基于尺度—小波能量谱和神经网络的内燃机故障诊断

为了对内燃机气门及活塞—连杆组故障进行有效地诊断,通过试验测取内燃机在不同故障下的振动信号,利用连续小波变换得到信号在不同尺度上的能量分布,即信号的尺度—小波能量谱。其能量主要分布于尺度范围1~32,且相同故障模式下的尺度—小波能量谱呈现出相似性,不同故障模式之间的尺度—小波能量谱存在很大的差异性,以此作为不同故障模式的信号特征,结合BP神经网络方法,实现了对待检信号的正确识别。

Fault Diagnosis of Internal Combustion Engine Based on Scale-wavelet Power Spectrum and Neural Network

In order to diagnose effectively the failures or faults of the valve and the piston connecting rod of ICE,the vibration signals under different fault models were measured by experiments.Through the continuous wavelet transform(CWT),the power distribution of signals in different scales,which was also called the scale-wavelet power spectrum(SWPS) of signals,was acquired.The wavelet power(WP) distribution was mainly in the scale of 1～32, the scale-wavelet power spectrums among same fault models were similar,and the scale-wavelet power spectrums among different fault models were diverse.Finally the detecting signals were discerned correctly according to the signal characteristic of different fault models under the assistance of back propagation neural network(BPNN).