基于小波分析与迁移学习的风电机组关键部件故障诊断系统研究  

Research on Fault Diagnosis System of Wind Turbine Key Components Based on Wavelet Analysis and Transfer Learning

作　　者：秦晓梅 王有杰 俞啸[2] Qin Xiaomei;Wang Youjie;Yu Xiao(Xuzhuang Street Technology Office,Taibaihu District,Jining,Shandong 273500,China;IOT Perception Mine Research Center,China University of Mining and Technology,Xuzhou,Jiangsu 221008,China)

机构地区：[1]济宁市太白湖区许庄街道办事处科技办公室,山东济宁273500 [2]中国矿业大学,物联网研究中心,江苏徐州221008

出　　处：《机电工程技术》2025年第5期134-138,147,共6页Mechanical & Electrical Engineering Technology

基　　金：国家重点研发计划项目(2017YFC0804400,2017YFC0804401)。

摘　　要：为进一步提高风电机组关键部件的在变工况场景中的故障诊断精度,满足工业现场诊断实际应用需求。首先设计了一种信号-图像转换方法,利用连续小波变换对旋转部件产生的振动信号进行时频分析处理,研究信号中低频故障频率和高频冲激频率分布特点,并采用图像融合方法构建CWT时频特征图,充分表达振动信号中包含的关键故障特征信息;其次针对变工况场景下故障诊断精度下降问题,设计出基于ConvNeXt和DJP-MMD的迁移故障诊断模型,实现对旋转部件运行状态的深度域不变特征的自适应提取,实验结果表明所设计的故障迁移诊断方法具有不错的诊断效果,平均准确率达到90.7%;最后设计了一套基于物联网架构的风电机组关键部件诊断应用系统,实现风电机组多源传感信息的采集、远程诊断和运维管理等功能,满足实际工作场景中风电机组状态在线监测与预测性维护的需求。In order to further improve the fault diagnosis accuracy of the key components of wind turbine in the variable working condition scenario,and to meet the practical application requirements of industrial field diagnosis.A signal-image conversion method is firstly designed to analyze and process the vibration signals generated by rotating parts in time-frequency analysis using continuous wavelet transform.The low-frequency fault frequency and high-frequency impulse frequency distribution characteristics of the signal are studied,and the image fusion method is used to construct the CWT time-frequency feature map,which fully expresses the key fault feature information contained in the vibration signal.Secondly,in view of the problem of decreasing fault diagnostic accuracy in the scenario of variable operating conditions,the migratory fault diagnostic model based on the ConvNeXt and DJP-MMD is designed to realize the adaptive extraction of invariant features of the depth domain of the rotating component’s operating state,which is the best solution to the problem.The experimental results show that the designed fault transfer diagnosis method has good diagnostic effect.The average accuracy reaches 90.7%.Finally,a set of wind turbine key component diagnosis application system based on the Internet of Things architecture is designed to realize the functions of wind turbine multi-source sensing information collection,remote diagnosis and operation and maintenance management,which can satisfy the needs of wind turbine condition online monitoring and predictive maintenance in the actual working scenarios.

关 键 词：风电机组 小波分析 深度学习 故障诊断 物联网 

分 类 号：TK83[动力工程及工程热物理—流体机械及工程]