Research On Wheel Size Prediction And Re-profiling Strategy For The Wheel-set Of High-speed EMU

With the vigorous development of high-speed EMUs all over the world,the safety of train operation has been attracted increasing attention of the society.A good wheel profile not only ensures the correct position of the train on the rail to ensure the safe operation,but also reduces the wheel-rail wear and prolongs the service life of the wheel,thus reducing the manufacturing and maintenance costs.Therefore,accurately predicting the size of the EMU wheel-set,and optimizing the wheel-set re-profiling strategy have important practical significance for train maintenance and railway safety.The main research object of this thesis is the train wheel-set,of which the analysis and research on the field measurement data is carried out.Due to the complexity of the wheel size variation affected by the operating environment and other factors,a prediction model based on Multi-Kernel Extreme Learning Machine optimized by Particle Swarm Optimization(PSO-MKELM)for wheel-set size data is proposed.A wheel-set wear model and a re-profiling strategy research model are established,and a reasonable re-profiling strategy is proposed based on the simulation results.The specific work includes the following aspectsFirstly,this thesis analyzes the variation law of the EMU wheel size data and applies the Kernel Extreme Learning Machine(KELM)algorithm to the prediction of the wheel size.The Multiple Kernel(MK)function,which is composed of the polynomial kernel function and the radial basis kernel function,is integrated into Extreme Learning Machine(ELM),and the Particle Swarm Optimization(PSO)algorithm is used to optimize the four key parameters of the model.For the wheel diameter and flange thickness data of the CRH2 high-speed EMU,the rationality and accuracy of the proposed method are verified by comparing the prediction results of different algorithms.Compared with the BP model,the ELM model,and three commonly used KELM model,the prediction results show that PSO-MKELM model has higher determination coefficient R2,lower Mean Squared Error,Mean Absolute Error and Mean Absolute Percentage Error,which illustrates the effectiveness of the PSO-MKELM modelBesides,based on the field measurement data analysis of CRH2A EMU,the wheel-set wear model was established,which estimated that when the flange thickness is about 30.3mm,the flange thickness wear rate is lower.Taking the longest service life of the wheel and the minimum times of re-profiling as indicators,the limited values of wheel diameter and the flange wear are the constraint conditions,research on the single wheel optimization re-profiling strategy.The optimal re-profiling strategy is concluded through the simulation calculation.Finally,the re-profiling amount of whole vehicle wheel-sets are analyzed under the condition of ensuring the limited values of wheel diameter difference.Related models and algorithms can be widely applied to the formulation of the EMU wheel-set re-profiling strategy.