PSO-K-Means Clustering-Based NSGA-III for Delay Recovery

With the development of the civil aviation industry, flight delay rates have risen rapidly. The flight delay recovery can enable the swift adjustment of plans, collaborative efforts for minimizing impacts, and early warning and preparedness. Therefore, an enhanced non-dominated sorting genetic algorithm III(NSGA-III) with the K-Means clustering algorithm with particle swarm optimization (PSO) (PKM), Cubic chaos and Penalization-based Boundary Intersection (PBI) aggregation (PKMCP-NSGA-III) is developed to realize a flight delay recovery method. In the PKMCP-NSGA-III, a novel PKM algorithm based on K-Means clustering and PSO is designed by clustering the initial reference points and optimizing the optimal number of clusters. Cubic chaos is employed to enhance the diversity within the populations and improve the optimal solution search. The PBI aggregation function in the niche preservation is defined to balance the convergence and diversity. Then, the objectives of the minimum total operating cost for airlines, the minimum delay losses of passengers and the minimum total delay time are constructed and a novel flight delay recovery method using PKMCP-NSGA-III is proposed. Finally, the results on Deb-Thiele-Laumanns-Zitzler(DTLZ) and Waterfall Function Group (WFG) show that the PKMCP-NSGA-III exhibits the superior diversity and search capability by comparing with six cuttingedge algorithms. The experiment results on actual airport flight delay data show that the proposed recovery method has great potential for decision-making in airport operations.