Considers the problem of scheduling rounds of a non-professional table-tennis league. We formalize the problem in terms of a timetabling optimization problem, then we solve this highly constrained problem with a permutation-based genetic algorithm for which feasibility-preserving operators are defined. Since coding and operators cannot warrant feasibility in every case, the fitness function penalizes constraint violations. This algorithm is compared to an even more elaborated variant, which additionally aims at repairing infeasible solutions produced by the genetic operators.
展开▼