOLUTION

RESULTS

CONCLUSION

REFERENCE
ABSTRACT
Artificial bee colony (ABC) is an algorithm, which has proved its importance in solving various
practical problems including engineering optimization problems. ABC algorithm is distinctively known
for being the youngest and most popular member of the family of population-based nature inspired
meta-heuristic swarm intelligence method. ABC has proved to be one of the most effective
algorithms over various other Nature Inspired Algorithms when being used for benchmark functions
and real world problems. The performance of search process of ABC depends on a random value
which tries to stabilize exploration and exploitation phase. With the purpose to enhance the
performance, it is required to balance the exploitation of optimal solution of the ABC and exploration
of search space. This report outlines 2 new hybrid of ABC algorithm with Genetic Algorithm that are
Genetically Mutated Bee Colony Optimization Technique (GMBC) and Tuning of GA parameters using
ABC. This report reviews Artificial Bee Colony (ABC) and Genetic Algorithm (GA), both of which are
two powerful meta-heuristics. This report explains some major defects of these two algorithms at
first then proposes 2 new hybrid models. Experimental results show that proposed hybrid algorithms
are effective and their performance including speed and accuracy beats other version.
Need of Combining Meta- Heuristic Algorithm
As we know that many practical control problems faced by control engineers are
non-linear in nature having higher order with time delay elements. The
Metaheuristic algorithm (GA & ABC) generally works good for linear system, but for
non-linear systems these algorithms converges to local maxima and many times they
even don’t give satisfactory result.
Genetic algorithm being very robust have high convergence rate for lower order
problems but for higher order practical problems convergence rate reduces and they
provide local maxima as optimal solution.
Artificial Bee Colony method being flexible, easy to implement and good exploration
of the solution but exploitation to found food sources is very bad and falls into local
optimum solution as a result of premature.
PROPOSED SOLUTION

1. Genetically mutated bee colony optimisation technique

2. Tuning of GA parameters using ABC


Genetically mutated bee colony optimisation
technique
1. Initialize the variables as mentioned above 9. Onlooker bees select the solution around which they explore for new food sources

2. Create initial bee population by assigning random position within search space  The selection is done probabilistically i.e. by Roulette Wheel selection

3. Calculate cost of each position, from the cost function defined in the program  The Roulette Wheel method requires fitness values which are computed from cost values

4. Now, sort population based on cost 10. Those bees food source whose abandonment counter exceeds abandonment limit will become Scout

5. Store details of 1st population in BestSol (variable to store best solution ever found) bees and they searches around the hive randomly.

6. Apply crossover and mutation to the population array 11. Now sort all the food sources explored based on cost and truncate the top food sources.

7. Initialize the abandonment counter values and set it to zero 12. Apply Crossover and Mutation on population so formed.

8. Now in the main loop 13. Store the best solution found in BestSol

 for employed bees search for the new position of food source nearby and calculate the cost. 14. Repeat the main loop for n iterations

 If calculated cost is better, then update the new bee position 15. Optimal values of Kp, Ki and Kd are stored in BestSol

 Else increment the abandonment counter


Tuning of GA parameters using ABC
1. Define cost function

 This cost function will optimize the values of GA parameters i.e. (pm, pc and gamma) 8. Onlooker bees select the solution around which they explore for new food sources

 GA parameters are optimized based on different cost function which will optimize PID  The selection is done probabilistically i.e. by Roulette Wheel selection

parameters based on weighted sum of integral square error (ISE), Maximum peak overshoot  The Roulette Wheel method requires fitness values which are computed from cost values

(Mp), Settling time (Ts) 9. Those bee’s food source whose abandonment counter exceeds abandonment limit will become

2. Initialize the variables as mentioned above Scout bees and they searches around the hive randomly.

3. Create initial bee population by assigning random position within search space 10. Now sort all the food sources explored based on cost and truncate the top food sources.

4. Calculate cost of each position, from the cost function defined in the program 11. For evaluating the cost, GA is evaluated for optimal value of mutation percentage(pm), crossover

5. Store details of Best Solution found in BestSol (variable to store best solution ever found in percentage(pc) and range factor for crossover(gamma).
randomly initialized population) 12. Store the best solution found in BestSol
6. Initialize the abandonment counter values and set it to zero
13. Repeat the main loop for n iterations.
7. Now in the main loop
14. Optimal values of Kp, Ki and Kd are stored in BestSol
 for employed bees search for the new position of food source nearby and calculate the cost.

 If calculated cost is better, then update the new bee position

 Else increment the abandonment counter


Advantages of the proposed solution
 Ability to overcome local maxima problem faced by Ga
 Easy to converge for complex problem
 Ability to handle multidimensional Cost Function
 Works even with Larger Population Size
 Converges to optimal solution within few iterations.
 Can be applied to problems of various practical domain
 Less dependent on user input variables
 Applied to higher order practical systems
 It guarantees optimality.
RESULTS:
PLANT 1:

The transfer function for G1(s) is given as:


plant = 4.228 / ((s + 0.5) *(s^2 + 1.64*s + 8.456))
Cost function as:
f = 10(ISE) +3(ts) +(Mp)
Step response of G1(s) from Genetic Algorithm
Step response of G1(s) from Artificial Bee Colony
Step response of G1(s) from GA-ABC-1
Step response of G1(s) from GA-ABC-2
Comparison of step response of different algorithms for plant 1

