Design of neural PID controller for reduced order model

Abstract

The aim of this paper is to design a PID controller for higher order systems using the proposed model reduction method with neural networks. In a linear time invariant continuous system (LTICS) the coefficient matrix cannot be stored explicitly in a computer memory. The matrix vector products can be computed relatively in a less expensive manner by using an approximation technique. A novel method is proposed to obtain a reduced model from a higher order linear time invariant continuous system. The proposed scheme is simple, computationally straight forward and does not involve any complex algebra. The reduced order model will always be stable if the higher order system is stable. Using the proposed method, PID controller is designed for the higher order linear time invariant continuous system to meet the performance specifications. The PID controller parameters are tuned by using Back propagation Network.