Research on nonlinear predictive control algorithm based on BP-ARMAX combination model

To solve the noise interference problem of multivariable system, a nonlinear predictive control algorithm based on BP-ARMAX combination model is presented. The algorithm builds the system model by the BP neural network and the recursive extended least square method, the dynamic ARMAX model can describe the characteristic of system noise, therefore the model's accuracy is enforced, and the predictive model can be more close to the actual system. We analysis and simulate the curve tracking features and control accuracy of the predictive control algorithm. Simulation results show that, the combined model predictive control algorithm has perfect control effort. © 2014 Binary Information Press.