基于RBF核支持向量机的轴箱磨损状态诊断系统

Fault Diagnosis System of Axle Box Wear Condition Based on Support Vector Machines with RBF Kernel

磨粒分析与识别是机务段对列车轴箱润滑状态检测与故障诊断的一个重要领域。文中提出了一种基于磨粒特征识别的轴箱润滑状态诊断技术,利用Matlab软件分别提取了8类典型磨损颗粒的特征参数,采用支持向量机原理设计了一个磨损颗粒分类器。将提取的磨粒特征参数进行归一化处理后作为分类器的输入进行模式诊断,并将磨损状态的编码作为目标输出。结果表明,该方法诊断轴箱磨损状态的准确率较高,具有较高的推广能力,可用于列车轴箱磨损状态诊断系统中。

Wear particle analysis and recognition is key to train axle box lubricating state detection and fault diagnosis in locomotive depot. The diagnosis technology of axle box lubrication based on the recognition of wear debris is put forward. Characteristic parameters for 8 kinds of typical wear particles are extracted by using Matlab, and a wear particle classifier is designed by the support vector machine principle. The extracted parameters of abrasive are normalized as input of classifier model diagnosis, and the wear condition is encoded as the target output. The results show that the accuracy of the method in the diagnosis of axle box wear state is high, and the method has high generalization ability, applicable in the diagnosis system of axle box wear condition.