Multi-Objective Optimization Of High-Speed Vehicle Suspension System Parameters Based On HAM

With the rapid development of high-speed railways and high-speed trains,high-speed train dynamics performance such as comfort and operation safety has become the focus of researchers’attention.The dynamic performance is closely related to the parameters of the suspension system.Multiple target optimization of the parameters of the suspension system can effectively improve high-speed train dynamic performance.Many scholars widely use the proxy model technology to replace the simulation model for the optimization and solution of dynamic properties,but the shortcomings are that most of the scholars adopt an offline update single agent model form.For the research on offline or online update,few.This article establishes a classic hybrid proxy model and online update mixed self-adaptive proxy model（HAM）on the offline update of high-speed train suspension system parameters and dynamic performance indicators."（GB/T 5599-2019）The medium dynamic indicator threshold is used as a constraint condition to build a multi-target optimization model,and the multi-target optimization problem of the parameters of the suspension system.The specific research work includes:（1）Establish a high-speed train dynamic simulation model.Based on the SIMPACK dynamic simulation software,a high-speed train dynamic simulation model is established,and the nominal calculation of the model is calculated to verify the rationality of the model.Add rail inspiration and different lines of the dynamic simulation model to calculate related dynamic performance indicators.（2）Select experimental design methods and identify key parameters.Taking plus or minus 50%of the initial design value of the suspension system parameters as the design space,the optimal Latin hypercube experimental design method was selected to extract the sample points and calculate the kinetic response.The correlation between each parameter and the dynamic performance index was analyzed,and the five key suspension system parameters with the highest correlation with dynamic performance were selected according to the analytic hierarchy method and fuzzy comprehensive evaluation method:the suspension stiffness of the first series spring K₂,the transverse/longitudinal stiffness of the air spring K₆,the vertical stiffness of the air spring K₇,the equivalent damping coefficient C₂of the second series transverse shock absorber and the vertical damping C₄of the air spring.（3）Build a hybrid agent model for high-speed trains.Five key suspension system parameters were taken as design variables,and the derailment coefficient,wheel load reduction rate,nonlinear critical speed,vertical smoothness,and comfort index were taken as the response targets,50%positive and negative of the initial design value are the design space,and the classic hybrid proxy models of offline update and online updated hybrid adaptive proxy model are set up.The use of two models for multi-goal optimization to lay the foundation.（4）Multi-target optimization of high-speed train suspension system parameters.Multi-target optimization of the above two models are optimized.The classic hybrid proxy model selects the adaptive simulation annealing algorithm for optimization and calculation.HAM continuously iterates to update multi-target optimization through the sample point.Finally,the optimization sets obtained by the original solution and the two methods were compared and analyzed.Compared with the comparison,it was found that the comprehensive optimization rate of HAM on dynamic performance was higher,and there were less"expensive points"required for building models.Therefore,HAM has better adaptability in multi-target optimization of high-speed train suspension system parameters.