Rail Wear Prediction And Profile Multi-Objective Optimization Design Of High-Speed Railway Based On Swarm Intelligence Optimization Algorithm

With the rapid growth of high-speed railway mileage and the increase of train operation density in China,the problem of wheel-rail wear becomes more and more serious,rail wear has become one of the reasons for the serious damage of track foundation,the wheel-rail will contact directly when the train runs,the state of rail profile will directly affect the running state of the train.Good wheel-rail contact can reduce wheel-rail wear,prolong rail service life and reduce maintenance costs.However,severe rail wear will affect the comfort of the train,operation stability and safety have also declined.Through the prediction of rail wear,the change trend of rail wear can be grasped to arrange the rail grinding cycle,the optimization of the rail profile can change the wheel-rail contact state to improve vehicle performance and reduce wheel-rail wear.In this thesis,the research status of wheel-rail wear prediction and profile optimization is analyzed and summarized.In view of the current research status,the high-speed train-track coupling dynamic model is established by UM software,and the optimization model and prediction model are programmed by MATLAB software.The research and analysis are carried out around the theory of vehicle-track coupling dynamics,the improvement of intelligent optimization algorithm,the influencing factors and changing rules of rail wear,the prediction of rail wear and the optimization of rail profile.The main research contents and results are as follows:(1)Based on the concepts of particle swarm optimization and sine cosine algorithm,the global optimal fitness value is calculated by combining the two optimization algorithms,and the value of the final fitness function is selected and determined according to the greedy strategy.An adaptive exponential decreasing inertia weight update mechanism is designed to change the original inertia weight value.In order to prevent the lack of population diversity in the later stage,the algorithm falls into local optimum.The particle swarm optimization and sine cosine algorithm are disturbed by Levy flight and Cauchy mutation.Seven test functions and four other comparison algorithms are used to verify the proposed HPSOSCA algorithm.It is found that the accuracy and speed of optimization are better than the other four.(2)Based on the NSGA-II algorithm,an improved NSGA-II algorithm(ChaosNormal-Non-dominated Sorting Genetic Algorithm-II,CN-NSGA-II)is proposed.Secondly,the adaptive selection crossover operator is used to improve the optimization ability and convergence speed of the algorithm.The crossover operator is the SBX operator of NSGA-II and the introduced NDX crossover operator.In the iterative process of the algorithm,the trade-off coefficient is used to determine the selection probability of different crossover operators.An update mechanism of adaptive scaling factor decreasing method based on exponential function is designed.Finally,in the mutation stage,the traditional polynomial mutation method is changed to Cauchy mutation method,and the performance of the algorithm proposed in this paper is verified.Seven test functions and four comparison algorithms are used to calculate three performance indicators.The performance indicators of the CN-NSGA-II algorithm under the three objective function are better than other comparison algorithms,and most of the performance indicators under the two objective function are better than other comparison algorithms.(3)The influencing factors of rail wear of high-speed railway are analyzed.The simulation experiments are carried out under different influencing factors(through total weight,curve radius and running speed).It is found that the wear amount increases with the increase of total weight,and the wear amount decreases with the increase of curve radius and running speed.The wear distribution is mainly concentrated in the top of the rail.The wear of the gauge angle is very small.The analysis of the influencing factors of rail wear is to determine the input in the wear prediction model.The HPSOSCA-MKLSSVM prediction model is proposed by optimizing the relevant parameters in the Multi-Kernel Least Squares Support Vector Machine(MKLSSVM)through the HPSOSCA algorithm.The total weight,curve radius and running speed are taken as the input of the prediction model,and the total wear amount is taken as the output.HPSOSCA-MKLSSVM is trained and predicted by HPSOSCA-MKLSSVM.In order to verify the prediction error of the new model and the prediction error of the wear data of the other four prediction models,it is found that the calculation value of the HPSOSCA-MKLSSVM model proposed in this paper under the four evaluation indexes is smaller than that of other prediction models,indicating that the model has higher prediction accuracy.(4)The optimization model of high-speed railway rail profile is established.Firstly,the rail profile parameterized by cubic NURBS curve is used as the optimization variable to solve the coordinates of discrete points on the rail profile.Then,the number of model points and specific coordinate points are determined,and then the coordinates of control points are solved and the weight factors are valued.After determining the above conditions,the cubic NURBS curve can be used to fit the profile.Next,determine the area to be optimized and the objective function to be optimized and formulate relevant constraints.The vehicle-track coupling dynamics model is used to calculate the objective function in the optimization model,and the CN-NSGA-II algorithm is used for iterative optimization.Finally,the multi-objective optimization of the rail profile is realized by MATLAB and UM joint simulation,and the Pareto optimal solution set is obtained.In order to avoid the randomness and subjectivity of manual selection,the optimal solution is selected by TOPSIS algorithm to obtain the optimal target profile.The maximum wheel-rail lateral force of the left and right wheels of the guide wheelset is reduced by 13.3 % and 13.71 % respectively,the maximum contact stress is reduced by 10.35 % and 3.21 %,and the maximum wear number is reduced by 36.09 % and 11.4 %.The safety and stability indexes of the vehicle are verified,and the calculation results are within the limit range of the relevant indexes.The adaptability of the optimized rail profile under different running speeds is analyzed,and the contact stress,wheel-rail lateral force,derailment coefficient and stability index of the guide wheelset are calculated.The results show that the calculated values under the corresponding calculation indexes of the optimized rail profile are lower than those of the original rail profile.