Fault Analysis Of EMU Air Conditioning System Based On Fault Tree And Neural Network

With the development of Chinese train technology,EMUs are becoming more and more popular nationwide.Its comfort and safety are highly valued by tens of thousands of passengers.The reliability of the EMU air conditioning system plays a decisive role in passenger comfort.Once the air conditioning system fails,the high temperature may cause other systems of the EMU to malfunction and greatly affect passenger comfort and safety.Therefore,it is of great significance and practical application value for the fault analysis of EMU air conditioning system.In this paper,the subject of fault analysis of EMU air conditioning system based on fault tree and neural network is studied.The main work is as follows:(1)Focus on the analysis of the functions and structural components of the refrigeration system,ventilation system,heating system,control and pressure protection system in the EMU air conditioning system,the principle of the pressure protection system.(2)Figure out the historical fault data of the EMU air conditioning system from 2012 to 2017,and the fault data is classified according to the structure of the air conditioning system.Based on the data statistics and the structure of the air conditioning system,a fuzzy fault tree is established for the air conditioning system.The fault tree analysis method is used to qualitatively analyze and quantitatively analyze the air conditioning system.It can be calculated that the most vulnerable component in the system is the air conditioning control unit..(3)According to the fault tree analysis method,the air conditioning control unit is the structural unit with the highest probability of fault.Using temperature as an important metric for fault of the train system.BP neural network and LSTM neural network are used to establish the fault diagnosis model of EMU air conditioning system respectively.The evaluation criteria of air conditioning faults in EMUs are established to diagnose the faults of air conditioning system in 2017~2018.The experimental results verify the effectiveness of the fault diagnosis model for EMU air conditioning system established by LSTM neural network.(4)Design the overall framework of the EMU air conditioning fault analysis system.Implement LSTM-based analysis algorithms and deploy on the network.Access and display via online web apps and Android mobile apps.The client displays historical fault data and use line graphs to show the outdoor temperature,the passenger room temperature,the train speed.The fault diagnosis algorithm of the LSTM neural network EMU air conditioning system can be called through the network service interface to realize the fault diagnosis of the online air conditioning system.