基于航空信息网络的枢纽机场航班延误预测模型

航班延误一直是机场运营管理的一大难题，本研究报告面向区域多机场，重点针对机场集团内枢纽机场的航班延误问题，提出基于航空信息网络的航班延误预测模型NBFDM. 该模型不仅使用了航班自身的相关属性，并且还考虑了航空信息网 络内其他机场的因素对航班延误的影响. NBFDM模型首先提取航班本身的特征和该航班飞行前一段时间内航空信息网络的特征，然后使用PCA进行降维，对降维处理后的特征再使用SVR方法，得到非线性回归模型，用于预测航班的延误时间. 实验表明本研 究报告所提模型NBFDM相比仅使用航班自身属性的模型，对航班延误时间的预测误差降低约20%.

Study on flight delay prediction model based on flight networks

Delay of flight has been regarded as one of the toughest difficulties in aviation control. How to establish an effective model to handle the delay prediction problem is a significant work. This study proposes NBFDM, a model based on flight information network, which performs flight delay prediction of hub airports according to the state of flight network. The model not only takes into consideration the properties of individual flights, but also considers the related factors which may affect the flight delay. NBFDM first extracts the properties of individual flights as well as the features of flight network before the flight departure, and then apply PCA method to reduce the dimension of feature. After that, the SVR method is employed to obtain the non-linear regression models which are used to predict flight delays. Experiments show that compared with the method that only considers flight properties, the prediction error of proposed model NBFDM is reduced by about 20%.