基于动态分级和邻域反向学习的改进粒子群算法

An improved particle swarm algorithm based on dynamic segmentation and neighborhood reverse learning

针对粒子群算法容易陷入局部最优解的问题,提出了一种基于动态分级和邻域反向学习的改进粒子群算法.该算法通过构建动态分级机制,将种群中的粒子动态地划分成3个等级,对不同等级内的粒子采取不同的扰动行为,使得粒子在增强种群多样性的同时保持向全局最优方向进化;采用粒子智能更新方式,提高了粒子的搜索能力;引入动态邻域反向学习点建立全局搜索策略,促使种群快速寻优.最后,利用多种典型测试函数对该算法进行仿真实验,结果表明,与其他几种优化算法相比,本算法具有较好的收敛性和稳定性.

In order to solve the problem that the particle swarm optimization algorithm is likely to fall into local optimum,an improved particle swarm algorithm based on dynamic segmentation and neighborhood reverse learning(DSNRPSO)is proposed.By setting up a dynamic segmentation mechanism,the algorithm divides the particles in the population into three grades,then employs different perturbation strategies for the particles in different grades,so that the particles maintain the evolution to the global optimal direction while the diversity of the population is enhanced.Furthermore,it adopts the method of particle intelligent updating to promote the search ability of particles,and introduces the dynamic neighborhood reverse point enabling a global search to improve the particle searching speed.The preliminary results show that the proposed algorithm has better convergence and stability than several other kinds of optimization algorithms.