A design of predictive PID control systems using GA and GMDH network

Abstract

PID control scheme has been widely used in most process control systems, because the PID controller has a simple structure, and three parameters at the most. However, since it is difficult to choose suitable parameters,lots of works have been reported with respect to tuning scheme of PID parameters. In this paper, a design scheme of predictive PID control system using a GMDH (group method of data handling) multilayer neural network and a genetic algorithm (GA) is proposed. Furthermore, the effectiveness of the newly proposed scheme is evaluated on a simulation example.