A proxy model of aerodynamic characteristics of high-speed trains,such as aerodynamic drag and aerodynamic lift,was constructed by machine learning algorithm,and a multi-objective optimization design of high-speed trains was completed based on the above aerodynamic characteristics.Firstly,the simplified three-dimensional model of CRH3 train is established,in which the locomotive part has the greatest influence on the aerodynamic performance of the train.Secondly,eight shape design variables were selected from the loco-motive by Sculptor.100 sets of sample data were extracted by using an optimal Latin hypercube test design.Fluent software was used to calculate the aerodynamic characteristics of the train head in the sample space.Thirdly,the BP neural network model is optimized by genetic algorithm to obtain the proxy model of aerody-namic characteristics such as aerodynamic drag and aerodynamic lift.Finally,NSGA-Ⅱ genetic algorithm was used to optimize the multi-objective problem.The aerodynamic drag is reduced by 4.3％and lift by 8.0％,and the aerodynamic characteristics of the train are effectively improved.
维普
万方数据
