A new dynamic matrix control algorithm based on the FNN TS fuzzy model

Abstract

This paper discusses a control design method for the fuzzy neural network (FNN) TS fuzzy model based on which a new dynamic matrix control (DMC) algorithm is presented. The basic interest is motivated by control problems in parameter varying systems because traditional control schemes, such as PID and DMC, do not achieve good performance under all working states in these systems. Based on previous work in identification methods for nonlinear systems and parameter varying systems, this paper further utilizes the mathematical expression characteristics of the TS fuzzy model and presents a new DMC algorithm where the unit step response prediction model is replaced by the TS fuzzy model. The most attractive feature of the resultant algorithm is that controllers designed by the proposed method can be used for all working states with good results. Simulation results for a second-order parameter varying system demonstrate the effectiveness of the suggested method.