摘要
粒子群优化算法本质上是一种全局随机优化技术,优化性能高但容易陷于局部最优,并且算法性能很大程度上依赖于参数设置。本文对该算法的3个控制参数进行数据实验和调查,分析参数设置对算法性能的影响规律,提出一种改进的粒子群优化算法,该算法在迭代的每一代中,惯性权重和加速系数都是在一定范围内随机产生:ω=rand(0.4,0.7),C1=rand(0.5,3.0),C2=rand(1,3.5)。由于该算法的控制参数不再固定取值;而且在一定范围内随机产生,从而增强了算法的多样性和遍历性,能够有效避免算法早熟收敛。通过标准函数的测试,验证了该算法性能优于固定参数粒子群算法和随机加速系数粒子群算法,具有更好的收敛性和稳定性。
Particle swarm optimization (PSO) is a powerful stochastic global technique, but easily trapped into local optimization, and its performance often depends heavily on the parameter settings. Based on analyzing the influence of the parameters setting in the experiment, this paper proposed a new particle swarm optimization algorithm which the inertia weight (ω) and acceleration coef-ficients (c1 and c2 ) are generated as random numbers within a certain range in each iteration process., ω= rand (0.4, 0.7), C1 = rand (0.5, 3.0), c2 = rand (1.0, 3.5). The proposed algorithms apply more particles' information, can easily jump out of local optimum and improve convergence performance. The experimental results demonstrate that the proposed algorithm is superior to the other two algorithms with a better astringency and stability.
出处
《重庆师范大学学报(自然科学版)》
CAS
CSCD
北大核心
2013年第6期123-127,共5页
Journal of Chongqing Normal University:Natural Science
基金
广东省自然科学基金(No.101754539192000000)
关键词
粒子群优化算法
惯性权重
加速系数
随机参数
particle swarm optimization (PSO)
inertia weight
acceleration coefficients
random parameters