APP下载

基于机器学习的铁路道岔故障识别

2021-07-15牛太冬

河南科技 2021年6期

牛太冬

摘 要:道岔的正常运转是保证列车正常运行的必备条件,传统的道岔故障检测方法主要来源于人的工作经验,根据电流的非正常变化来判别道岔是否发生故障,消耗较多的人力资源与物力资源。为了提升资源的有效利用率,本文运用概率主成分分析法提取数据的主要特征,分别采用支持向量机模型和[k]近邻模型作为道岔故障分类器,然后使用十折交叉验证法作为模型的评价标准,以达到智能识别铁路道岔故障的目的。

关键词:概率主成分分析;支持向量机;故障识别;[k]近邻法

中图分类号:U284.92 文献标识码:A 文章编号:1003-5168(2021)06-0033-03

Railway Turnout Fault Recognition Based on Machine Learning

NIU Taidong

(Tianjin University of Science & Technology,Tianjin 300457)

Abstract: The normal operation of the switch is a necessary condition to ensure the normal operation of the train, traditional turnout fault detection methods are mainly derived from human work experience, it judges whether the turnout is malfunctioning according to the abnormal change of the current, which consumes more human resources and material resources. In order to improve the effective utilization of resources, this paper used the probabilistic principal component analysis method to extract the main characteristics of the data, respectively used the support vector machine model and the [k]-nearest neighbor model as the turnout fault classifier, and then used the ten-fold cross validation method as the evaluation standard of the model to achieve the purpose of intelligently identifying the railway turnout fault.

Keywords: probabilistic principal component analysis;support vector machine; fault identification;[k]-nearest neighbor method

目前,大部分地區通过微机监控系统采集道岔开闭时的电流值来判断铁路道岔是否发生故障。转辙机正常动作时的电流曲线如图1所示,发生故障时的转辙机动作电流曲线如图2至图6所示。由图像可以看出,除了故障时转辙机动作电流与正常时转辙机动作电流不同外,不同情形下的故障电流也不相同。

随着人工智能行业的发展和完善,人们可以使用机器学习算法进行铁路道岔故障识别,减少人力和物力的浪费,提高铁路道岔故障识别的准确性,减少故障识别的时间成本。唐维华[1]等利用LSTM(Long-Short Term Memory)电流数据的特征,将神经网络算法应用到道岔动作电流曲线分类器中。程宇佳[2]以核方法为基础,研究高速铁路道岔故障诊断方法。可婷等[3]利用主成分法提取道岔工作电流特征的主成分,并利用查准率和查全率构造道岔识别性能指标。……

登录APP查看全文