Crew Scheduling in Civil Aviation Based on Multiobjective Genetic Algorithms

This work addresses the problem of Crew Rostering in civil aviation, which requires the conciliation between operational efficiency, safety and labor standards. The complexity of the problem lies in the vast number of variables and constraints, making empirical methods susceptible to failures and inefficiencies that increase costs and risks. In view of the unfeasibility of exact methods in large instances, a heuristic model based on the NSGA-II Genetic Algorithm is proposed from the perspective of Multiobjective Optimization. Using real data from the Flightradar24 website, the model sought to simultaneously minimize the number of pilots used and the standard deviation of the workloads. For a network of 166 flights, the algorithm reached the limit of six pilots, presenting a balanced distribution of hours and the absence of contractual penalties.