基于双重力模型和人工神经网络的空中交通流量组合预测

Air Traffic Flow Combinational Forecast Based on Double Gravity Model and Artificial Neural Network

基于四阶段法,采用双重力模型,预测全国空域的交通流量OD分布.针对历史数据的随机性和周期性,建立灰色广义回归神经网络组合模型,得出空中交通流量的预测结果.采用马尔可夫链预测模型,分析了组合预测结果.算例分析表明,对比华北地区空中交通流量统计数据,与回归分析和广义回归神经网络模型的结果相比,本文模型预测结果的精度更高、更可信.

The four stage method and a double gravity model were used to forecast the OD （origindestination） distribution of air traffic flow in the whole airspace of China. In view of the randomicity and periodicity of historical data of air traffic flow, a GM-GRNN （gray model and generalized regression neural network） combinational model was built to obtain the forecast results. The forecast results were then analyzed by Markov chain forecast model. A case study shows that, compared with the statistical data of air traffic flow OD distribution in north China airspace, the forecast results by the proposed combinational model are more precise and credible than the regression analysis and the traditional GRNN model.