Management-Action-Embedded Sector-Demand Prediction Models

Abstract

airtrafficcontrolandairlineactions,andthataccountsforbothshortterm (less than 30 min) and midterm (30 min to 2 h) predictions. The model consists of two parts: the open-loop prediction and the TFM action model. The open-loop predictions, similar to the traditional methods, are used to determine the possibility of demand-capacity imbalances at a future time, and help decide whether to activate the TFMaction.TheTFMactionmodelsimulatesthedemandreduction caused by the TFM actions. The closed-loop prediction represents the net result of the open-loop prediction and the TFM actions. The periodic autoregressive model and its variants [7,8] were used to build the model. The model considers both historical traffic flows to capture the midterm trend and flows in the near past to capture the transientresponse.Inaddition,forsevereweathercases,theweatherimpacted TFM action was modeled using weather forecast information. The proposed model provides both open- and closed-loop sector-demand predictions. Open-loop prediction is adequate for short durations. When looking at predictions for long durations, open-loop models produce large errors due to their inability to capture traffic initiatives and airline actions during the planning period.Acombinationofclosed-loopandopen-loopmodelsprovide decision-makers the full range of traffic behavior. The remainder of the paper is organized as follows. Section II provides the sector-demand data and a description of the open- and closed-loop sector-demand prediction models. Next, in Sec. III, a weatherfactorisintroducedandtheTFMactionmodelthatconsiders weather is described. The results and performance of the models are demonstrated in Sec. IV. Finally, a summary and conclusions are presented in Sec. V.