A Neural Networks Based Direct Adaptive Controller for Multivariable Systems

This paper presents a direct multivariable adaptive controller using neural network which adapts to the changing parameters of the multivariable nonlinear system with nonminimum phase behavior, mutual interactions and time delays. It base on the theory which a nonlinear multivariable systems to be controlled is divided a linear part and a nonlinear part. The controller parameters of the linear part are obtained by the recursive least square algorithm at the parameter estimation stage, whereas the nonlinear part is achieved the through the Back-propagation neural network. This controller is performed on-line. In order to demonstrate the effectiveness of the proposed algorithm, the computer simulation results are presented to adapt a nonlinear multivariable system with nonminimum phase, noises and time delays and with changed system parameter after a constant time. The proposed method is effective compared with the conventional direct multivariable adaptive controller using neural network.