Based On Type Ⅱ T-s Model And Predictive Control Of Chaotic Particle Swarm Optimization Algorithm

Predictive control is a new control technology based on the controlled model. Since it hasgood control performance, strong robustness and can handle variables’ constraints effectively, ithas successfully been applied to industrial control process, and show great superiority. Butpredictive control algorithm for nonlinear systems is hard to obtain good control performance.For nonlinear systems, the predictive control algorithm often faces two main difficulties:(1) How to obtain an accurate model for the controlled nonlinear system.(2) How to solve the nonlinear optimization problems during the sampling period.In this paper, for a class of nonlinear system, a predictive control algorithm is given basedon type-2T-S fuzzy model with satisfaction clustering mothed and chaotic particle swarmoptimization.Type-2T-S fuzzy modeling method based on the G-K clustering algorithm is often usedunder the assumption which the number c of parameters’ clustering centers is given in advance.In fact, since lack of enough knowledge, it is hard to give a reasonable classification number c.In this paper, based on satisfaction clustering method, a new type-2T-S modeling method isgiven.In this paper, based on the type-2T-S fuzzy model and the chaotic particle swarmoptimization algorithm, a predictive control algorithm for the controlled nonlinear system isgiven. The chaotic particle swarm optimization algorithm combines the particle swarmoptimization algorithm with the chaos optimization algorithm. It has the advantages which arethe quickly searching ability for the optimal solution of the particle swarm algorithm, the strongjumping-out ability from local extremum of the chaos optimization algorithm, and the abilityavoiding to fall into the local extremum shortcomings. It improves precision of the particleswarm algorithm and chaos optimization algorithm. The chaotic particle swarm optimizationalgorithm is used as the receding optimization strategy to realize the continuous optimization. Itcan effectively avoid to solve a large number of matrix inverse problem and relative complexgradient calculation, and it can quickly get the most effective solution.For a numerical example, the simulation results are given using Matlab simulation tool toshow that the proposed method is feasible and valid.