Fractional factorial analysis to the configuration of simulated annealing applied to the multi-objective optimization of master production scheduling problems

Searching for the global optimal solution in a Master Production Scheduling problem usually demands an effort most industries are not willing to pay. Therefore, the use of meta-heuristics that generates good solutions in reasonable computer time becomes an attractive alternative. However, such strategies are usually complex to implement and configuring their parameters is not a trivial task because of the number of usually conflicting objectives involved. The use of statistical methods that facilitate the set-up of the heuristic's parameters becomes therefore necessary. Knowing which parameters are more important, that is, the ones that really affect the solution quality, and those that are irrelevant, is very important for chosen technique performance. This work presents how fractional factorial analysis can be applied to the configuration of simulated annealing used for optimization of Master Production Scheduling problems. Two scheduling scenarios illustrate the use of the proposed method.