Data mining parameters' selection procedure applied to a multi-start local search algorithm for the permutation flow shop scheduling problem

摘要

In this paper, a new metaheuristic algorithm is developed, suitable for solving combinatorial optimization problems, such as the job shop scheduling problems, the travelling salesman problem, the vehicle routing problem, etc. This study focuses on permutation flow-shop scheduling problem. The proposed algorithm combines various techniques used in local search. As various elements of the proposed algorithm may be tuned, a systematic data mining procedure is followed and utilizes data from a number of executions in order to build models for the suitable parameterization for every problem size. The results, using the model suggested parameter combinations, are presented using benchmark instances for the permutation flow-shop scheduling problem from the literature. The results show that the followed parameter control procedure improved vastly the efficiency of the proposed algorithm.