NSGA-II implementation details may influence quality of solutions for the job-shop scheduling problem

Abstract

The helper-objective approach for solving the job-shop scheduling problem using multi-objective evolutionary algorithms is considered. We implemented the approach from the Lochtefeld and Ciarallo paper using NSGA-II with the correct implementation of the non-dominated sorting procedure which is able to work with equal values of objectives. The experimental evaluation showed the significant improvement of solution quality. We also report new best results for 16 out of 24 problem instances used in the considered paper.