基于CART决策树的柴油机故障诊断方法研究

Fault diagnosis of diesel engines based on a classification and regression tree (CART) decision tree

采用一种自适应局部有效值(RMS)计算方法提取柴油机缸盖振动信号时域特征,结合分类回归树(CART)算法建立故障分类模型并进行柴油机的状态识别。通过实验获取柴油机失火和撞缸两种故障工况及正常工况下的振动数据,计算出原始信号的局部RMS后,根据自适应阈值确定点火冲击区域和非点火上止点冲击区域提取局部特征,最后将特征输入CART算法中构建分类模型来验证所提取特征的有效性。结果表明:柴油机在3种状态下的识别率均达到100%,基于CART算法和局部特征提取的方法能够有效诊断柴油机故障。

An adaptive local root mean square（ RMS） calculation method has been used to extract the time-domain characteristics of a diesel engine cylinder head vibration signal. The classification model of the diesel engine was established based on the classification and regression tree（ CART） algorithm. The vibration data for diesel engine misfiring,crashing cylinders and the normal operating conditions were obtained through experiments. After calculating the local RMS of the original signal,the local characteristics were extracted based on the self-adaptive thresholds to determine the ignition impact area and non-ignition top dead center impact area. Finally,these features were input into the CART algorithm to build a classification model in order to verify the validity of the extracted features.The results show that the recognition rate of the diesel engine reaches 100% in all three states. Thus the CART algorithm and local feature extraction method can effectively diagnose diesel engine faults.