融合分类优化与拓展策略的粒子群优化算法

Fusion Classification Optimization and Expansion Strategies-based Particle Swarm Optimization Algorithm

针对传统的粒子群算法易陷入局部最优、后期收敛速度慢、精度低等缺点,提出了一种融合分类优化与拓展策略的粒子群优化算法.该算法对易陷入局部最优的粒子群采用分类优化拓展策略,淘汰劣质解,并采用拓展策略生成新的优质解,以提高粒子群优化算法的收敛精度,同时通过采取一种正态演化变异策略,搜索当前最优粒子的邻域空间的方式增强局部开采能力,以尽量避免算法陷入局部最优.实验针对6个经典函数利用智能优化搜索算法对其求最小值问题上进行仿真测试,结果表明本文提出的改进算法在解的精度上明显优于一些知名的改进粒子群优化算法,尤其在多峰函数上表现更为突出.

As we all know, traditional particle swarm optimization algorithm is easy to fall into local optimum and has slow convergence ,low accuracy and some other disadvantages. To overcome the disadvantages, a fusion classification optimization and expansion strategies-based particle swarm optimization algorithm is proposed. The algorithm not only uses a classification optimization strategy to eliminate inferior solutions for the particle swarm which falls into local optimum, but also adopts a expansion strategy to generate high quality solutions to improve the convergence precision of particle swarm optimization algorithm. Meanwhile, a normal evolutionary mutation strategy is presented to search the optimal particle neighborhood space, since this may enhance the capacity of local exploration and avoid the algorithm trapping into local optimum. The experiments are conducted on 6 classic functions to find the minimum issue using intelligent search algorithm, andthe results show that the improved algorithm is better than the recent improved particle swarm optimization algorithm inconvergence speed, especially in multimodal function is more outstanding.