基于PSO-SVM的车载ATC设备故障诊断

PSO-SVM-based fault diagnosis of vehicle ATC equipment

地铁列车车载自动控制系统(Automatic Train Control,ATC)设备的故障排查诊断大多依赖人工经验,存在效率低下问题.针对车载ATC设备的故障诊断问题,采用一种基于粒子群算法优化支持向量机(Particle Swarm Optimization-Support Vector Machine,PSO-SVM)的地铁列车车载ATC设备故障诊断方法 .根据历史故障数据记录表得到故障特征词汇集,引入粗糙集理论的知识对故障特征词汇集进行属性约简.利用PSO-SVM算法对约减后的故障特征词汇集进行分类对比,实验结果表明:在相同训练测试集下,PSO-SVM算法相对于SVM、神经网络算法具有更高的故障诊断准确率,并且更具稳定性.

Most of the troubleshooting and diagnosis of Automatic Train Control(ATC) equipment on subway trains rely on experience of staffs, which is inefficient upon opearation. Aiming at the fault diagnosis problem of onboard ATC equipment, a fault diagnosis method for ATC equipment on subway train based on Particle Swarm Optimization-Support Vector Machine(PSO-SVM) is adopted. The knowledge of rough set theory is adopted to reduce the attributes of the fault data feature set, the fault feature set is obtained according to the historical fault data record table. The PSO-SVM algorithm is used to classify and compare the historical fault data feature set. The experimental results show that under the same training test set, the PSO-SVM algorithm has higher fault diagnosis accuracy and is more stable than SVM and neural network algorithms.