高超声速飞行器分解集成轨迹预测算法

针对无动力滑翔高超声速飞行器的轨迹预测问题,提出了分解集成轨迹预测模型。依据运动轨迹的周期跳跃特性,运用先集成再分解的轨迹预测思路,首先将运动轨迹序列分解为具有趋势性、周期性和随机性特征的子序列,再针对每项子序列的特征采用相应的子轨迹预测模型,最后将各子轨迹预测模型预测结果的集成作为最终预测值。由于子序列与子轨迹预测模型具有更高的契合度,使得分解集成轨迹预测算法相对于使用单一模型的轨迹预测算法更具优势。仿真实验表明,分解集成轨迹预测算法显著提高了轨迹预测精度。

Decomposition ensemble trajectory prediction algorithm for hypersonic vehicle

Aiming at the trajectory prediction of unpowered gliding hypersonic vehicle and its application requirements, a decomposition ensemble trajectory prediction model is proposed.Firstly, according to the periodicity leap characteristic of the hypersonic vehicle, the trajectory sequence is decomposed into sub-sequences with trend, periodicity and randomcharacteristics by using the thought of decomposition and integration. Then, using the corresponding trajectory prediction model to predict each trajectory sub-sequence on thebasis of sub-sequence characteristics. Finally, the ensemble of the prediction result of each trajectory prediction sub-model is taken as the final prediction value.Due to the high degree of fit between the sub-sequence and the trajectory prediction sub-model, the decomposition ensemble trajectory prediction algorithm has moreadvantages than the trajectory prediction algorithm using a single model. The simulation results show that the decomposition ensemble trajectory prediction algorithm canimprove the accuracy of trajectory prediction effectively.