小波分解及图像处理在内燃机振动诊断中的应用研究

APPLICATION RESEARCH OF WAVELET DECOMPOSITION AND IMAGE PROCESSING FOR FAULT DIAGNOSIS IN INTERNAL COMBUSTION ENGINES

内燃机缸盖振动具有明显的非平稳时变特点 ,因此在监测诊断时应尽量利用信号中的信息。为了克服时域和频域特征提取方法的不足 ,从而充分利用振动信号所包含的信息 ,本文提出了一种新的内燃机故障诊断方法 :对缸盖振动信号进行小波包分解 ,得到信号的时 -频分布图 ,并利用软件方法定义图象 ,然后提取图象的各种特征参数实现诊断。将该方法用于 613 5柴油机气阀机构的故障诊断 ,取得了很好的效果。结果表明此方法简单有效 ,诊断精度较高 ,且对信号采样的要求不高 ,易于实用。

As the cylinder head vibration signal is time-varying and nonstationary, the abundant information in it should be utilized as fully as possible when using it to carry out diagnosis and condition monitoring for internal combustion engines. In order to overcome the shortcomings of the commonly used time and frequency feature-extracting methods and take full consideration of the information in signals, a diagnosis method is presented. Decomposing the vibration signals with wavelet packet algorithm provides the time-frequency distribution images, and the images are then defined by software and various image features are extracted to be applied in diagnosis. The method produces good results when diagnosing different faults of the valve train in four-stroke six-cylinder diesel engine. It is shown that low requirement of signal sampling is needed and relatively high diagnosing accuracy and feasibility can be achieved.