Optimized PI/PD/PID Controller Design for DC Motor Speed Control: A Comparative Study of Random Forest-Based Tuning & Classical Methods

This Study explores ways to improve the speed of DC motors through comparison of traditional tuning methods, viz. Ziegler-Nichols and Cohen-Coon with a machine learning approach using Random Forest. The paper focuses on tuning three types of controllers, PI, PD, and PID to achieve better system performance by reducing errors and achieving stability within the system. The DC motor behavior can be represented mathematically, and the performance of the controllers is evaluated using key parameters such as rise time, settling time, steady-state error, and other error measures. The results show that while Random Forest can predict system behavior fairly well, the traditional methods are more effective in minimizing errors and optimizing the performance of the system. The overall best performance was given by the PID controller. This research gives the value of classical tuning methods, meanwhile bringing evidence of the potential from machine learning to further improve DC motor control.