基于改进粒子群优化神经网络的电力变压器故障诊断  

Power Transformer Fault Diagnosis Based on Neural Networks with Improved Particle Swarm Optimization

作　　者：程加堂[1] 熊伟[1] 徐绍坤[1] 艾莉[1] 

机构地区：[1]红河学院工学院,蒙自661100

出　　处：《高压电器》2012年第2期42-45,共4页High Voltage Apparatus

基　　金：红河学院科研项目(10XJY117)~~

摘　　要：为了提高电力变压器故障诊断的准确性,采用了一种自适应变异粒子群优化神经网络的方法,用于BP网络的权值优化。并根据变压器的故障特征,用优化好的BP网络进行故障诊断。该算法修正了粒子个体行动,克服了标准粒子群和BP网络易陷入局部极小的问题。实例仿真结果表明,该方法具有较好的分类效果,具有一定的实用性。In order to improve the accuracy of fault diagnosis of power transformer,a method of neural networks optimized by adaptive mutation particle swarm optimization （AMPSO） is adopted, in which the improved PSO is combined with the BP algorithm to optimize the weight of BP neural network. According to transformer fault feature, the fault diagnosis is accomplished via the optim/zed neural network. This algorithm amends individual particle＇s act, and overcomes the problem that the basic PSO and BP network are vulnerable to local minima. Simulation results show that the method is of better classification effectiveness and nracticalitv_

关 键 词：电力变压器 粒子群优化算法 自适应变异 神经网络 故障诊断 

分 类 号：TM41[电气工程—电器]