基于小波包与支持向量机结合的齿轮故障分类研究  

A New and Effective Method of Gear Fault Diagnosis Using Wavelet Packet Transform Combined with Support Vector Machine

作　　者：李永龙[1] 邵忍平[1] 曹精明[1] 

机构地区：[1]西北工业大学机电学院,陕西西安710072

出　　处：《西北工业大学学报》2010年第4期530-535,共6页Journal of Northwestern Polytechnical University

基　　金：国家自然科学基金(50575187);航空科学基金(01I53073);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)资助

摘　　要：文章通过对齿轮系统在不同的运转状态下故障类型进行试验测试分析,采集了有关的振动测试信号,对振动特征信号进行了小波阈值去噪,运用小波包方法对信号进行分解,然后对分解后的各层信号进行重构,并计算各层的能量,将它作为故障特征,在此基础上将各层信号特征作为输入,运用支持向量机对它们进行分类,将所得结果与神经网络分类的结果进行了比较。研究表明,去噪处理后的效果比没有去噪的信号特征更加明显,而采用小波与支持向量机结合的方法,对于单一故障和复合故障都能够进行很好地区分与诊断,其诊断成功率均在92%以上。该方法不仅可对实际工程工作的齿轮系统进行故障诊断,而且可用于其它故障诊断领域。After briefly analyzing past research,we propose a new and effective method.Section 1 of the full paper briefs the decomposition and reconstruction of wavelet packet transform.Section 2 briefs the classification principles of a support vector machine.Section 3 presents our method of gear fault diagnosis.Its core consists of：（1） we analyze the vibration features of testing signals of a gear system in different running conditions by wavelet de-noising with thresholds;（2） we decompose the feature signals into different frequency bands with the wavelet packet transform（WPT） and then calculate the energy percentage of every frequency band component to obtain its fault detection index used for fault diagnosis by the support vector machine（SVM）.The fault diagnosis results,obtained with the method proposed by us and their comparison with the method that uses only neural network,given in Figs.2,3 and 4 and Tables 1 through 4,show preliminarily that：（1） the vibration features of the denoised signals are superior to those of signals that are denoised with the methods that do not use the WPT and SVM;（2） all the diagnosis rates for various kinds of faults are over 92%.What is worth mentioning in particular is that our method can also effectively diagnose compound faults.

关 键 词：小波阈值去噪 小波包变换 支持向量机 特征提取 故障诊断 齿轮系统 

分 类 号：TH113.1[机械工程—机械设计及理论]