文献类型：期刊文献

中文题名：基于XGBoost-LightGBM-LSTM的风机齿轮箱轴承故障预警

英文题名：Fault Warning of Wind Turbine Gearbox Bearings Based on XGBoost-LightGBM-LSTM

作者：俞国燕[1];李少伟[1];董晔弘[2]

机构：[1]广东海洋大学机械工程学院,广东湛江524000;[2]中国船舶重工集团海装风电股份有限公司,重庆401122

年份：2023

期号：6

起止页码：140

中文期刊名：轴承

外文期刊名：Bearing

收录：CSTPCD、、Scopus、北大核心、北大核心2020

基金：广东省区域联合基金资助项目(2019B1515120017);广东省海洋经济发展(海洋六大产业)专项资助项目(GDNRC[2021]42)。

语种：中文

中文关键词：滚动轴承;风电轴承;风力发电机组;滑动窗口;残差;故障;预警

外文关键词：rolling bearing;wind turbine bearing;wind turbine;sliding window;residual;fault;warning

中文摘要：针对风电机组齿轮箱温度预测准确性较低,泛化能力差的问题,提出一种极端梯度提升树(XGBoost)、轻量梯度提升机(LightGBM)和长短时记忆网络(LSTM)加权融合的组合模型对齿轮箱轴承温度进行预测。采用灰色关联度(GRA)选取与齿轮箱轴承密切相关的特征参数作为组合预测模型的输入,利用训练好的组合模型预测齿轮箱轴承正常工作温度,计算与实际温度值之间的残差,并用滑动时间窗口设置预警阈值,从而进行齿轮箱轴承故障预警。通过江苏某海上风场5 MW风机实际数据验证表明,该组合模型对风电机组齿轮箱轴承温度预测精度较好,并能提前进行故障预警。

外文摘要：Aimed at the problems of low accuracy and poor generalization ability of temperature prediction in wind turbine gearboxes,a combined model of extreme gradient boosting(XGBoost),light gradient boosting machine(LightGBM)and long short-term memory(LSTM)weighted fusion is proposed to predict the temperature of wind turbine gearbox bearings.The grey relation analysis(GRA)is used to select the feature parameters closely related to the bearings as input for combined prediction model.The trained combined model is used to predict the normal working temperature of the bearings,and the residual between predicted value and actual value of temperature is calculated.The sliding time windows are used to set the warning threshold,so as to carry out the fault warning of the bearings.The actual data validation of a 5 MW wind turbine in an offshore wind farm in Jiangsu Province shows that the combined model has good prediction accuracy for temperature of the bearings and can provide fault warning in advance.