IMPROVED MULTI-OBJECTIVE PARTICLE SWARM OPTIMIZATION FOR JOB-SHOP SCHEDULING PROBLEMS

Abstract

The Job-shop Scheduling Problems (JSP) is a typical production scheduling problem widely applied as a single-objective optimization in existing research. However, this is not suitable for cases in the real world, which normally consist of multi-objective criteria. In this paper, a multi-objective Particle Swarm Optimization (MOPSO) for solving JSP is developed, where it involves three key MOPSO attributes to be improved as identified from the literature which are diversity of swarm solutions, exploitation/exploration mechanisms throughout the search process and premature convergence. In order to address the issues related to these attributes, improvement strategies are implemented that include reinitialization of particles, systematic switch of best solutions and Tabu search-based mutation. The computational results in solving benchmark instances demonstrated that the improved MOPSO performs well in terms of finding non-dominated solutions in different regions of the Pareto fronts with a wider spread and producing a higher percentage of solutions in comparison with other established techniques.