摘要
针对粒子群优化算法在训练小波网络存在的早熟收敛问题,提出一种改进的粒子群优化算法.该算法通过引入多粒子信息共享策略,使种群中各粒子共享多个粒子的有用信息,以期增加种群多样性,减少各粒子在进化早期被吸引到最优粒子附近的可能性,提高小波网络的建模质量.仿真表明,相对于BP算法和标准粒子群优化算法,本文算法在训练小波网络方面估计精度更高,收敛速度更快,并有效解决了早熟收敛和局部最优问题.
For prematurity convergence in applying Particle Swarm Optimization(PSO) algorithm to training of Wavelet Neural Network(WNN),an improved PSO algorithm is proposed.To ameliorate population diversity,a multi-particles information sharing strategy is introduced into Simple PSO algorithm in order to reduce the likelihood that each particle in population is attracted to the neighborhood of the optimum particle during the early period of evolution.Simulations show that,by contrast with BP and Simple PSO algorithm,this algorithm is better in accuracy,convergence rate and prematurity elimination in WNN training.
出处
《昆明理工大学学报(理工版)》
北大核心
2010年第5期52-55,65,共5页
Journal of Kunming University of Science and Technology(Natural Science Edition)
关键词
粒子群优化
小波网络
信息共享
早熟收敛
particle swarm optimization
wavelet neural network
information sharing
prematurity convergence