Cloud Model Evolutionary Multi-objective Flexible Job-shop Scheduling Based on Improved Non-dominated Sorting

Abstract

A multi-objective evolutionary algorithm is applied to research the flexible job-shop scheduling problem, which sets maximum makespan, the total workload of machines and maximum workload of machine as optimization goals. Aiming at improving performance of NSGA- at distinguishing and comparing nonⅡ-dominated individuals, a modified non-dominated sorting algorithm is designed so that it can distinguish non-dominated individuals rapidly, eliminate dominated individuals and enhance the conformation efficiency of non-dominated sets. Combining the characteristics of flexible job-shop scheduling problem and properties of evolutionary algorithm, and introducing evolutionary strategy based on cloud model, a multi-objective flexible job-shop scheduling algorithm based on improved non-dominated sorting is proposed. Applying the excellent characteristics of cloud model in both fuzziness and randomness to maintain evolutionary populations and ameliorate the distribution breadth and uniformity of non-dominated solution. Making use of multi-attribute decision model based on weighted grey target to select the most satisfied schedule. Using the proposed algorithm to test the benchmark problems and analyze the results, and testified its feasibility and effectiveness. The proposed algorithm is applied to a real production case study, and the most satisfied schedule is generated.