An evolutionary goal-programming approach towards scenario design for air-traffic human-performance experiments

Air traffic controllers are responsible for maintaining a safe and efficient flow of air traffic in controlled airspace. Many aspects of air traffic control are impacted by the performance of air traffic controllers such as separation assurance tasks. In order to design more advanced air traffic management systems, there is a need for more experiments to be conducted which evaluate the impact of human performance on the system. The design of scenarios that satisfy/meet specific traffic characteristics needed by the analyst is a daunting task. For example, it is often required to design scenarios for a specific sector that have a specific number of conflicts to evaluate human task load. To schedule the aircraft within the time specified for the experiment, and given all the constraints imposed by the route structure and airspace design parameters for the sector in question, are far from trivial problems. In this paper, an evolutionary goal programming approach has been presented which generates a set of scenarios for use in these experiments. The evolutionary goal programming system aimed to generate scenarios meeting the criteria of a set number of conflicts in each of four conflict angle groups. Differential evolution was employed in addition to three modified methods for the optimization of the problem. It was found that the three modified methods outperformed the standard method by producing a greater number of scenarios meeting the set criteria.