张红;赵龙庆;唐送;

1:西南林业大学机械与交通学院

2:江南工业集团有限公司民品技术研究所

    摘要：本文在阐述了发动机气门漏气声学特性及其振动诊断机理的基础上,针对发动机缸盖振动信号的特点,运用小波包对采集的振动信号进行3层分解、重构、提取特征向量。然后将特征向量作为概率神经网络的输入,构建网络模型。再用测试数据验证诊断模型的正确性。诊断结果表明该方法是可行的,并取得了较好的效果。

    关键词： 气门漏气;;小波包分析;;概率神经网络;;故障诊断

    Abstract: Basing on acoustic characteristics,the vibration mechanism of diagnosis and characteristics of cylinder head vibration signal of valve leakage,the paper uses three- layer wavelet packet to decompose the vibration signal,and constructs the wavelet packet energy eigenvector. Then, the eigenvector is input into PNN to diagnose the fault type. The results show that this method is feasible and practical.

    Keywords:Valve Leakage;;Wavelet Packet Analysis;;PNN;;Fault Diagnosis

参考文献：

[1]肖云魁.汽车诊断学.北京理工大学出版社,2001.

[2]曹龙汉.柴油机智能化故障诊断技术.国防工业出版社,2005.

[3]Takeshi K,Keiichi N,Koji M.Diagnosis of gas leakage from exhaust valve in marine diesel engine by accustic sound.Bulletin of the MESJ,1989,17(1):4-11.

[4]黄小念,夏勇,张振仁.发动机气门漏气诊断的关联积分法.小型内燃机与摩托车.2002,31(2).

[5]王裕鹏,赵龙庆,张华伟.基于小波改进阈值消噪的缸盖振动信号分析[J].柴油机,2007,29(5):37-39.