上下限領域に閉じ込めた Particle Swarm Optimization の力学系の分岐特性と収束特性

概要

論文の詳細を見る
In this paper, we propose two types of dynamics of Particle Swarm Optimization (PSO) embedded into upper and lower bounds, which have more powerful abilities of searching the global optimal solution on their bounded constraints. The new type of dynamics called "nonlinear operator model" and "nonlinear variable transformation model" are inertial models, which are introduced by discretizing the convolution integral form on continuous time. After certifying stability and/or instability including chaos of dynamics by the bifurcation diagrams, the infuences of model parameters settings on the global search are investigated to achieve better performance than the orthodox PSO dynamics in copmutational simulations for a few benchmark objective functions.
社団法人電気学会の論文
2006-07-01