A Method of Flight Off-Block Time Prediction Based on LSI-CNN Model
A Method of Flight Off-Block Time Prediction Based on LSI-CNN Model
复制标题
基于LSI-CNN模型的航班离区时间预测方法
DOI：
10.1109/iccasit48058.2019.8973223
复制
发表时间：
2019
期刊：
影响因子：
--
通讯作者：
Zhang Yang
中科院分区：
文献类型：
--
作者：
Mao Jian;Dang Zhengyang;Liu Yang;Li Dingliang;Deng Dingyu;Zhang Yang
为了提高机场、航空公司和空管部门的运行效率和协同决策效率，提出了一种基于潜在语义索引（LSI）和卷积神经网络（CNN）结构的LSI-CNN模型。该模型利用潜在语义索引技术识别飞行保障过程中与离档时间相关的时间节点。将特征重构后的输入数据输入到特定的卷积神经网络中，最终的预测是得到离块时间。实验结果表明，在（±）5 min、（±）10 min和（±）15 min的误差范围内，LSI-CNN模型的平均预测精度高于基于全连接神经网络的模型，具有较好的鲁棒性，能够准确识别与航班离挡时间相关的重要时间节点。具有较高的预测精度。
In order to improve the operational efficiency and collaborative decision-making efficiency of airports, airlines and air traffic control departments, a LSI-CNN model based on latent semantic index(LSI) and convolutional neural network(CNN) structure is proposed in this paper. This model identifies the time nodes related to off-block time in flight support process by using latent semantic index technology. The input data after feature reconstruction is input into a specific convolutional neural network, and the final prediction is to get the time of off-block. The experimental results show that the average prediction accuracy of LSI-CNN model is higher than that of the model based on full-connected neural network in the error range of (±) 5 min, (±) 10 min and (±) 15 min. The model has good robustness and can accurately identify the important time nodes related to flights off-block time. It has higher prediction accuracy.