基于机器学习的高超声速飞行器双曲率前缘气动热预测方法  

Prediction method of aero-heating of hypersonic vehicle bi-curvature leading edge based on machine learning

作　　者：杨帆[1,2] 林明月 胡宗民 罗长童[1,2] YANG Fan;LIN Mingyue;HU Zongmin;LUO Changtong(State Key Laboratory of High-Temperature Gas Dynamics,Institute of Mechanics,Chinese Academy of Sciences,Beijing 100190,China;School of Engineering Science,University of Chinese Academy of Sciences,Beijing 100049,China)

机构地区：[1]中国科学院力学研究所高温气体动力学国家重点实验室,北京100190 [2]中国科学院大学工程科学学院,北京100049

出　　处：《北京航空航天大学学报》2024年第9期2826-2834,共9页Journal of Beijing University of Aeronautics and Astronautics

基　　金：国家自然科学基金(12172365,12072353,12132017);国家重点研发计划(2019YFA0405204)。

摘　　要：高超声速气动热预测技术是高超声速飞行器发展的关键技术之一,气动热环境的精准预测对飞行器热防护系统设计及气动布局优化具有重要意义。为快速获得高超声速飞行器表面的热流分布情况,缩短飞行器设计周期,基于具有广义可分离特性、可实现强非线性数据快速建模的多层分块(MBB)算法,提出一种针对高超声速飞行器双曲率前缘气动热分布的快速预测方法。通过数值计算获得双曲率前缘驻点区的气动热分布作为训练集数据,基于MBB算法提出预测热流分布的显式表达式,对表达式预测结果的统计分析显示,表达式预测值与测试集数据的偏差低于2%,这表明其具有较高的预测精度;将驻点区热流分布表达式进行外推,验证了机器学习公式在不同几何外形下的适用性。在双曲率前缘构型的防热设计及气动外形优化阶段,所提表达式可实现气动热环境的精准、快速预测。The prediction technology of hypersonic aero-heating is one of the key technologies for the development of high-speed vehicles.Creating an efficient method for predicting the hypersonic thermal conditions is highly important for designing thermal protection systems and optimizing aerodynamics.In order to obtain the heat flux distribution on the surface of hypersonic vehicles quickly and to shorten the vehicle design cycle,a fast prediction method for the aerothermal environment of the bi-curvature leading edge of hypersonic vehicles is proposed based on the multi-level block building(MBB)algorithm.The MBB algorithm is distinguished by its generalized separability,which enables it to efficiently represent highly nonlinear data.First,numerical simulations are conducted to obtain the database composed of the aero-heating data of the bi-curvature leading edges of the vehicles in the training set.Based on the MBB algorithm,an explicit expression for predicting the distributions of heat flux is given.The statistical analysis results demonstrate that the discrepancy between the estimated value and the observed value is below 2%,suggesting that the formula given in this study exhibits a high level of predictive precision.Further,an extrapolation of the formula is performed to verify its applicability for different geometric shapes.At the stage of thermal design and aerodynamic optimization of the bi-curvature leading edge configuration,the formulation proposed in this paper enables accurate and rapid prediction of the aerodynamic thermal environment.

关 键 词：高超声速流动 气动热 多层分块算法 快速预测方法 驻点热流 双曲率前缘 

分 类 号：V221.3[航空宇航科学与技术—飞行器设计] TB553[理学—物理]