Applications of Multi-objective Evolutionary Algorithms to Cluster Tool Scheduling

In this paper, we propose a method of using Multiobjective Evolutionary Algorithm (MEA) to obtain an optimal deadlock-free schedule during the flexible process of the cluster tool. The MEA approach, a method of combining the genetic algorithm with the multi-objective method, can consider the relation of the parameter and the solution space in the same time to explore the optimum solution. To solve deadlock and re-entrant problems, once the deadlock of scheduling occurs and a high penalty value will be assigned to the makespan. Therefore, we have take advantage of fitness value and variance integrating with Method of Inequalities and Improved Rank-based Fitness Assignment Method to transfer rank value into Pareto curve and to eliminate unfeasible solution after evolution. In conclusion, MEA can build mathematic model easily, global searching for all solutions, and also achieving optimal solution.