GA – Genetic Algorithm

ABC – Artificial Bee Colony optimizatio

GA-ABC-1 – Genetically Mutated Bee

Colony optimization technique

GA-ABC-2 – Tuning of GA parameters

via ABC
Comparison of different parameters of plant 1

GA ABC GA-ABC-1 GA-ABC-2


Kp 2.7299 2.6622 2.7021 2.1075
Ki 1.0661 1.1131 1.0920 1
Kd 2.2590 2.5633 2.4573 1.2396
Cost function 18.6857 18.5883 18.5015 18.1574
Settling time 5.0527 4.8786 4.6423 4.1460
Rise time 0.4749 0.4471 0.44506 2.2364
Overshoot(%) 0.6453 1.1.2366 0.9060 1.7402
PLANT 2:

The transfer function for G1(s) is given as:


Plant = 27/((s + 1)*(s + 3)^3);
Cost function as:
f = ISE +5(ts) +(Mp) + tr
Step response of G2(s) from Genetic Algorithm
Step response of G2(s) from Artificial Bee Colony
Step response of G2(s) from GA-ABC-1
Step response of G2(s) from GA-ABC-2
Comparison of step response of different algorithms for plant 2

GA – Genetic Algorithm

ABC – Artificial Bee Colony optimizatio

GA-ABC-1 – Genetically Mutated Bee

Colony optimization technique

GA-ABC-2 – Tuning of GA parameters

via ABC
Comparison of different parameters of plant 2

GA ABC GA-ABC-1 GA-ABC-2


Kp 2.3456 2.3430 2.3378 1.8699
Ki 1.4367 1.4698 1.3924 1.2647
Kd 1.9954 1.9348 2.0348 1
Cost function 16.0457 16.5926 15.9564 9.5458
Settling time 2.9183 2.8914 2.9696 1.6767
Rise time 0.6581 0.6686 0.6535 1.0127
Overshoot(%) 0.3377 1.0030 2.4662 X 10-4 1.67 X 10-7
PLANT 3:

The transfer function for G1(s) is given as:


Plant = exp(-0.1*s)/ ((1+s) *(1+ 0.5*s) *(1 + 0.25*s) *(1 + 0.125*s ))
Cost function as:
f = ISE +5(ts) +(Mp) + tr
Step response of G3(s) from Genetic Algorithm
Step response of G3(s) from Artificial Bee Colony
Step response of G3(s) from GA-ABC-1
Step response of G3(s) from GA-ABC-2
Comparison of step response of different algorithms for plant 3

GA – Genetic Algorithm

ABC – Artificial Bee Colony optimizatio

GA-ABC-1 – Genetically Mutated Bee

Colony optimization technique

GA-ABC-2 – Tuning of GA parameters

via ABC
Comparison of different parameters of plant 3

GA ABC GA-ABC-1 GA-ABC-2


Kp 2.3518 2.3554 2.0768 1.9494
Ki 1.3937 1.4473 1.3196 1.3664
Kd 1.7281 1.8147 1 1
Cost function 16.3863 16.7707 9.5766 9.073
Settling time 2.7967 2.7097 1.4142 1.5191
Rise time 0.5748 0.5534 0.8240 0.8674
Overshoot(%) 1.3424 2.1921 1.0900 0.0141
CONCLUSION
Combination of Genetic Algorithm (GA) and Artificial Bee
Colony (ABC) algorithm was employed to tune PID controllers
for plants of high order and with time delay. The method
optimized PID gains as design variables in both single- and
multi-objective approaches. The objective functions taken
from literature were important performance indices of ISE
and ITAE as well as overshoot and settling time. Results
clearly expressed that the utilized method has been
successful in comparison to Genetic Algorithm (GA) and
Artificial Bee Colony (ABC) alone; and can be considered as a
powerful tuning scheme for controllers.
REFERENCE
[1]. M. Abachizadeh, M. R. H. Yazdi and A. Yousefi-Koma, "Optimal tuning of PID controllers using Artificial
Bee Colony algorithm," 2010 IEEE/ASME International Conference on Advanced Intelligent Mechatronics,
Montreal, ON, 2010, pp. 379-384.
[2]. D. C. Meena and A. Devanshu, "Genetic algorithm tuned PID controller for process control," 2017
International Conference on Inventive Systems and Control (ICISC), Coimbatore, 2017, pp. 1-6. doi:
10.1109/ICISC.2017.8068639.
[3]. Zhen-Yu Zhao, M. Tomizuka and S. Isaka, "Fuzzy gain scheduling of PID controllers," in IEEE Transactions on
Systems, Man, and Cybernetics, vol. 23, no. 5, pp. 1392-1398, Sep/Oct 1993. doi: 10.1109/21.260670.
[4]. Sandeep Kumar, Sharma Kumar, Vivek, Rajani Kumari “A Novel Hybrid Crossover based Artificial Bee Colony
Algorithm for Optimization Problem” International Journal of Computer Applications, vol. 82, issue 8, pp.
18-25
[5]. M Gopal,Control Systems(Principles and Design),Tata McGraw Hill,2008.
THANK YOU

Potrebbero piacerti anche