基于OLS-RBF神经网络的进场飞行时间预测

Arrival Flight Time Prediction Based on OLS-RBF Neural Networks

航空器预计到达时刻(ETA)是航空器进场排序与调度的基础,因此进场航空器飞行时间的快速与准确预测显得尤为重要。基于历史雷达轨迹分析,通过RBF(Radial Basic Function)神经网络构建进场航空器进港时的高度、速度、进场飞行距离与进场飞行时间的映射关系,利用正交最小二乘算法设计基于RBF神经网络的进场飞行时间预测模型。以上海浦东机场VMB进港点进场航班为例进行仿真验证,在考虑航空器机型的情况下,可将航空器飞行时间预测的均方根误差控制在50 s以内。仿真结果表明,提出的方法能够有效地实现进场飞行时间的快速与准确预测。

Estimated Time of Arrival （ETA） plays a great role in arrival sequencing and scheduling, therefore it is particularly important to predict the arrival flight time quickly and accurately. Based on the analysis of historical radar track, with the help of RBF （Radial Basic Function） Neural Network, the map- ping relationship is constructed between the arrival aircraft＇s altitude/speed at the metering point, flight distances and flight time. And then, the orthogonal least squares （OLS） algorithm is adopted to design the RBF- NN based arrival flight time prediction model. Taking the arrival aircrafts via VMB to Shanghai Pud- ong Airport as examples, the RMSE between estimated and actual time of arrival is controlled within 50 s with consideration of the same aircraft type. The simulation results indicated that the proposed approach is able to predict arrival flight time quickly and accurately.