基于EEMD-CC和PCA的风电齿轮箱状态监测方法  

Condition Monitoring of Wind Turbine Gearbox Based on EEMD-CC and PCA

作　　者：马越[1] 陈捷[1] 洪荣晶[1] 潘裕斌[1] MA Yue;CHEN Jie;HONG Rong-jing;PAN Yu-bin(College of Mechanical and Power Engineering,Nanjing Tech University,Jiangsu Nanjing 210009,China)

机构地区：[1]南京工业大学机械与动力工程学院,江苏南京210009

出　　处：《机械设计与制造》2019年第5期67-71,共5页Machinery Design & Manufacture

基　　金：2013国家自然科学基金(51375222);2014年度高校"青蓝工程"中青年学术带头人;江苏省研究生教育教学改革课题(KYCX17_0937)

摘　　要：齿轮箱是风力发电机组的核心传动部件,不仅结构复杂制造成本高,而且故障率高维修费用巨大,对其进行状态监测具有重要意义。针对风电齿轮箱在复杂工况下运行所产生的非线性、非平稳振动信号,提出了一种基于EEMD-CC和PCA的风电齿轮箱状态监测方法。该方法先对含有大量噪声的风电齿轮箱振动信号进行集合经验模态分解和相关系数(EEMD-CC)降噪处理。然后,将降噪后的正常信号数据进行主分量分析(PCA)建模,并以T^2统计量和SPE统计量作为信号异常的评判指标。最后,把降噪后的测试数据带入PCA模型中,分别判断T^2和SPE值是否超出阈值,实现风电齿轮箱的状态监测。试验结果证明,该方法能够有效地监测风电齿轮箱的状态。As the key part of wind turbine,gearbox consists of a large amount of complex parts and has a high incidence failure,and the costs of production and maintenance are especially expensive. Thus,it is meaningful to monitor the status of it. The vibration signals generated by the wind power gearbox under complex operating conditions are non-stationary and nonlinear. A wind turbine gearbox condition monitoring method based on Ensemble Empirical Mode Decomposition and Correlation Coefficien t(EEMD-CC)and Principal Component Analysi s(PCA)is proposed to analyze the vibration signals.Firstly,the vibration signal of wind power gearbox was denoised by EEMD-CC. Then,a model for normal vibration signal data was established by PCA and using T^2 and SPE statistics as evaluation indexes. Finally,the denoised test data was brought into the PCA model,and determined whether the T^2 and SPE values exceed the threshold. The experimental results show that the proposed method can effectively monitor the status of wind turbine gearbox.

关 键 词：风电齿轮箱 EEMD-CC降噪 主分量分析 状态监测 

分 类 号：TH16[机械工程—机械制造及自动化]