Many-objective low-cost airline cockpit crew rostering optimisation

This research addresses the cockpit crew rostering problem in a low-cost airline. Through interviews with the aircrew planner, four objectives are considered vital for effective roster preparation; minimisation of nautical mile cost, balance workload among cockpit crews, maximisation of preferential requests from senior pilots and minimisation of the number of repeated flight patterns flown by individual pilots. Since the problem is NP-hard with many conflicting objectives that need to be optimised simultaneously, multi-objective evolutionary optimisation is an effective technique to solve this problem. As a result, the hybridisation between MOEA/D and HBMO algorithms, namely MOEA/D-HBMO, is developed. The proposed algorithm is compared to MOEA/D and HBMO. It is observed that MOEA/D-HBMO outperforms the counterparts in convergence related metrics and it could discover the best extreme points to meet every objective.