摘要
传统电子网络环境下的神经网络故障搜索算法,粒子群停滞于局部极值点,故障检测率低。提出电子网络环境下故障数据粒子群融合搜索算法,在基本PSO算法的基础上引入进化速度因子,得到改进的带扰动项PSO算法,避免算法停滞粒子处于局部极值点。在改进PSO算法中设计加速因子,使得每个粒子快速集合到局部最优解,以提高收敛速度。将模式搜索法与改进PSO算法相融合,引导粒子群搜索最优位置,实现电子网络环境下的故障数据搜索。为减少计算量,初始步长使用可伸缩的模式搜索法。实验结果表明,所提算法具有较低的误差、较高的收敛速度。
The traditional neural network fault search algorithm in the electronic network environment,particle swarm optimization( PSO) stagnates at local extreme points,and the fault detection rate is low.A particle swarm optimization( PSO) algorithm for fault data fusion in electronic networks is proposed.Based on the basic PSO algorithm,an improved PSO algorithm with perturbation term is proposed by introducing the evolutionary speed factor to avoid the stagnant particles at local extreme points. In order to improve the convergence speed,an acceleration factor is designed in the improved PSO algorithm so that each particle can quickly converge to the local optimal solution. Combining the pattern search method with the improved PSO algorithm,the particle swarm optimization( PSO) is guided to search the optimal position,and the fault data search in the electronic network environment is realized. In order to reduce the computational complexity,Scalable mode search method is adopted for initial step size. The experimental results show that the proposed algorithm has lower error and higher convergence speed.
作者
宋定宇
SONG Ding-yu(Nanyang Institute of Technology,Nanyang 473000,China)
出处
《中国电子科学研究院学报》
北大核心
2018年第5期590-594,共5页
Journal of China Academy of Electronics and Information Technology
关键词
电子网络
故障数据
粒子群
扰动项
初始步长
模式搜索
electronic network
fault data
particle swarm
disturbance term
initial step length
patternsearch