基于小波包和概率神经网络的气门漏气故障诊断研究

Study on Diagnosis of Valve Leakage Fault Based on Wavelet and PNN

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。

Basing on acoustic characteristics, the vibration mechanism of diagnosis and charac- teristics of cylinder head vibration signal of valve leakage, the paper uses three - layer wavelet pack- et 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 fea- sible and practical.