Smith predictive double controllers design based on dynamic neighbor particle swarm optimization algorithm

For the unknown time-delay system of predictive compensation control, a dynamic neighborhood topology particle swarm optimization (PSO) algorithm is presented to optimize the parameters of dynamic neural networks, which is taken as a predictor and identifier in the new double-controller Smith predict structure, respectively. By using the particle swarm optimization space search capability index, the neighborhood topologies of PSO algorithm are dynamically created to optimize the neural network parameters. After that, the combination model is applied to the new two double-controller structure, which separates the load disturbance and fixed value control, and improves the control precision and robustness of Smith predictive compensation model. Finally, simulation results show the effectiveness of the proposed method.