Nonlinear Multi-step Predictive Control Based on Taylor Approximating method

Based on the neural recursive multi-step predictive strategy, the process´ multi-step predictive outputs are available. Under Taylor series expansion, the process predictive values can be approached more precisely. By minimizing the multistage cost function, a sequence of future control signals is obtained. Compound neural networks are adopted during the processes of identification and recursive prediction. The stability condition of the closed-loop neural network-based predictive control system is demonstrated based on Lyapunov theory. Simulation studies have shown that this scheme is simple and has good control accuracy and robustness.