采用变频器故障树为训练样本的BAM神经网络故障诊断方法

A Fault Diagnosis Method for Inverter Based on BAM Neural Network and Fault Tree

研究故障树分析(FTA)和双向联想记忆(BAM)神经网络在故障诊断中的应用,提出了一种融合FTA和BAM的故障诊断方法。利用FTA得到系统所有的故障模式,进而由故障模式和根据维修经验的故障分析归纳出BAM的学习样本,即故障模式和故障分析之间的对应。BAM通过联想记忆矩阵并行联想,得到诊断结果,扩展综合故障诊断能力。用上述方法对变频器故障诊断进行仿真分析,结果表明该方法用于解决变频器故障问题是有效的。

The application of Fault Tree Analysis（FTA） and Bidirectional Associative Memory（BAM） network in fault diagnosis was studied,then a fault diagnosis method based on FTA and BAM was proposed.All of the fault modes of a system were obtained by using FTA.On which basis,the learning samples of BAM were summarized according to the fault modes and the maintenance experience,which were the corresponding relation between the fault modes and the fault analysis.The results of diagnosis were obtained through associative memory matrix,thus could expand the general ability of fault diagnosis.The method was used for fault diagnosis simulation of an inverter,and results showed that the approach is effective.