Air Traffic Performance Improvement of Congested Terminal Airspace with Genetic Algorithm based Optimization

Abstract

With the increase of air traffic demand, the terminal airspace becomes more and more congested. Future air transport systems are expected to handle increasingly heavy demand on air traffic, especially in a highly constrained terminal airspace. Therefore, the realizable capacity of current terminal airspace is a challenge for future air transport development. The objective of this thesis is to understand and identify the system vulnerability of congested terminal airspace, and then optimize aircraft landing sequence and runway allocation in order to maximize the TMA capacity and minimize delay. Thereby an algorithm for optimal arrival flight sequencing and runway assignment in a terminal airspace has been proposed. In this project, the optimization scheme selected to tackle this problem is genetic algorithm. Genetic algorithm is a problem solving system based on principles of evolution and heredity. It consists of an iterative procedure to identify and preserve the candidate solutions with higher fitness, as well as generating new possible candidates through chromosome operations. After a reasonable time of iteration, a feasible optimal solution will be generated. As demonstrated in this thesis, genetic algorithm is well adapted for resolving the congestion of terminal air traffic. The outcome of this optimization scheme is only the assignment of each aircraft with a respective landing time and runway, however it does not provide instructions on how aircraft can achieve this in real-world actions. Although the separation limits are in force for the initial positing and landing sequence of aircraft, there exist a risk that the separation limit may be violated during the periods in between. Therefore this is one of the major motivations to design a simulation scheme and incorporate speed adjustment and holding within it. In daily operations, altering the airspeed is a common tool that air traffic controllers used to achieve the desired arriving time of the aircraft. Hence, it is considered as the primary means of aircraft adjustment in the simulation. To visualize the optimization result and adapt for realistic situations, a dynamic simulation was designed. The fundamental logic of the dynamic simulation is to incorporate a quick optimization scheme into the simulation loop. As a result, the computational performance will be significantly improved; bring it closer to be implemented in the real world air traffic control dynamic environment. In this thesis, the proposed optimization scheme and simulation design is validated by using the arrival routes of Hong Kong International Airport. Accordingly, the genetic algorithm based optimization design is capable of generating a rather optimum landing sequence and runway allocation with minimal computation time. Therefore, the proposed scheme may be implemented into the real world air traffic control procedure to fill the void left in strategic decision making during high workload periods. This is not only crucial to efficient air traffic management, but also a cost effective solution to deal with the increasing flight demands.