基于多小波和峭度准则的风力发电机滚动轴承故障检测  

Early Fault Detection of Wind Turbine Rolling Bearings Based on Multi-Wavelet and Kurtosis Criterion

作　　者：聂永辉[1] 徐明文 张译丹[1] 

机构地区：[1]东北电力大学电气工程学院

出　　处：《东北电力大学学报》2019年第6期15-23,共9页Journal of Northeast Electric Power University

基　　金：国家自然科学基金项目(51507029);国家自然科学基金重大项目(2018YFB0904203)吉林省教育厅“十三五”科学技术研究项目(JJKH20180445KJ);吉林市科技创新发展计划项目(201750203);吉林省教育厅科技计划项目(JJKH20180435KJ)

摘　　要：风力发电是全球未来最重要的代替能源,由于其风电机组工作在恶劣的条件下,易造成风力发电机局部出现故障.风力发电机组的滚动轴承故障振动信号呈现非线性和非平稳特点,大量背景噪声污染导致故障特征难以有效识别,提出了多小波和谱峭度相结合的风力发电机滚动轴承故障特征提取方法.首先对振动信号进行多小波降噪,计算其峭度值,评判风机轴承是否产生故障;其次依据快速峭度图算法的自适应选择性获得最优的滤波器参数,滤波后对其进行平方包络分析;最后提取高频共振信号中包含的低频信息,判断风机的故障类型.通过仿真实验结果表明,对于风机轴承微弱的故障诊断,该方法能排除强烈的噪声干扰,保留易丢失的有用信号,明显提高信噪比,精确识别出故障特征频段,有效地的进行故障诊断.Wind power generation is the most important alternative energy source in the world in the future,Due to the harsh conditions of wind turbines.The early fault vibration signals of rolling bearings of wind turbines exhibit nonlinear and non-stationary characteristics.A large number of background noise pollutions make it difficult to identify fault features effectively.A method for extracting fault characteristics of wind turbine rolling bearings based on multi-wavelet and spectral kurtosis is proposed,Firstly,multi-wavelet noise reduction is applied to the vibration signal,and the kurtosis value is calculated to judge whether the fan bearing is faulty.Secondly,according to the adaptive selectivity of the fast kurtosis graph algorithmthe optimal filter parameters are obtained,and the filter is squared.Network analysis,finally extract the low frequency information contained in the high frequency resonance signal.The results show that the method can eliminate the strong noise interference in the early weak fault diagnosis of the fan bearing,retain the useful signal that is easy to lose,significantly improve the signal-to-noise ratio,accurately identify the fault characteristic frequency band,and effectively diagnose the fault.

关 键 词：风力发电机组 滚动轴承 多小波变换 谱峭度 故障检测 

分 类 号：TM76[电气工程—电力系统及自动化]