Research On Prediction Method Of Bearing Temperature Rise Trend Of High Speed Train

With the rapid development of high-speed railway and railway informatization,as well as the high cost performance of high-speed trains in China,the "China high speed railway" and "China Technology" have come out of China and into the world.High speed railway brings great convenience to people's travel,at the same time,it makes people pay more attention to the operation safety of high-speed trains.Bearings are the key components of the high-speed train transmission system,and working conditions of the bearing directly affect the safety operation of the trains.At present,high-speed trains use on-board axle temperature monitoring systems to monitor the real-time changes in axle temperature.The system alarm based on pre-set thresholds.Then the staff take corresponding remedial measures.However,as the maintenance mode of high-speed trains gradually changed from "planned repair" to "state repair" and even "preventive repair",according to the current running state of high-speed train parts,evaluating failures and preventing them in advance have gradually become the main direction of research.Therefore,based on the temperature-related data of high-speed train axlebox bearings,large gearbox bearings and small gearbox bearings,this paper studies the prediction method of temperature rise trend of high-speed train bearing,in order to effectively guarantee the safety of train operation and reduce the maintenance pressure.The main research contents of this paper are as follows:In the first place,by analyzing the axle temperature monitoring data of high-speed trains,and then solving the problems of the huge amount of original data and some missing parts,the data preprocessing is realized.And analyze the characteristics of bearing temperature changes of the same high-speed train running on the same line.According to the time series characteristics of high-speed train bearing temperature data,the ARIMA temperature prediction model is established by using the historical axle temperature data of high-speed train through Eviews software.Then carry out one-step and multi-step predictive analysis on bearing temperature data of high-speed trains.In the second place,according to the characteristics of the high-speed train axle temperature affected by many sensitive factors,the correlation analysis is used to complete the extraction of axle temperature related data features and establish a sample set.The prediction model of high-speed train axle temperature based on support vector machine is established by MATLAB software.Then the PSO algorithm is used to optimize the kernel function and penalty factor parameters,and the optimization model of the bearing temperature prediction of the high-speed train is established.Finally,use the actual service data of high-speed trains to verify the accuracy of the model.The results show that the optimized high-speed train bearing temperature prediction model can better reflect the relationship between the axle temperature-related characteristic parameters and the axle temperature change,and the model's prediction effect and generalization ability are better.In the end,aiming at the shortcomings such as insufficient use of information in a single model and the decrease in prediction accuracy during multi-step prediction,this paper uses the weighted mixing method and Kalman fusion algorithm to fuse and analyze the ARIMA model and PSO-SVM model.The results show that the kalman filter fusion algorithm reduces the error accumulation of single model and further improves the overall prediction effect of bearing temperature sequence of high-speed train.The fusion algorithm can accurately predict the future temperature change trend by using the historical bearing temperature data of high-speed trains.