摘要
通过对标准PSO算法中惯性权重和全局最好值的分析,提出了一种根据全局最好值的变化而自适应变化的随机惯性权重的方法。通过对5个典型的Benchmark函数的测试,结果表明此方法在收敛速度和全局收敛性方面都较线性递减的惯性权重的方法有所改进。最后,将改进的PSO算法应用于分类问题,与标准PSO算法与C4.5的结果相比,分类精度和速度都有所提高。
Based on the analysis of inertia weight and global best fitness of the standard PSO, a PSO method is described with selfadaptive stochastic inertia weight by the change of the global best fitness （WPSO）. By the experiments of five Benchmark function, the results show the performance of WPSO improved more clearly than that of the standard PSO. Finally, the WPSO are applied to mine classification and experiment results compared with C4.5 on 5 data sets indicate better accuracy and more rapid speed.
出处
《计算机工程与设计》
CSCD
北大核心
2006年第24期4677-4679,4706,共4页
Computer Engineering and Design
基金
教育部重点科研基金项目(204018)
关键词
PSO算法
惯性权重
全局最好值
自适应随机惯性权重
分类
PSO algorithm
inertia weight
global best fitness
self-adaptive stochastic weight
classification
