摘要
针对粒子群算法搜索精度不高特别是对高维函数优化性能不佳问题,提出了一种动态扩散粒子群算法(DDPSO)。该算法通过非线性函数调节惯性权重,在粒子速度更新方式上增加一个动态随机数加强粒子的搜索能力,提高算法的性能,同时在一定条件下对粒子进行重新扩散,保证种群的多样性。实验结果表明,DDPSO算法的寻优能力明显高于基本PSO及其他一些改进的PSO算法,并且该算法性能稳定,更加适合高维复杂函数优化问题。
Dynamic Diffusion Particle Swarm Optimization （DDPSO） was proposed to improve the poor search quality of the standard PSO for optimizing high-dimensional function. A nonlinear function was introduced to adjust the inertia weight and it added a dynamic random number in the updating mode of particle velocity to enhance the searching ability. Meanwhile, particles were diffused again under certain conditions in order to ensure diversity of the swarm. Simulations show that dynamic diffusion particle swarm optimization has outstanding performances in high-dimensional function optimization compared with standard or other modified PSOs.
出处
《计算机应用》
CSCD
北大核心
2010年第1期159-161,共3页
journal of Computer Applications
基金
宁波市自然科学基金资助项目(2008A610002
2009A610090)
浙江教育厅项目(Y200803228)
关键词
粒子群算法
动态随机数
惯性权重
扩散
Particle Swarm Optimization （PSO）
dynamic random number
inertia weight
diffusion
