基于小波包神经网络的级联式变频器故障诊断仿真研究  

Fault diagnosis simulation study of cascaded converter based on wavelet packet neural

作　　者：王新[1] 徐娟[1] 

机构地区：[1]河南理工大学电气工程与自动化学院自动化研究所,河南焦作454000

出　　处：《河南理工大学学报（自然科学版）》2012年第6期706-711,共6页Journal of Henan Polytechnic University（Natural Science）

基　　金：河南省高校科技创新人才支持计划项目(2008HASTIT022);河南省重点科技攻关项目(082102240008)

摘　　要：以级联式变频器为研究对象,重点研究功率器件开路故障的诊断方法.对变频器输出电压进行小波包分解,寻求变频器正常和故障情况下的输出电压频带能量的变化规律.选取输出电压能量变化大的频带能量值作为特征向量,利用BP神经网络进行故障诊断.研究结果表明,小波包神经网络在级联式变频器故障诊断中具有较高的可行性和有效性.The power component open circuit faults diagnosis method of a cascaded converter is mainly stud- ied. According to the output voltage signal features of a frequency-domain, an energy eigenvector was estab- lished by the means of a wavelet packet. Then fault pattern recognition of the cascaded converter was presen- ted using BP neural network. The result shows that wavelet packet neural network in the cascaded converter fault diagnosis has high feasibility and validity.

关 键 词：级联式变频器 小波包分析 BP神经网络 故障诊断 

分 类 号：TH165.3[机械工程—机械制造及自动化]