Solving a Large-Scaled Crew Pairing Problem by Using a Genetic Algorithm

This paper presents an algorithm for a crew pairing optimization, which is an essential part of crew scheduling. The algorithm first generates many pairings and then finds their best subset by a genetic algorithm which incorporates unexpressed genes. The genetic algorithm used employs greedy crossover and mutation operators specially designed to work with chromosomes of set-oriented representation. As a means of overcoming the premature convergence problem caused by greedy genetic operators, the chromosome is made up of an expressed part and an unexpressed part. The presented method was tested on real crew scheduling data.