基于小波包与LS-SVM的气阀故障诊断  

Valve Train Fault Diagnosis Based on LS-SVM and Wavelet Packet

作　　者：王旭平[1] 陈小虎[1] 王汉功[1] 周永涛[1] 

机构地区：[1]第二炮兵工程学院501教研室,陕西西安710025

出　　处：《机电工程技术》2009年第5期42-44,95,共4页Mechanical & Electrical Engineering Technology

摘　　要：基于小波包原理,对柴油机的缸盖振动信号进行小波包分解,利用"频带能量"的特征提取方法得到特征向量,并作为LS-SVM的输入进行训练和分类检验,提出了一种基于小波包和LS-SVM的气阀故障诊断方法。结果表明不同状态下的气阀漏气故障能得到识别和分类,且具有较高的精度。Based on the algorithm of wavelet packets, Wavelet packet decomposition is applied to the cylinder head vibration signals, feature vectors are obtained by a feature extraction method with'frequency band energy'. The feature vectors are used for training and classification as the input s of LS-SVM. Hence, a new valve train fault diagnosis method based on LS-SVM and wavelet packet is established. The tests indicate that the leakage fault of exhaust valve in different conditions can be identified and classified, and the result is highly accurate.

关 键 词：气阀机构 故障诊断 小波包 LS-SVM 

分 类 号：TH165.3[机械工程—机械制造及自动化]