基于小波包特征提取和模糊熵特征选择的柴油机故障分析  

Fault diagnosis of diesel engines based on wavelet packet energy spectrum feature extraction and fuzzy entropy feature selection

作　　者：蒋佳炜 胡以怀[1] 柯赟 陈彦臻[1] JIANG Jiawei;HU Yihuai;KE Yun;CHEN Yanzhen(Shanghai Maritime University,Shanghai 201306,China;Harbin Engineering University,Harbin 150001,China)

机构地区：[1]上海海事大学,上海201306 [2]哈尔滨工程大学,哈尔滨150001

出　　处：《振动与冲击》2020年第4期273-277,298,共6页Journal of Vibration and Shock

摘　　要：船舶动力设备因故障监测信号样本少、变化缓慢、数据特征呈非线性,使得设备故障模式的准确识别和状态预测比较难。尤其是柴油机振动信号的故障诊断,由于柴油机振动信号噪声多,诊断信号难以进行特征选择的问题,提出了基于小波包能量谱特征提取和模糊熵特征择的柴油机故障诊断方法。利用模糊熵对小波包能量谱提取出的特征集进行特征选择,将选择后的特征参数输入LS-SVM进行故障模式识别。试验结果表明,该方法可以提高故障识别准确率。在该试验中,故障识别准确率达到了99.36%,相比于未进行特征选择的特征集,识别准确率提高了0.72%。Ship power equipment makes fault pattern recognition and state prediction more difficult due to few samples,slow changes,and the nonlinear structure of data of fault monitoring signals.Especially for the diesel engine vibration signal fault diagnosis,due to the noise of the diesel engine vibration signal,the diagnosis signal is difficult in feature selection.This paper presents a diesel engine fault diagnosis method based on wavelet packet energy spectrum feature extraction and fuzzy entropy feature selection.The feature set extracted from the wavelet packet energy spectrum is selected by fuzzy entropy,and the selected feature parameters are input to the LS-SVM for fault pattern recognition.Experimental results show that this method can improve the accuracy of fault recognition.In the experiment of this paper,the fault recognition accuracy rate reaches 99.36%.Compared with the feature set without feature selection,the recognition accuracy rate is increased by 0.72%.

关 键 词：小波包分析 模糊熵 特征选择 支持向量机 柴油机故障诊断 故障模式识别 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]