基于多传感器信息融合的列车转向架机械故障诊断方法

MECHANICAL FAULT DIAGNOSIS METHOD OF TRAIN BOGIE BASED ON MULTI-SENSOR INFORMATION FUSION

针对单一传感器所含信息不能完全表达故障状态的局限性,提出一种支持向量机分类器和DS证据理论相结合的多传感器信息融合方法。将支持向量机的硬输出通过Platt模型转化为概率输出,用混淆矩阵来评估分类器的识别能力;将分类器局部可信度作为DS融合时的折扣因子,建立基于支持向量机和DS结合的多传感器信息融合模型。在列车转向架故障诊断中的实验结果表明,该方法在实际问题中有效且合理,能够获得比单一传感器更高的分类准确率,且对不同速度下列车转向架故障的识别结果都较好。

Aiming at the limitation that the information contained in a single sensor cannot fully express the fault state,we propose a multi-sensor information fusion method combining support vector machine classifier and DS evidence theory.The Platt model converted the hard output of support vector machine into probability output,and the confusion matrix was used to evaluate the recognition ability of the classifier;the local credibility of the classifier was taken as the discount factor of DS fusion,and a multi-sensor information fusion model based on support vector base and DS combination was established.The experimental results of the train bogie fault diagnosis show that our method is effective and reasonable in practical problems.It can obtain a higher classification accuracy than a single sensor,and the identification results of train bogie fault at different speeds are better.