Abstract: Proportional-Integral-Derivative (PID) controllers have been widely used in process industry for decades from small industry to high technology industry. But they still remain poorly tuned by use of conventional tuning methods. Conventional technique like Zeigler-Nichols method does not give an optimized value for PID controller parameters. In this paper, we optimize the PID controller parameter using Genetic Algorithm (GA), which is a stochastic global search method that replicates the process of evolution. Using genetic algorithm the tuning of the controller will result in the optimum controller being evaluated for the system every time. The GA is basically based on an iterative process of selection, recombination, mutation and evaluation. The performance of Advanced Genetic Algorithm (AGA) is compared with Guo Tao's Algorithm (GTA) and Elite Multi-Parent Crossover Evolutionary Optimization Algorithm (EMPCOA). AGA has a different replacement strategy as compared to EMPCOA which helps to maintain the population diversity and thus reducing the computational time which is proved by the results presented here. The effectiveness of the AGA is also verified for a system with an unstable plant. The PID controller is also tuned with different error criteria viz. Integral Time Absolute Error (ITAE), Integral Square Error (ISE) and Integral Absolute Error (IAE).
Keywords: PID tuning; Genetic Algorithm; Multi-parent crossover; Elite crossover; Discrete recombination.