Direct Nonlinear Controller Design Based on Virtual Reference and Support Vector Machine

Abstract

This paper proposed a new direct nonlinear controller design method based on virtual reference(VR) and support vector machine(SVM), which allows to directly design nonlinear controller on the base of input/output data with no need of a model of the plant. Firstly, the relation between virtual reference feedback tuning(VRFT) and internal model control(IMC) was analyzed. Then, the structure and design procedure of the proposed nonlinear controller was given. Simulation results demonstrate this method can effectively deal with the noise and nonlinearity and eliminate the steady-state error. Moreover, the amount of calculation decreases apparently compared to normal indirect model reference control method using neural network(NN).