基于双谱的柴油发动机活塞销故障诊断  

Bispectrum Analysis of Fault Diagnostics of Piston-Pin of Diesel Engine

作　　者：肖云魁[1] 李会梁[1] 王保民[1] 程广涛[1] 张威[1] 司爱威[1] 

机构地区：[1]军事交通学院汽车工程系,天津300161

出　　处：《内燃机学报》2008年第4期369-373,共5页Transactions of Csice

基　　金：总装备部预研项目(40407030302)

摘　　要：为了从柴油发动机振动信号中提取出活塞销故障特征，采用双谱对振动信号进行了分析，并在双谱模域内沿平行于对角线的直线，按特定步长搜索双谱特征频率面，计算其平均幅值得到信号特征参数，最后利用GA—BP神经网络成功地对故障进行了模式识别。试验结果表明，对于EQ6BT发动机，诊断活塞销故障的最佳部位为发动机缸体下部左右两侧机体，最佳转速为1800r／min，特征频率为1．3—3kHz；信号特征不仅存在于对角线上，还大量存在于对角线以外的区域；采用双谱分析提取故障特征，结合神经网络进行故障诊断，效果良好。To extract the fault characteristics of piston-pin from the vibration data of diesel engine, the method of bispectrum is used to analyze the data. Meanwhile, bispectral characteristic planes are searched along the parallel to the diagonal line at certain step in the bispectral modulus field. The mean magnitude is calculated to get the characteristic parameters. The GA-BP artificially neural network is used to successfully diagnose the fault. Experimental result shows that, for EQ6BT engine, the best positions to diagnose the piston-pin fault are the low places on both sides of engine cylinder; the best engine speed is 1 800 r/ min, and the feature frequency is 1.3 -3 kHz. The signal characteristics are also existed away from the diagonal line. Combination of bispectral analysis and artificially neural network in fault diagnosis of pistonpin is an effective method.

关 键 词：双谱 柴油发动机 活塞销 神经网络 故障诊断 

分 类 号：TK428[动力工程及工程热物理—动力机械及工程]