Abstract:

A novel generalized predictive control (GPC) strategy based on the Hammerstein-Wiener model is proposed. The dynamic characteristics of the nonlinear system are described by the Hammerstein-Wiener model based on the support vector machine, so a prediction model of the controlled object is obtained. Furthermore, an optimization algorithm of chaotic particle swarm combined with quasiNewton trust region (QN-TR) is proposed in order to avoid the deficiency of slow convergence speed and low accuracy of the genetic algorithm and the chaotic particle swarm optimization （CPSO） algorithm, so a rolling optimization strategy of the predictive control is obtained. Function tests and rolling optimization of the GPC to the nonlinear object reflect the superiority of the algorithm. Finally, the resultsof the simulation example for the generalized predictive controller show that it can meet the demand of real-time and stable operation of the system, and a good control effect is obtained.