摘要
微粒群算法是相对较新颖的优化算法,已成功应用于许多优化问题,但该算法容易陷入局部极值。惯性权值的选择方案的好坏,起到举足轻重的作用,本文提出三种惯性权值的改进方案。通过对4种常用测试函数进行测试,结果表明这些改进方案比经典惯性权值选择方案具有更低的平均最好适应值,快速收敛到全局最优解,优化效率明显提高。
PSO algorithm is a relatively new optimization algorithm and has been successfully used in many optimization problems,but the algorithm is vulnerable to local extreme.The paper proposed three improvement ways based on the selection of inertia weight scheme.The test results proved the improvement ways shows that the average of the best fitness of the algorithm is lower than classic inertia weight.The ways can rapidly converge to the global optimal solution;the optimization efficiency is increased significantly.
出处
《北京电子科技学院学报》
2009年第4期55-61,共7页
Journal of Beijing Electronic Science And Technology Institute
关键词
PSO
优化
群智能
微粒群
惯性权值
PSO
optimization
swarm intelligence
multi-particle swarms
inertia weight
