带粒子释放和速度限制的粒子群算法  

Particle swarm optimization with particle release and speed limit

作　　者：吴正科[1] 杨青真[1] 施永强[1] 李岳锋[1] 

机构地区：[1]西北工业大学动力与能源学院,西安710072

出　　处：《计算机应用研究》2013年第3期682-683,687,共3页Application Research of Computers

摘　　要：针对粒子群算法存在的收敛性与多样性之间的矛盾,在惯性权重线性递减的基础上,提出了带粒子释放和速度限制的算法。用粒子释放来增加种群多样性,引导算法跳出局部最优;用速度限制来促使释放后的粒子快速收敛到全局最优,进而平衡粒子的释放速度和收敛速度,保证算法的收敛。通过五个测试函数的实验,证实了粒子释放和速度限制的综合应用显著提高了算法的寻优能力和收敛精度。To resolve the conflict between the convergence and the diversity of particle swarm optimization（ PSO）, this paper proposed an algorithm with particle release and speed limit based on linearly decreasing of the inertia weight. It used particle release to increase swarm diversity, leading the algorithm to leap out of local optimums. It inducted speed limit to make re- leased particles converge to global optimum quickly, and thus to ensure the convergence of the algorithm. The experiments to optimize 5 benchmark functions show that the possibility of finding a best solution is improved by introducing the method of particle release and speed limit, so as to the precision.

关 键 词：粒子群优化 线性递减 速度限制 粒子释放 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]