摘要
为了有效提高道岔装置运行时的安全性与可靠性,减少人工失误,采用机器学习方法,对道岔运行故障诊断系统的设计进行初步研究。对系统应具有的功能进行分析,从通信链路状况和通话管理两个方面进行设计以保证数据的正确采集。采用LSTM(long-short term memory)模型,从原始的电流序列里自动提取特征,再根据特征利用神经网络分类器来对道岔动作电流曲线进行智能故障识别,并针对每一个诊断结果给出相应的处置意见和操作步骤,提高故障诊断系统的准确性。
In order to effectively improve the safety and reliability of the switch machine operation on rail transit and to reduce the manual error, the machine learning method was used to conduct a preliminary design of the switch machine operation fault diagnosis system. By analysis of the functions of system, the communication link status and communicate management were designed to ensure collecting data correctly. Meanwhile the system automatically extracted features from original current sequences by using LSTM model, and then the neural network classifier was used to identify the fault for switch machine. The corresponding disposal opinions and operation steps were given for each diagnosis resulting to improve the accuracy of the fault diagnosis system.
作者
唐维华
李德敏
Tang Weihua;Li Demin(College of Information Sciences and Technology, Donghua University, Shanghai 201620, China;Casco Signal Ltd., Shanghai 200040, China)
出处
《计算机应用与软件》
北大核心
2019年第9期37-40,共4页
Computer Applications and Software
关键词
道岔装置
数据采集
机器学习
故障诊断
应急处置
Switch machine
Data collection
Machine learning
Fault diagnosis
Emergency disposal
