基于SAE与改进LightGBM算法的笼型异步电机故障诊断方法  

Fault detection method of cage asynchronous motor based on stacked autoencoder and improved LightGBM algorithm

作　　者：许伯强[1] 何俊驰 孙丽玲[1] XU Bo-qiang;HE Jun-chi;SUN Li-ling(School of Electrical and Electric Engineering,North China Electric Power University,Baoding 071003,China)

机构地区：[1]华北电力大学电气与电子工程学院,河北保定071003

出　　处：《电机与控制学报》2021年第8期29-36,共8页Electric Machines and Control

基　　金：国家自然科学基金(51277077)。

摘　　要：针对笼型异步电动机故障诊断传统的特征值方法的局限性,提出了一种基于栈式自编码(SAE)和改进的轻型梯度提升机算法的笼型异步电动机定子绕组匝间短路和转子断条故障的联合诊断方法。该方法主要包含如下两个步骤,第一步骤是应用SAE故障检测算法对电机定子电流与电压采样数据进行处理,并通过添加惩罚项(稀疏项和加噪环节)对输入数据进行降维编码,获取数据重构值,以实现特征自动提取;第二步骤是针对难以处理的权重赋值问题、误分类代价和过拟合等问题改进LightGBM故障检测算法,将上述第一步骤得到的数据重构值引入LightGBM分类器进行电机故障多分类,从而实现定子绕组匝间短路和转子断条故障的同时、联合诊断。实验结果表明,该方法是有效的。Because of the limitation of traditional eigenvalue method for fault diagnosis of squirrel cage induction motor,a combined diagnosis method based on stacked autoencoder and improved LightGBM algorithm is proposed.Firstly,SAE fault detection algorithm was applied to process the sampled data of motor stator current and voltage,and penalty terms(sparse term and noise adding link)were added to reduce the dimension of the input data and obtain the data reconstruction value to realize automatic feature extraction.Then,the LightGBM fault detection algorithm was improved for the problems of weight assignment,misclassification cost and over fitting,which are difficult to deal with.The data reconstruction value obtained was introduced into LightGBM classifier for multi classification of motor faults to realize the simultaneous and joint diagnosis of stator winding inter turn short circuit and rotor broken bar faults.The experimental results show that the method is effective.

关 键 词：异步电动机 故障诊断 自编码 轻型梯度提升机 定子绕组匝间短路 转子断条 

分 类 号：TM343[电气工程—电机]