航空器飞行状态预测的混合模型研究

Research on hybrid model for aircraft flight state prediction

航空器的飞行状态预测是飞行冲突探测的核心问题,也是保障飞行安全的关键所在。为了准确、高效地预测航空器的飞行状态,提出了一种HMM-BP混合模型。首先深入分析了航空器的飞行特点,从不同角度定义飞行状态并建立几何判定方法;然后通过HMM模型分别对航空器的飞行高度、航向以及速度特征进行时序建模;最后利用BP神经网络对航空器的飞行状态进行了推理预测。研究结果表明,该方法通过分析航空器在扇区内最初5min的雷达航迹数据,能够准确地预测其在扇区剩余时间的飞行状态,且计算速度快、预测效率高,可以有效协助管制员正确掌握航空器的飞行状态。

To predict an aircraft＇s flight state is not only the core issue for flight conflict detection,but also the key point to ensure flight safety. In order to predict the flight state accurately and efficiently,an HMM-BP hybrid model is proposed. Firstly,the flight characteristics are deeply analyzed,and the flight states are defined separately while the geometry-based methods for detection of flight states are developed.Secondly,temporal models for the characteristics of altitude,heading and speed are established separately by HMMs. Finally,the flight state is predicted by using the BP neural network. The result indicates that the proposed method can analyze the initial 5 minutes radar trajectory data in one sector,and predict the aircraft＇s flight state within the remaining time in the same sector accurately. With high calculation speed and prediction efficiency,it can help the controller to know the flight state well.