Bộ điều khiển PID dựa trên mạng nơ-ron hàm cơ sở xuyên tâm

Abstract

This paper is to proposed an algorithm for estimating parameters (i.e., Kp, Ki, Kd) of a PID (proportional integral derivative) controller using RBF (radial basis function) neural network. The proposed algorithm is then applied to control direct current motor and a degree-of-freedom robot arm. MATLAB based simulation results show that the RBF based PID gives better performance than the conventional PID for both cases: the fixed and changed parameters of the DC motor and 1 DOF robot arm. It is noteworthy that the proposed approach gives the step response with insignificant overshoot, fast settling time (2 s), and the rise time (0.8 s) is shorter than the conventional PID.