Air Traffic Flow Reduction Using Genetic Algorithms in Brazilian Airspace

摘要

Air Traffic Flow Management (ATFM) is a process to treat an online problem, which is related to many complex attributes. The air traffic controllers are responsible to handle data and acquire knowledge from currently airspace scenario to detect possible risks situations and take some actions to reduce air traffic flow congestions. Recently, Reinforcement Learning (RL) and Multiagent System (MAS) were used to solve this kind of problem. As an alternative, Genetic Algorithms (GA) technique was also used to select the best group of actions, using an evolutionary approach when happens new situations. This paper presents a case study applied in Brazilian Airspace with better suggestions of restrictive measures. The main goal is reduce the impact of the action on airspace scenarios using genetic algorithms to choose the actions for specific scenario. As initial studies, Brasilia International Airport (SBBR) and Congonhas Airport (SBSP) in Sao Paulo, are involved. The developed method improved the ATFM with the better performance of 8% to 21% with the implementation of Decision Support System (DSS).