WA与RBFN在气密性故障诊断中的应用研究

Fault Diagnosis about Cylinder Gas Tightness Based on WA and RBFN

提出了根据检测发动机起动时电瓶电压的波形来分析气密性的试验方法。在基于小波分析 (WA)的基础上对信号进行消除趋势项和除噪处理 ,并提出了包括周期延长和主频推移 2个新的判据在内的 7个特征参数。通过对径向基神经网络 (RBFN)的训练 ,证明该神经网络能够较好地进行故障模式辨识 ,从而为发动机气密性故障诊断提供了一个系统方案。

In this paper a new experiment method for analyzing cylinder gas tightness from the waveform of the storage battery voltage,which was collected when engine was driven,was discussed.The signal tendency and noise were canceled based on wavelet analysis.And 7 fault symptom parameters including two new parameters (period extension and main frequency delay) were put forward.The Radial Basis Function Network(RBFN)was trained by putting these symptom parameters in and this network could distinguish the fault modes preferably.Then a system scheme was formed to diagnose the cylinder compression ratio fault.