基于小波包分析和LS-SVM的柴油机故障诊断方法  

Fault Diagnosis Approach of Diesel Engine Based on Wavelet Packets Analysis and LS-SVM

作　　者：左来[1] 

机构地区：[1]中南林业科技大学机电工程学院,湖南长沙410004

出　　处：《计算机测量与控制》2009年第11期2150-2152,共3页Computer Measurement &Control

基　　金：中南林业科技大学青年基金重点项目(05003A)

摘　　要：针对某柴油机检测样本小,难以准确估计故障的状况,提出了一种基于小波包分析和最小二乘支持向量机的柴油机故障诊断方法;利用小波包分析对柴油机缸盖振动提取频谱能量并对干扰信号进行处理,从而获得故障征兆样本集;由于柴油机故障的征兆样本集有限性,提出了采用最小二乘支持向量机分类方法构建柴油机故障分类器;结果表明,经过小波处理过后的振动信号再经过LS-SVM辨识网络训练,能够准确地诊断和预测故障。In view of diesel engine diagnosis system which has small testing samples and is difficult to accurately estimate the situation of failure, fault diagnosis approach based on wavelet packets analysis and least square support vector machine is presented. The frequency spectrum energy of the cover was extracted as evidential samples by wavelet packets decomposition. In reality the diesel engine fault diagnosis evi- dential samples is limited, least square support vector machine approach is presented to construct status diagnosis classifiers. Simulation resuits demonstrate that after have being dealt with by wavelet packet, the vibration signals be trained by LS--SVM identification network can accurately diagnosis and predict fault failure.

关 键 词：柴油机 最小二乘支持向量机 故障诊断 小波包 

分 类 号：TP206.3[自动化与计算机技术—检测技术与自动化装置]