Research On Wheel Wear Predicting And Re-profiling Strategy Based On Intelligence Analysis

With the vigorous development of Chinese railway transportation,the safety of train operation has become an increasing concern to the society.Train wheels,which are the key components of the train,are not only important to train safety,but also has been a hot spot for scholars at home and abroad.Taking the wheelset as the main object,this thesis builds the LS-SVM prediction model optimized by PSO-GA-LM,establishes wheel wear model and single-wheel multi-objective optimization model,improves the classical NSGA-II algorithm,and designs vehicle wheel re-profiling strategy.Specific work includes the following aspects:Firstly,in order to predict the wheel wear volume,the PSO-GA-LM optimization algorithm is proposed to solve the problem that hyper-parametric optimization process of the LS-SVM model being easily trapped into the local optimum.Related experimental results show that the PSO-GA-LM can effectively prevent the LS-SVM from falling into the local optimum,the prediction accuracy is improved by 9.92% and 12.73% compared with the PSO-GA and PSO,and 42.19% compared with that of BP neural network.The LS-SVM model optimized by PSO-GA-LM can effectively predict the state of wheels.Secondly,based on the analysis of historical data in locomotive SS4-0997 from Taiyuan North Locomotive Depot of Taiyuan Railway Bureau,the wheel wear model is established.With the model,considering single wheel service life and re-profiling times,a single-wheel multi-objective optimization model is built.At the same time,aiming at the characteristics of wheel re-profile,the improved NSGAII algorithm is proposed to improve the global search ability and convergence of classical NSGA-II.The experimental results show that the proposed algorithm can find a re-profiling strategy with five reprofiling times and a service life of 11.22 years.Relevant models and algorithms can be widely used in single wheel and the ideal case of the whole wheel sets re-profiling strategy development.Thirdly,considering the wheel wear characteristics and the wheel diameter difference between all wheelsets on one locomotive,a model which aims at minimizing the cut off volume of all wheels during the re-profiling process is established,At the same time,preliminary vehicle wheel re-profiling strategy is put forward.Based on the on-site re-profiling results of SS4-0997 locomotive,the re-profiling strategy proposed in this thesis reduces 48.15% wheel diameter re-profiling volume compared with the currently adopted method,and can keep the locomotive in the best status.Based on the data of wheel-sets,this thesis points to practical applications and gives effective suggestions for the prediction,maintenance and repair of train wheelsets in China,which can ensure the safety of train operation to a certain extent,and has great practical significance.