Safety and efficiency are two classical conflicting objectives in the air traffic system: an increase in efficiency may come at the cost of increasing density of aircraft in the space, which increases collision risk and controllers' workload. Nationwide air traffic network flow optimization (ATNFO) is an effective way to pursue trade-offs between safety and efficiency by optimizing flight departure time-slots and routes within a given time period and under the latest airspace resources. Solving a national ATNFO problem is usually bedeviled by "the curse of dimensionality" as it consists of a huge number of variables. This paper presents a specific "divide-and-conquer" based approach, namely H-COEA, to solve it. Firstly, an effective chromosome representative scheme, which can be naturally divided into 3 sub-components, i.e., the departure time-slots, the heuristic for selecting flight route, and the timetabling indicating the order and fairness for flight to select route, is employed. And then, the corresponding 3 sub-populations are co-evolved based on a Cooperative Co-evolution (CC) paradigm. Four different-scale ATNFO problems are solved with H-COEA and the state-of-the-art multi-objective evolutionary algorithms. Results show that H-COEA obtains better trade-offs between safety and efficiency, making CC paradigm appropriating for solving large-scale ATNFO problem.
Keywords: Air traffic network flow optimization; Cooperative co-evolution; Large-scale; Multi-objective optimization.
© 2025. The Author(s).