基于数据融合的风电机组主传动链典型故障分层诊断方法  

Hierarchical Diagnosis Method of Typical Faults ofWind Turbine Main Drive Chains Based on Data Fusion

作　　者：程凯 王鹏宇 王晓东[2] 包涛 吴宇航 杨苹[3] CHENG Kai;WANG Pengyu;WANG Xiaodong;BAO Tao;WU Yuhang;YANG Ping(Digital Grid Research Institute,China Southern Power Grid,Guangzhou 510630,China;SPIC Guangxi Electric Power Co.,Ltd.,Nanning 530000,China;Guangdong Key Laboratory of Clean Energy Technology,South China University of Technology,Guangzhou 510641,China)

机构地区：[1]南方电网数字电网研究院,广州510630 [2]国家电投集团广西电力有限公司,南宁530000 [3]华南理工大学广东省绿色能源技术重点实验室,广州510641

出　　处：《噪声与振动控制》2023年第3期125-131,174,共8页Noise and Vibration Control

基　　金：中国南方电网有限责任公司科技资助项目(SYYKJXM20210054);广东省重点领域研发计划资助项目(2021B0101230003)。

摘　　要：风电机组主传动链故障是影响风电机组年运行时间的主要故障类型,提高其诊断精度是确保风电场稳定可靠运行的关键。为提升风电机组主传动链故障诊断精度,在主传动链上增加高频振动监测系统对其振动信号进行精细化分析。由于分别根据高频振动监测信号与风电机组数据采集与监视控制系统(Supervisory Control and Data Acquisition,SCADA)实时监测信号独立进行故障诊断时,难以发现风电机组主传动链的早期故障。为此,提出一种基于3类数据源融合的智能故障诊断方法,通过融合SCADA实时监测系统振动数据、非振动数据和振动监测系统振动数据3类不同时间尺度数据,建立基于自编码网络的典型故障分层诊断模型。实际诊断案例表明,基于3类数据源融合的典型故障分层诊断模型可准确定位风电机组的典型故障。The main drive chain fault of wind turbine is the main fault type affecting the annual operation time of wind turbines.Improving its diagnosis accuracy is the key to ensure the stable and reliable operation of wind farms.In order to improve the fault diagnosis accuracy of the main drive chain of wind turbines,a high-frequency vibration monitoring system is added to the main drive chain to make a fine analysis of its vibration signal.When using the high-frequency vibration monitoring signal and the real-time monitoring signal of wind turbine’s Supervisory Control and Data Acquisition(SCADA)to carry out fault diagnosis independently,it is difficult to find the early fault of wind turbine’s main drive chain.Therefore,an intelligent fault diagnosis method based on the fusion of three kinds of data sources is proposed.By fusing the vibration data and non-vibration data of SCADA real-time monitoring system and the vibration data of vibration monitoring system,a typical fault hierarchical diagnosis model based on autoencoder network is established.The actual diagnosis case shows that the typical fault hierarchical diagnosis model based on the fusion of three kinds of data sources can accurately locate the typical faults of wind turbines.

关 键 词：故障诊断 风电机组 数据融合 典型故障 智能诊断 

分 类 号：TM315[电气工程—电机]