Loading...

An Improved Ant Colony Optimization Algorithm for Solving Airline Crew Pairing Problem

Feizi Karim Abadi, Sina | 2023

5 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 56957 (01)
  4. University: Sharif University of Technology
  5. Department: Materials Science and Engineering
  6. Advisor(s): Varmazyar, Mohsen
  7. Abstract:
  8. In airlines, the problem of ‘flight crew scheduling’, as a part of the organization resources, holds great importance. Furthermore, considering the numerous and specific regulations regarding the working hours of flight crew, as well as constraints resulted by origin, destination, and time of each flight, this problem is also highly complex. The most significant challenge in solving this problem lies in the “construction and selection of the crew pairings” In this research, first, according to the literature, definitions and regulations necessary for the problem are stated. Then the problem is modeled as a "set covering problem," in which the rows of the problem correspond to flights which the airline is supposed to cover within a given time horizon. Additionally, the columns of the set covering problem represent all feasible crew pairings. Each crew pairing is a sequenced subset of flights, constructed based on the mentioned regulations. Due to the NP-completeness of this problem in terms of complexity, in this research an "improved ant colony optimization algorithm" is proposed, in which a new formula for calculating the probabilities of ants’ decisions is used. Furthermore, real and diverse problem instances are solved using the proposed algorithm, the mathematical model of the problem, and the best existing evolutionary algorithms and other meta-heuristics in the literature, including two versions of "genetic algorithm," a version with binary variables of the "particle swarm optimization algorithm". Then, by presenting the obtained numerical results, it is shown that the improved ant colony optimization algorithm has performed better in terms of the "average objective function value" compared to the other mentioned algorithms, with an average improvement of 3.59% in small instances, 2.58% in medium instances, and 4.96% in large instances. Also, in terms of the "best objective function value," it has an average distance of 3.73% from the optimal values obtained from solving the mathematical model of the problem in small and medium-sized problem instances. Finally, using statistical tests, the meaningfulness of the performance difference between these algorithms is demonstrated.
  9. Keywords:
  10. Set Cover Problem ; Crew Pairing Problem ; Genetic Algorithm ; Particles Swarm Optimization (PSO) ; Ant Colony Optimization (ACO) ; Airline Crew Pairing Problem

 Digital Object List

 Bookmark

No TOC