基于群决策的道岔控制电路故障诊断方法

Fault Diagnosis for Railway Turnout Control Circuit Based on Group Decision Making

高速铁路道岔是与高速列车直接接触的重要信号设备,其控制电路的故障检测手段目前仍停留在简单仪器与人的经验相结合的方式.为了实现道岔控制电路故障的智能诊断,提高故障诊断的准确率并降低单一诊断方法带来的不确定性,本文提出一种基于群决策的诊断方法:首先根据道岔控制电路的特点,总结了典型的11个故障模式和对应的8个故障特征;其次,分别采用模糊理论、神经网络和支持向量机(Support vector machine,SVM)对道岔控制电路进行故障诊断;然后引入群决策理论将三种方法视为决策专家,通过群基数效应集结方式实现决策级上的信息融合从而得到群专家综合评判的诊断结果.从仿真数据的验证来看,该方法比单一方法的故障诊断的准确率要高,表明了本文所提方法能够实现三种方法的互补融合,也提高了故障诊断的准确率,在该领域有着良好的应用前景.

High speed railway turnout is an important signal device that directly touches the high speed train. However,it still depends on simple instruments and human experience to deal with the faults of the control circuit. In order to realize intelligent fault diagnosis for the turnout control circuit, improve diagnosis accuracy and decrease uncertainty that a single method may bring about, a fault diagnosis method based on group decision making strategy is proposed. Firstly11 typical fault modes and 8 corresponding fault features are summarized according to the characteristics of the control circuit. Secondly, fuzzy theory, neural network and support vector machine（SVM） are adopted to conduct the diagnosis process, respectively, then group decision making strategy is introduced, which regards the above three methods as three different experts. Ultimately, the final comprehensive diagnosis result is achieved by utilizing the group cardinal utility method on the three experts. Simulation result shows that compared with all three individual methods, the proposed method achieves a better performance on the diagnosis accuracy, indicating that the proposed method can integrate the advantages of the three methods and have a great application prospect in the field.