Research On The Optimization Of Freight Car Axle Temperature Prediction In The Operating Environment Of Heavy Load And Long Ramp

In recent years,with the continuous development of the railway freight market,the number of railway freight cars nationwide has steadily increased,and the safety problems that have arisen during the operation of vehicles have also increased.Bearings are the key equipment for carrying weight on trains,and their safety and reliability directly affect the safe and stable operation of freight cars.Axle temperature is an effective indicator that reflects the health of freight train axles.In order to monitor the axle temperature of freight trains,the railway bureau currently mainly relies on the Trace Hotbox Detection System(THDS)to monitor train bearings in real time.However,in the actual operation of railway freight cars,misreporting of hot axles often occurs,resulting in a large waste of human,material,and financial resources.In order to solve this problem,this paper will optimize the existing methods of predicting the axle temperature of trucks to reduce the false alarms in the process of predicting the axle temperature of trucks.The dissertation first uses the powerful statistical analysis function of the SPSS software,and selects the correlation analysis method in the software to carry out correlation analysis and research on all the factors that may affect the axle temperature change obtained from the collection and analysis,and finds the position,speed,and time of the truck,Model,load,environmental temperature,line characteristics and other factors that have greater correlation with axle temperature,pave the way for the application of the adaptive adjustment method of axle temperature abnormality judgment threshold based on deep Q learning algorithm.Taking the special operating environment and line characteristic conditions of the Wari Line as an example,this article introduces the line overview of the Wari Line with ultra-long downslope and the three continuous THDS axes of the Puxian Downstream Station,the Longma Downstream Station,and the Xinbao Downstream Station.Condition characteristics of the temperature monitoring site.Collected and sorted the axle temperature monitoring data information of the above three stations.Through the analysis of the data information,it was found that there were many false alarms in the axle temperature monitoring of the Longma downlink station.And the false alarm rate of C80 trains is higher than that of other trains.Therefore,the axle temperature monitoring and collection of the Longma downlink station was selected.The data is used as an actual calculation example of the first line of the road bureau.Finally,the theoretical knowledge of the deep Q-learning algorithm and the specific methods applied to the prediction of truck axle temperature are introduced in detail.In order to verify the applicability and effectiveness of the axle temperature prediction method proposed in this dissertation for actual working conditions,taking the Longma downstream THDS axle temperature monitoring station in the Wari line as an example,the axle temperature prediction method based on the deep Q learning algorithm is applied.The calculation results of the axle temperature forecast on the Wari line have been reduced to two events,of which two strong thermal events that should have been normal conditions were only reduced to slightly thermal conditions.Therefore,the optimization method of freight train axle temperature forecast proposed in this dissertation can better adapt to the actual working conditions of railway bureaus,significantly reduce the false alarm rate of freight car axle temperature forecast,and effectively promote the intelligent development of railway maintenance and construction.