Comparison of Machine Learning Algorithms for DC Motor PID Control with Genetic Algorithm

Abstract

Proportional-Integral-Derivative (PID) is a closed loop control system based on dynamic system feedback commonly employed in industrial control systems that require continuous modulated control. The purpose of this project was to improve the efficacy of PID control, a branch of classical control theory, to allow for the application of PID controllers in more complex and non-linear environments. The research compares the results of different machine learning algorithms with genetic algorithms to optimize PID control parameters to achieve more precise control of 12V DC motors. Genetic Algorithm (GA) is a metaheuristic algorithm used to solve search and optimization problems in machine learning. GAs are developed based on the process of natural selection and genetics, relying on biologically inspired operators. After a series of selections and crossovers, the genetic algorithm selected the fittest generation of PID constants that would result in the optimal PID controller which minimizes system error value and linearizes the system behavior. Optimizing PID parameters to enhance the efficacy of classical control theory could increase machining precision in industrial production and improve the robustness of integrated PID controllers in complex environments such as the human body, enabling PID controllers to be an adaptive, simple, and viable option to be implemented in implantable artificial organs and motorized prosthetics. The results of the study suggest that GA optimized PID parameters are effective in regulating the system behavior of DC motors and the enhanced PID control system could be implemented in a wide range of rapidly changing and non-linear applications.