【摘要】 动车组外形设计后需要进行气动性能分析,以验证模型是否符合气动性能要求.针对传统试车实验和CFD方法成本高、用时长的缺点,提出了一种基于高斯过程回归的动车组气动性能预测模型.选取不同动车组的实体模型,利用STAM-CCM+仿真软件分别得到各模型的气动阻力模拟值,以此作为训练样本集合,然后采用高斯过程回归对训练样本进行学习,探寻动车外形与气动阻力关系,并预测明线动车组气动阻力,最后通过预测残差拟合验证了高斯过程回归预测模型的合理性和准确性.更多还原

【Abstract】 Aerodynamic performance analysis is needed to verify whether the model meets the aerodynamic performance requirements after the design of high-speed multiple unit.Aiming at the shortcomings of the traditional test experiment and CFD method,Gaussian process regression model is introduced into aerodynamic performance evaluation of high-speed multiple unit model.Different high-speed multiple model schemes were chosen,and the aerodynamic drag simulation values of each model were obtained by using STAM-CCM+simulation software as the training sample set,then Gaussian process regression was used to study the training samples so as to explore the relationship between the shape of motor vehicle and aerodynamic drag,and to predict the aerodynamic resistance of high-speed multiple unit.Finally,the rationality and accuracy of Gaussian process regression evaluation model were verified by the prediction of residual fitting.更多还原