摘要
为了有效避免粒子群算法(PSO)早熟和局部收敛的现象,在深入分析PSO算法的基础上,提出了一种基于高斯白噪声扰动变异的粒子群优化算法(GMPSO)。该算法以一定的概率选中粒子进行基于高斯白噪声扰动的变异,并重新随机产生飞离搜索区域的粒子,以克服粒子群后期多样性严重下降的缺点。通过对Benchmark函数的测试表明:GMPSO算法无论是搜索精度、速度还是稳定性均显著优于PSO算法。
The particle swarm optimization （PSO） is difficult to deal with the problem of premature and local convergence, so an improved PSO algorithm based on mutation of Gaussian white noise disturbance （GMPSO） is proposed. In GMPSO, the mutation is undertaken by selecting the particles with certain small probability, and the particles that fly out the field of solution will be regenerated so as to overcome the disadvantage of the droping of diversity in the later development of the PSO algorithm. The experimental results on Benchmark functions show that GMPSO algorithm is obviously superior to the PSO in convergence precision, convergence rate and stability.
出处
《华东理工大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2008年第6期859-863,共5页
Journal of East China University of Science and Technology
基金
国家杰出青年科学基金(60625302)
国家自然科学基金面上项目(60804029)
国家863计划课题(2007AA04Z193,2007AA04Z192)
上海市科委项目(07JC14016,08DZ1123100)
高等学校学科创新引智计划(B08021)
上海市重点学科建设项目(B504)
上海市国际科技合作基金项目(08160710500)
关键词
粒子群优化算法
高斯白噪声
变异
多样性
particle swarm optimization
Gaussian white noise
mutation
diversity
