基于灰度图像纹理分析的柴油机失火故障特征提取

Fault feature extraction for diesel engine misfires based on the gray image texture analysis

柴油机失火是其常见的故障模式,传统的诊断方法不仅参数获取困难且原始信号易受噪声污染导致准确性较差。针对此问题,提出了一种基于灰度图像纹理分析的二维故障特征提取模型,可以有效地降低噪声污染,简化计算过程。将时域振动信号转化为灰度图,通过局部二值模式对灰度图进行局部纹理分析,提取其局部特征,并通过二维傅里叶变换识别灰度图的特征频率,达到降噪及识别特征频率的目的。以三缸四冲程柴油机为研究对象,设计了柴油机失火故障的预置试验,采集排气噪声和缸盖振动信号对提出的方法进行验证。结果表明,该方法能有效降低信号噪声,识别柴油机的故障特征。

Misfire is a common fault in engines,and the traditional fault diagnosis approaches are of the problems of noise pollution and difficulty in parameters obtaining,which make a bad accuracy.Aiming at these problems,a 2D fault feature extraction model based on the gray image texture analysis was proposed,which can be used to reduce noise pollution and simplify the calculation process.A 1D time-domain signal was converted into a 2D gray image,then the texture analysis on the gray image was introduced based on the Local Binary Patterns(LBP)to find out local features.The fault features of the gray image were then extracted based on the 2D FFT with the purpose of reducing the noise pollution.Taking a three-cylinder four-stroke diesel engine as an object,a predesigned misfire fault experiment for the diesel engine was carried out to collect the exhaust noise and cylinder head vibration signals for fault diagnosis,which demonstrates the accuracy of the proposed method.