基于GA-BP神经网络的汽车空气阻力系数预测研究

针对整车空气动力性能开发中数值计算耗时长的问题，提出一种基于GA-BP神经网络的汽车空气阻力系数预测方法。将汽车部分特征参数作为输入变量，经外流场仿真得到的空气阻力系数作为输出变量，获取数据集。采用遗传算法对BP神经网络进行参数寻优，最终建立基于遗传算法的BP神经网络模型，验证不同训练集数量对模型预测精度的影响。结果表明，GA-BP神经网络在训练样本较少时也能维持较高的预测精度，可用于汽车空气阻力系数的快速测。

Prediction of Vehicle Drag Coefficient Based on GA-BP Neural Network

In the development of vehicle aerodynamic performance, traditional numerical simulations proved to be time-consuming. Therefore the GA-BP neural network method was proposed to predict the vehicle drag coefficient. Some typical characteristic parameters were chosen as the input variables, while the drag coefficient obtained from the external flow field simulation served as the output. The BP neural network parameters were optimized using a genetic algorithm, and the genetic algorithm-BP neural network model (GA-BP) was finally established. The effect of varying training set sizes on the prediction accuracy of the model was evaluated. The results show that the GA-BP neural network model has high prediction accuracy even with smaller datasets, and is suitable for predicting the drag coefficient.