基于PCA与DHMM的发动机故障诊断研究  

Fault Diagnosis for Diesel Engines Based on Principal Component Analysis and Discrete Hidden Markov Model

作　　者：黄家善[1] 张平均[1] 

机构地区：[1]福建工程学院电子信息与电气工程系,福建福州350108

出　　处：《厦门大学学报（自然科学版）》2010年第5期617-621,共5页Journal of Xiamen University：Natural Science

基　　金：福建省教育厅基金(JA08160);福建省科技厅高校预研基金(GY-Z0880)

摘　　要：提出了主元分析方法(PCA)与离散隐马尔可夫模型(DHMM)相结合的发动机故障诊断方法.首先利用PCA进行发动机振动信号的特征提取,设计了基于PCA的振动信号故障特征提取算法;然后采用DHMM进行故障诊断,构建了基于DHMM的故障分类器和诊断数据库,设计了柴油机振动信号的故障诊断策略.实际应用结果显示,该方法是可行的,对柴油机故障的诊断精度,达到了94.17%,具有工程应用价值.Fault diagnosis based on principal component analysis（PCA） and discrete hidden markov model（DHMM） for engine are studied in the paper.First,the vibration signals′ feature extraction from the diesel engine is realized by PCA;next,the vibration signal feature extraction algorithm is designed;then DHMM is applied for fault diagnosis;furthermore,a fault classifier based on DHMM with diagnostic databases is developed;and,finally,the fault diagnosis strategies of diesal vibration signal is conceived.The practical application results show that the method proposed in this paper is feasible for the accuracy of diesel engine fault diagnosis is 94.17%.

关 键 词：主元分析方法 离散隐马尔可夫模型 故障诊断 发动机 工程机械 

分 类 号：TK421.6[动力工程及工程热物理—动力机械及工程]