基于改进PSO-BP混合算法的电力变压器故障诊断  

Power transformer fault diagnosis based on improved PSO-BP hybrid algorithm

作　　者：杨道武[1] 李海如[1] 向卫东[2] 任卓[1] 李哲文 

机构地区：[1]长沙理工大学化学与生物工程学院,湖南长沙410004 [2]长沙电业局,湖南长沙410015 [3]湖南省超高压输变电公司,湖南长沙410100

出　　处：《电力科学与技术学报》2011年第1期99-103,共5页Journal of Electric Power Science And Technology

基　　金：长沙市重点科技攻关项目(k0802126-21-63)

摘　　要：采用粒子群算法和反向传播神经网络建立一种新型变压器故障诊断网络模型,设计故障诊断方法.仿真分析结果表明:基于该网络模型的诊断方法与传统的三比值法相比较,具有较好的故障识别与分类能力,显著提高了诊断准确率,将在电力设备故障诊断中有良好应用前景.A new network model and method are established for power transformers fault diagnosis in this paper.The Particle Swarm Optimization（PSO） technique is used to integrate with Back Propagation（BP） neural networks in this new network model.Compared with the conventional three-ratio method,the fault diagnosis method has better results for power transformers faults diagnosis and classification.Furthermore,the diagnostic accuracy is much improved.Simulation results demonstrate that this method has wide application prospects in the fault diagnosis of power equipments.

关 键 词：变压器 故障诊断 粒子群算法 反向传播网络 

分 类 号：TM407[电气工程—电器]