A meta-heuristic algorithm to generate Pareto frontiers for a simulation-based flexible flow shop scheduling problem

This thesis develops a meta-heuristic algorithm to generate Pareto frontiers for a real world scheduling problem. The problem can be modeled as a flexible flow shop with sequence dependent setup times, common due dates and stochastic processing times. The objective is to minimize the total tardiness and make-span so that it is a multi-objective scheduling problem. Simulation technique is used to represent the constraints and stochastic features existing in the flexible flow shop so that the robustness and effectiveness of the solutions can be validated by the verified simulation model. In order to obtain high quality solutions, a meta-heuristic for randomized priority search (Meta-RaPS) algorithm based on ATC (ATC-Meta-RaPS) is constructed and evaluated by comparing the Pareto frontiers with the objective results obtained from simple dispatching and composite dispatching rules. Analysis concerning the effectiveness of ATC-Meta-RaPS algorithm is completed by comparing the Pareto frontier with that from other Meta-RaPS algorithms. From the comparison, ATC-Meta-RaPS algorithm seems effective and efficient for providing good solutions for both objectives defined in the thesis. A hybrid Meta-RaPS algorithm is also potentially effective for multi-objective scheduling problem but its efficiency is not desirable.;Key words: multi-objective, Meta-RaPS, simulation, scheduling, Pareto frontier.