带有复形法局部搜索的粒子群算法求解最优控制问题

摘要

对最优控制问题的求解是实现计算机控制的前提条件,但有效地求出控制系统的全局最优解是困难的.现今各种求解此问题的算法需要增强算法的优化性能与简洁性.遗传算法求此类问题需要进行离散化;复形法、粒子群算法求解此类问题容易陷入局部极值.为此提出用带有复形法局部搜索的粒子群算法求解此类问题.此算法充分考虑粒子群算法与复形法的特性,将复形法的局部搜索与粒子群算法的全局搜索结合起来,以提高算法搜索能力,克服粒子群算法与复形法易陷局部极值的不足.通过性能测试效果良好,同时算法简便、可行、高效.最后将所提算法用于求解Park-Ramirez生物反应器补料流率的动态优化,取得了满意的效果.%It is presuppose for using computer to control the optimal control system that solving out optimal control problem. But it is difficult to effectively find out the global optimization solution of the control system. Now, the existed algorithms for this problem are need to be improved in optimal performance and simplicity. It is need to discrete when using genetic algorithm for this problem,and particle swarm optimization is easy to be trapped into local minima. Aiming at these problems,proposed a new algorithm,combined the complex local searching,the particle swarm optimization with complex local searching( PSOC)to solve the problem. The algorithm overcame PSO and MC were easy to be trapped into local minima in optimizing multimodal function. The result of a case study shows that PSOC is convenient, feasible, and efficient. When appling to dynamic optimizing feed-rate of Park-Ramirez bioreactor, the results are satisfied.