基于多传感器融合的异步电动机故障诊断研究  

Fault Diagnosis of Asynchronous Motors Based on Multi-sensor Fusion

作　　者：古玉锋[1] 肖子叶 燕钢强 黎程山 李昆鹏[1] GU Yufeng;XIAO Ziye;YAN Gangqiang;LI Chengshan;LI Kunpeng(Key Laboratory for Highway Construction Technology and Equipment of Ministry of Education,Chang'an University,Xi'an Shaanxi 710064,China)

机构地区：[1]长安大学道路施工技术与装备教育部重点实验室,陕西西安710064

出　　处：《机床与液压》2025年第7期16-23,共8页Machine Tool & Hydraulics

基　　金：国家自然科学基金项目(52205249);陕西省自然科学基础研究计划项目(2022JQ-434)。

摘　　要：针对电动机故障诊断方法中存在的单一传感器信号所含故障信息有限以及浅层学习模型故障诊断准确率较低等问题,提出一种基于多传感器融合的卷积神经网络(CNN)和长短期记忆(LSTM)网络的故障诊断方法。通过多传感器同步采集电动机的多源信息,并结合多源同类传感器信息的性质和特点,采用基于熵权法的数据加权融合方法,实现了电动机多源同类信息的数据层融合。构建CNN-LSTM故障诊断模型,自动提取多源异类信息的特征,完成特征层融合。最后,通过搭建三相异步交流电动机故障模拟实验平台,对该故障诊断算法进行实验验证。结果表明:该方法可有效实现电动机定子、转子及轴承的故障诊断,平均准确率达到99.53%,与1D-CNN、LSTM及仅使用单一振动信号的CNN-LSTM模型相比,准确率分别提高了6.41%、9.11%、28.39%。In view of the limited fault information contained in a single sensor signal,and the low fault diagnosis accuracy of shallow learning model in motor fault diagnosis,a fault diagnosis method based on multi-sensor fusion,combining the advantages of convolutional neural network(CNN)and long short-term memory(LSTM)network was proposed.Multiple sensors were used to collect multi-source information synchronously,and combined with the properties and characteristics of multi-source similar sensor information,the data-weighted fusion method based on the entropy weight method was used to realize the data layer fusion of multi-source similar information of motors.The CNN-LSTM fault diagnosis model was constructed to automatically extract the features of multi-source dissimilar information and complete the feature layer fusion.Finally,the three-phase asynchronous AC motor fault simulation experiment bench was built to verify the fault diagnosis algorithm.The results show that the method can effectually realize the fault diagnosis of the motor stator,rotor and bearings,and the average accuracy rate is 99.53%,and the accuracy is improved by 6.41%,9.11%and 28.39%,compared with the 1D-CNN,LSTM and CNN-LSTM models which only use one single vibration signal,respectively.

关 键 词：异步电动机 故障诊断 多传感器融合 

分 类 号：TM343[电气工程—电机] TP206[自动化与计算机技术—检测技术与自动化装置]