基于EMD小波阈值去噪和时频分析的齿轮故障模式识别与诊断  

Gear fault pattern identification and diagnosis using Time-Frequency Analysis and wavelet threshold de-noising based on EMD

作　　者：邵忍平[1] 曹精明[1] 李永龙[1] 

机构地区：[1]西北工业大学机电学院,西安710072

出　　处：《振动与冲击》2012年第8期96-101,106,共7页Journal of Vibration and Shock

基　　金：国家自然科学基金(50575187);航空科学基金(01153073);陕西省自然科学基金(2004E219);西北工业大学研究生创业种子基金(Z2010024)

摘　　要：建立齿轮故障系统试验装置,对齿轮传动系统在各种转速与故障状态下进行测试分析,获取有关振动信号,对齿轮系统的无故障、齿根裂纹、分度圆裂纹、齿面磨损四种状态信号进行特征提取,并对提取的信号进行基于经验模态EMD分解的小波阈值去噪处理,然后对预处理后的信号进行时频分析与诊断。结果表明,采用基于EMD的小波阈值去噪方法比单纯采用小波阈值去噪对测试信号进行预处理,能提高信噪比,并更加有效的提取出故障特征,而在EMD的小波阈值去噪的基础上,再与时频分析方法相结合能够较好的识别不同运转状况下不同种类的故障,如齿根裂纹、分度圆裂纹、齿面磨损等,可用于对实际工程工作的齿轮系统进行故障诊断。The testing equipment of a fault gear system was established. By measuring vibration signals of a gear system at different rotating of signals including signals face abrasion. As the speed for different faults, the test was conducted. The features were extracted from four kinds with no fault, those with tooth root crack, those with pitch circle crack and those with tooth signals of the transmission system were often corrupted by noise, so they were preprocessed using the wavelet threshold de-noising based on empirical mode decomposition （EMD）. The preprocessed signals were investigated using time-frequency analysis. The results showed that the wavelet threshold de-noising based on EMD is better than the wavelet threshold de-noising, and the former can improve the signal-to-noise ratio （SNR） to extract fault features better. After signal preprocessing based on EMD, the results of time-frequency analysis showed that the proposed method is effective for diagnosis of different fault kinds, such as, tooth root crack, pitch circle crack and tooth face abrasion.

关 键 词：经验模态分解 小波阈值去噪 时频分析 损伤检测 故障诊断 齿轮传动系统 

分 类 号：TH113.1[机械工程—机械设计及理论]