摘要
针对基本粒子群优化算法(PSO)易陷入局部极值点,进化后期收敛慢,精度较差等缺点,提出了一种改进的粒子群优化算法。该算法用一种无约束条件的随机变异操作代替速度公式中的惯性部分,并且使邻居最优粒子有条件地对粒子行为产生影响,提高了粒子间的多样性差异,从而改善了算法能力。通过与其它算法的对比实验表明,该算法能够有效地进行全局和局部搜索,在收敛速度和收敛精度上都有显著提高。
A modified particle swarm optimization algorithm （MPSO） is proposed for improving the disadvantages of basic PSO, as tending to trap into a local optimum, converging slowly in last period of evolution, possibly bringing a consequence in low precision and so on. A random and unconditional mutation strategy which substitutes for previous velocity is presented, and the effect which the best position of neighbor particle has conditionally on the particle behavior is considered. It efficiently increases diversity of particles and improves the performance of algorithm. Comparing with other optimization algorithms in the contrast experiment, the new algorithm can greatly accelerate convergence speed and improve convergence precision by exploring the local and global minima efficiently.
出处
《计算机工程与设计》
CSCD
北大核心
2008年第11期2893-2896,共4页
Computer Engineering and Design
关键词
粒子群优化
变异
进化计算
函数优化
群智能
particle swarm optimization
mutation
evolutionary computation
function optimization
swarm intelligence
