基于SVR方法的飞行仿真气动数据辨识及预测

Identification and prediction of aerodynamic data in flight simulation based on SVR

气动数据的辨识是全状态飞行仿真中的关键点。辨识数据模型的有效性与完整性将直接决定飞行运动仿真的准确度并将影响后续工作的效率。由于实验获得的数据具有离散性和不完整性,往往不能获得飞机整个飞行包线内各种状态下的气动数据。分析了飞机气动参数的分布类型,提出一种基于ε-支持向量机理论的气动数据辨识及预测方法,通过对离散气动数据的曲线和曲面拟合验证了该方法的可行性、精确性和鲁棒性。

Aerodynamic data identification is a key point in flight simulation,the integrity and useableness of the acquired data model have a directly effect on flight simulation precision and the efficiency of follow-on jobs.Due to the incompletion and randomness of the date acquired through experiment,the aerodynamic data can＇t be obtained fully in a whole flight envelope.The distributing style of the aerodynamic data was analyzed,and the identification and prediction method based on ε-SVR was put forward.The feasibility,precision and robustness are validated by matching the random aerodynamic data in two dimensions and three dimensions.