Study Of Turnout Equipment Fault Diagnosis Based On Machine Learning

There are three main railway signal outdoor equipments: turnout switching machine equipment,track circuit and signal machine.The turnout switching machine equipment is the key equipment in the rail transit industry that directly affects the forward direction of the train.The circuit that controls the movement of the switch and indicates the position of the turnout is called the control circuit.In order to monitor the running status of the key equipment such as the switch machine in real time,the railway department uses the centralized signaling monitoring system to monitor and record the operation of the equipments in real time,technician analyzes and studies the status information of the equipment.At present,the management and monitoring capabilities of equipment are still not up to the level of state repair.However,with the application of technologies such as artificial intelligence,data fusion and big data,diagnostic techniques for device status will become more and more intelligent,and the improvement of diagnosis technology will also promote the maintenance mode of railway signal equipment to the state repair.At present,status diagnosis for switch machine mainly relies on the technician to assess the key parameter curve of centralized signaling monitoring system for empirical judgment.At the transportation hub,the technician needs to assess a large number of curves,and the accuracy and timeliness are affected.The level of technician's ability is different,causing the level of fault diagnosis to be uneven in the different place.In this paper,the switch machine of signal outdoor equipment is selected as the research object.The main goal is: Firstly,from the perspective of electricity,select several typical fault cases of the switch to analyze and realize the running state of the switch.The second is to select the appropriate algorithm for the fault diagnosis modeling of the switch,to realize the intelligent diagnosis of the switch,and to reduce the heavy workload of the reviewers.For the above objectives,the main contents to be completed in this paper are as follows:(1)Clarify the basic composition and working principle of the mechanical and electrical parts of the turnout system,master the characteristics of the normal operating current curve,the emergency handling process of the switch,and the typical fault phenomena and causes.(2)The failure mode and effect analysis method are used to analyze the failure mode and cause of the switch,and according to EN50126 standard documents analysis the risk rank.Rate the severity,frequency and detectability of typical failure case,thus the risk priority value is calculated,which provides a certain reference for the setting of the model parameters.(3)Briefly introduces several algorithms of machine learning.Based on the practical application scenarios of the research objects,the advantages and disadvantages of each algorithm,select the right classifier algorithm,and the neural network is selected as the model for the fault diagnosis of the switch.(4)Design the network structure of the error back propagation neural network and the baseline and segmentation techniques are proposed.The feature extraction is performed on each segment,and 30 time domain eigenvalues are extracted.The error back propagation neural network based on the baseline is modelled.Optimize the algorithm and verify that the model fault diagnosis accuracy rate is up to 94.6%.This paper mainly proposes the concept of baseline for the switch machine,and uses the BP neural network method to identify the state of the switch.The simulation results show that the method has certain feasibility and effectiveness,provides a reliable guarantee for switch machine operating,and contributes to the evolution of the state of the switch machine equipment maintenance mode.