In this study, the discrete adjoint approach with the grid deformation technique based on radial basis function was presented to solve the tough problem that the complex shape of the defrosting duct is very difficult to be optimized with multiple parameters in the improvement of the defrosting performance of vehicle. Firstly, the defrosting performance of a light truck was analyzed. Then the discrete adjoint approach was applied by taking the average Nusselt-number as the objective function. The normal surface sensitivity was used in the deformation of the defrosting duct. After 10 optimization cycles, the optimal model was acquired. The results show that, after optimization, the sum of the average Nusselt-number is increased from 3164.16 to 3350.54, which has a positive effect on improving the convection and heat transfer capacity on the windshield effectively. Compared with the initial model, the airflow of the outlet near the windshield is increased from 95.07 g/s to 96.94 g/s, the defrosting time required by the optimal model is reduced by more than 20s. Therefore, the discrete adjoint approach can optimize the complex shape of defrosting duct with multiple parameters effectively, which provides an effective method in the study of automobile defrosting performance.

本研究提出了一种基于径向基函数的网格变形技术的离散伴随方法，解决了除霜管道复杂形状难以通过多个参数进行优化来改善除霜性能的难题。车辆。首先，分析了轻型卡车的除霜性能。然后以平均努塞尔数为目标函数，采用离散伴随法。法向表面灵敏度用于除霜管道的变形。经过10个优化周期，获得了最佳模型。结果表明，优化后，平均Nusselt数之和从3164.16增加到3350.54，这对有效提高挡风玻璃的对流和传热能力有积极作用。与初始模型相比，挡风玻璃附近出口的气流从95.07 g / s增加到96.94 g / s，最佳模型所需的除霜时间减少了20多秒。因此，离散伴随方法可以有效地优化多参数化霜管道的复杂形状，为研究汽车除霜性能提供了一种有效的方法。