在载机和无人机适配的方案阶段,分离边界的确定需要准确的快速预测方法。目前的分离轨迹预测方法仅考虑无人机质心处的载机干扰流场,对姿态角的预测不准,由于轨迹-姿态的耦合作用,进一步影响了轨迹预测的准确性。文中首先构建了基于神经网络的气动响应模型,研究了计算平均气流角与动压的影响因素,提出了一种基于双参考点的无人机离机轨迹和姿态高精度快速预测方法。针对翼展为4 084 mm载机投放翼展为300 mm无人机的投放系统,与基于嵌套网格求解Navier-Stokes方程的计算结果相比,文中方法的计算结果在空间位置上相差小于5 mm,包括滚转角在内的3个姿态角最大相差小于0.24°,相较于其他的预测方法,对姿态的预测精度提高了1个数量级;采用数值模拟方法计算一条分离轨迹需3~5 d,文中方法只需约120 s,大大提升了计算效率。文中的方法适用于大型载机空投小型带升力面投放物的分离轨迹与姿态的快速计算,可为型号方案设计阶段的分离安全性评估提供参考依据。

In the phase of conceptual design, it is required to obtain an prediction method which is time-saving and accurate in determining the separation boundary. The current prediction methods only refer to the interaction flow field of the carrier at the UAV centroid, resulting in inaccurate attitude prediction. Due to the trajectory-attitude interaction, it is further affected that the accuracy of trajectory prediction. In this paper, the aerodynamic response model based on neural network is firstly constructed, on the basis of studying the calculation method of equivalent angle of attack, angle of side-slip and dynamic pressure, a fast and precise prediction method of UAV separation trajectory and attitude considering double-reference-points is proposed. Compared with the results of CFD-6DOF, for a system about the 300 mm wingspan UAV airdropped from the 4 084 mm wingspan airborne, the deviations of the method are less than 5mm in space position and 0.24° in attitude angle. The prediction accuracy of attitude is improved by an order of magnitude compared with other prediction methods. It takes 3~5 d to calculate a separation trajectory using the numerical computational method, but only about 120 s using the method presented in this paper, which greatly improves computing efficiency. The method is suitable for the fast calculation of the separation trajectory and attitude of small stores with lifting surface dropped from large airborne, and can provide a reference for the separation safety evaluation in the phase of conceptual design for aircraft.