Real time prediction of worst case air traffic sector collision risk using evolutionary optimization

Abstract

Given the unprecedented growth in air transport, air traffic controllers and safety managers are exploring new approaches for collision risk assessment specifically in real time or near future term. In this paper a real time predictive approach for identifying worst case sector collision risk is proposed. An evolutionary framework is introduced to evolve flight maneuvers that may lead a traffic scenario to high collision risk. The proposed methodology discretizes the execution time of a baseline air traffic scenario into discrete time intervals based traffic scenarios. These traffic scenarios are then initialized at their given time intervals in a sector which are then perturbed with flight maneuvers to maximize collision risk. Flight maneuvers are evolved using an evolutionary framework with the objective of maximizing the collision risk for a given traffic scenario. Results indicate the effectiveness of the proposed methodology to successfully identify flight maneuvers which may increases the collision risk for a look ahead time. Results also indicate that when most of the flights are in the first half of their flight path, DESCEND and TURN RIGHT maneuvers may increase the collision risk. Also maneuvers, when flights are exiting the sector, do not significantly effect the collision risk. It was also found that traffic flow should be managed when flights enter the sector or when flights are exiting the sector (to facilitate coordination with neighboring sector controller) as this may reduce the collision risk.