摘要
标准粒子群算法是一种有效的寻找函数极值的演化计算方法,它简便易行,收敛速度快。但算法也存在收敛精度不高,易陷入局部极值点的缺点。针对这些缺点,在原有算法的基础上,提出一种动态改变惯性因子的粒子群优化算法(DCW-PSO),使得粒子在迭代过程中惯性因子随粒子进化度和聚合度的变化而改变。文中通过对测试函数的仿真实验,并与自适应改变惯性因子的粒子群算法(ACWPSO)以及标准粒子群算法比较,其结果表明这种改进的粒子群算法在性能上明显优于这两种粒子群算法。
The normal PSO algorithm is a validated evolutionary computation way of searching the extremum of function , which is simple in application and quick in convergence , but low in precision and easy in premature convergence. Because of the limitation , a dynamically changing inertia weight PSO algorithm is proposed based on the normal PSO algorithm . The inertia weight is changed in every loop according to the swarm evolution degree and aggregation degree factor . Compared with ACWPSO and the normal PSO , the optimization results of testing function show that the performance of the DCWPSO algorithm is more excellent.
出处
《计算机仿真》
CSCD
2007年第5期154-157,共4页
Computer Simulation
基金
江苏省自然科学基金项目(BK2005012)
关键词
优化算法
动态惯性因子
进化度
聚合度
Optimization arithmetic
Dynamical intera weight factor
Evolution degree
Aggregation degree
