基于支持向量机的接触网故障识别方法研究

Research on Identification Methodof Catenary Fault Based on SVM

接触网作为牵引供电系统中关键的子系统,一旦发生故障将直接影响行车安全,因此接触网故障的智能识别就尤为重要。基于铁路接触网结构模型及动力学特性,建立了垂向弓网耦合模型,对不同接触网故障情况下的受电弓振动响应进行了动力学仿真及分析,并提出了一种基于支持向量机和受电弓振动响应的接触网故障识别方法,为了提高故障的识别率及分类准确率,在支持向量机参数选择上使用了粒子群算法。将所提出的新方法应用于接触网故障识别仿真中,分析受电弓动力学响应中相应的故障特征,故障识别准确率达到83. 3%且没有漏检,表明了上述方法的有效性及实用性。

Catenary works as a key subsystem in the traction power supply system and any failure in it will directly affect the driving safety. So the intelligent identification for catenary fault becomes particularly important. Based on railway catenary structure model and dynamic characteristics,a vertical coupling model between pantograph and catenary was established,then the dynamics simulation and analysis of pantograph vibration response were carried out under different catenary fault condition. Further,a catenary fault identification method was put forward based on support vector machine（ SVM） and vibration response of pantograph. In order to improve the fault identification and classification accuracy,the particle swarm algorithm was used in the parameter selection of SVM. The proposed new method was applied to the simulation of catenary fault recognition,which can analyse the corresponding fault characteristics of pantograph dynamic response. The fault identification accuracy can reach 83. 3% and there is no leak,which shows that the method is effective and practical.