Control of Time Varying Nonlinear System Based on RBFNN-DMC Algorithm

In this paper, a novel dynamic matrix control algorithm based on RBF neural network for time varying nonlinear system is presented. RBFNN is used for system model identification, as well as DMC is adopted as optimized controller. Besides, the predictive initiative value is solved by multi-steps prediction of RBF neural network, and nonlinear dynamic matrix coefficients are derived correspondingly. Compared to regular DMC algorithm, the RBFNN-DMC algorithm not only effectively overcomes the large disturbance but also be very robustness. At last, the algorithm is applied in a time varying, high nonlinear Continuous Stirred Tank Reactor (CSTR) pH process model and presents a better control performance.