Study On Service Performance Of High Speed Railway Bogies Based On Machine Learning

The system of high-speed train bogie is an significant part of high-speed train,its performance largely determines the stability and safety of train operation.With the growth of high-speed train operating mileage,the performance and safety status of bogies are always in a state of change.This poses a great challenge to the safety and maintenance of bogies.Accurately analyzing the service performance can guide the speeding and speeding up of high-speed trains effectively,at the same time,it can optimize the maintenance cycle of the bogie,which is of great significance for the economy and safety of high-speed trains.In this paper,we propose a new method based on machine learning to study the service performance of high-speed rail bogie,and set up the website of bogie real-time monitoring system based on Django framework,and this method is verified to be practical.There are mainly 3 research results as follows:(1)In the basis of the modern signal processing theory,dynamics theory,operation data of high-train and machine learning regression algorithm,a research method of service performance of high speed train bogies is proposed.The research method can be divided into three processes.Firstly,the bogie dataset,which collected during operation,is transformed by wavelet method and then normalized.Secondly,machine learning algorithm including linear regression,backpropagation neural network,radial basis function neural network,decision tree regression and random forests were used to predict the vibration data of bogie,compare the regression performance of each algorithm.The experimental results show that the random forest regression model is the best model in terms of the vibration beam data of Bogie.(2)Through the verification of experimental,the experimental model that proposed can reflect service performance of bogie components effectively.There are mainly two applications based on experimental model.On the one hand,the model is used to monitor the current status of bogie components in real time.Under the premice of a certain mileage,when the train speed or acceleration exceeds the limit range,the monitored vibration value will above the safety threshold,if take measures timely to this dangerous circumstance,that can reduce unnecessary damage of property.On the other hand,the model can predict abnormal trends of the bogie components as well.When the using time of bogie components has reach its limit life,the vibration value of the train bogie will above the safety threshold,so the bogie of the train needs to be replaced promptly.Therefore,the model can make a more reasonable maintenance cycle to the bogie,which can reduce the costs of train operating.(3)In order to make the results of this research method more practical,we developed the real-time monitoring system website of bogie based on Django framework,in order to visualize all of the experimental results.This website can screen the vibration data of the bogie components according to the requirements,and the different bogie components will meets the different prediction model of machine learning algorithm.