基于LSTM网络的风机齿轮箱轴承故障预警  

Fault Early Warning of Fan Gearbox Bearing Based on LSTM Network

作　　者：王超 李大中[1] WANG Chao;LI Dazhong(School of Control and Computer Engineering,North China Electric Power University,Baoding 071003,China)

机构地区：[1]华北电力大学控制与计算机工程学院,河北保定071003

出　　处：《电力科学与工程》2020年第9期40-45,共6页Electric Power Science and Engineering

基　　金：河北省自然科学基金资助项目(F2011502001)。

摘　　要：提出一种将互信息特征选择、深度学习和残差滑动窗口分析相结合的风机齿轮箱轴承故障预警方法。首先,对风电场SCADA数据进行预处理,采用互信息法筛选与齿轮箱轴承温度关联度高的特征。在此基础上,依据所选特征建立长短期记忆神经网络(LSTM)深度学习模型对齿轮箱轴承温度进行预测,并通过滑动窗口对预测残差进行分析处理,设定合适的报警阈值和规则。以西北某风电场现场数据对提出的方法进行验证,结果表明该方法可以对风机齿轮箱轴承故障进行有效预警。A fault warning method for fan gearbox bearing is proposed,which combines mutual information feature selection,deep learning and sliding window residual analysis.First of all,the wind farm SCADA data are preprocessed,and the mutual information method is used to screen the characteristics with high correlation with gearbox bearing temperature.On this basis,according to the selected characteristics,the long short term memory neural network(LSTM)deep learning model is established to predict the gearbox bearing temperature,the prediction residual is analyzed and processed through the sliding window,and the appropriate alarm threshold and rules are set.The field data of a wind farm in northwest China are used to verify the proposed method.The results show that the method can effectively warn the fan gearbox bearing fault.

关 键 词：SCADA数据 深度学习 LSTM网络 齿轮箱轴承 故障预警 

分 类 号：TH17[机械工程—机械制造及自动化]