Gradient descent based linear regression approach for modeling PID parameters

Abstract

The paper proposes a method for developing a linear model for Proportional Integral Derivative (PID) parameters of a PID controller. The approach is based on linear regression and it uses gradient descent algorithm for parameter optimization. A deep perusal of the literature shows that there have been many improvements made in developing PID tuning methods, and improving its performance. But a mention of a linear modeling of PID parameters is not noticed in any of the literature. The highlight of this method is the dependency of the tuning parameters on certain selected transient specifications. The speed control system of a dc motor is identified for the present study.