小波神经网络法在柴油机故障诊断中的应用

Diesel Engine Fault Diagnosis Based on Wavelets and Neural Networks

用小波分析作信号处理手段提取柴油机振声信号特征量 ,以神经网络作为故障模式识别手段 ,进行了柴油机故障的振声诊断方法研究。针对柴油机振声信号的非平稳时变特性 ,应用小波理论中的小波包方法对其进行处理 ,结果表明小波分析是比傅里叶分析更为有效的处理柴油机振声这类非平稳信号的方法。在此基础上 ,研究了用神经网络实现根据小波包分解结果识别柴油机故障状态的方法。

It is of important significance to study fault diagnosis methods of diesel engine.As a kind of reciprocating machine,diesel engines fault diagnosis is generally recognized difficult problem,and the studying level is far lower than that of rotary machine.Vibration fault diagnosis of diesel engine is studied using wavelet as the signal processing method to extract the characteristic and neural networks as the mode recognition method.Because the diesel engines vibration signal is unstable and timely varying,wavelet packet are used to process the engines vibration signal.Results indicate that wavelet analysis is more effect than Fourier analysis when processing unstable signal such as the diesel engines.Then,apply neural networks to recognize the diesel engines faults according to the results of wavelet was studied.