Self-Adaptation in Genetic Algorithms for Control Engineering: The Case of Time Delay Systems

In this paper we present a mechanism of self-adaptation for genetic algorithms with application in the field of control engineering. In recent years, along with the increase in computational power, evolutionary optimization techniques have been increasingly used for the design of controllers. Some methods of design perform parameter tuning of controllers with a chosen structure. PID controllers, widely used in industry and practical applications, have mainly benefited from the use of genetic algorithms for their tuning, especially for systems that are not linear time invariant. In this study we consider the tuning of PID controllers for time delay systems by introducing a specially designed performance index and a self-adaptation mechanism that makes use of this index. Results obtained in simulation and experimentally in a laboratory setting validate the proposed approach. Moreover, the self-adaptation mechanism performs well for various evolution operators, which makes the approach generalizable for other problems.