Research On Optimal Design Method Of Curved Rail Profile For Heavy-haul Railway

With the rapid development of the world economy,the increasing demand for railway transportation has led to an increasing axle load of heavy-duty trucks,and the increase in axle load will accelerate the wear between wheels and rails,especially in curved section rails.Wheel-rail profile wear changes the wheel-rail contact geometry,affecting the safety and curve passing ability of the train during operation.Optimizing the wheel-rail profile can effectively improve the wheel-rail matching relationship,slow down the wheel-rail wear,and improve the service life of the rail.Therefore,this paper analyzes the wheel-rail wear of heavy-load curved lines,and establishes a multi-objective optimization method for the asymmetric curve rail profile of heavy-haul railway based on neural network and genetic algorithm.The optimal rail profile under actual line parameters was obtained by applying this method,and the adaptability of the optimal profile under different line parameters was studied.The main work content and conclusions of the paper are as follows:（1）Using the multibody dynamics software UM to establish a 25t axle load C80 truck vehicle-track coupling dynamic model,combined with KP multi-point contact theory and Archard wear calculation model,a rail wear prediction model was established,and the optimization goal of the minimum wear rate for the rail profile was determined.（2）Based on the circular arc parameters of the rail profile,the area with dense distribution of wheel-rail contact points is selected as the optimization area,and the simplified 5-dimensional rail profile independent variable is obtained.A large number of rail profile samples based on circular arc parameters were extracted by Latin hypercube sampling method,and a nonlinear mathematical model of rail profile and rail wear rate was obtained by combining neural network training,and the mean squared error MSE of the model reached 1×10^-8,and the predicted relative error goodness-of-fit was 0.973,with an error of less than 3%.（3）The genetic algorithm is used to solve the nonlinear mathematical model of rail profile and wear rate,and the wear rate of the optimal profile is 0.1852mm²·MT^-1,compared with the standard 75kg/m track,the optimization effect reaches 18.9%.The maximum grinding capacity is 0.9758 mm for the outer rail and 0.0892 mm for the inner rail.Then,the safety performance and curve passing ability of the vehicle under the silhouette are evaluated,and the results show that the safety meets the requirements and the curve passing ability is good.（4）Select the line parameters that have a great influence on the wear rate of the rail:rail bottom slope,ultra-high,friction coefficient,gauge,and determine the variation range of the parameters according to the relevant specifications.The difference in wear rate and dynamic performance of rail under single parameter variation was analyzed.The results show that the bottom slope,ultra-height and friction coefficient of the four parameters have great influence on the wear rate,and the wear rate increases accordingly.The wear rate and dynamic performance of the optimal profile are better than the standard 75kg/m rail profile under the change of the four types of line parameters.