Comparison of performance the neighborhood operators and the ratio of search local-global in minimizing makespan using an GA-VNS collaboration to solve reactive production scheduling

• Published in: IECON 2016 - 42nd Annual Conference of the IEEE Industrial Electronics Society pp. 66 - 71
• Main Authors: Tuma, Carlos C.M., Morandin, Orides, Carida, Vinicius F.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01-10-2016
• Subjects: Biological cells; Context; Genetic algorithms; hybridization; Job shop scheduling; local search; neighborhood operator; ratio local/global; Sociology; Statistics
• DOI: 10.1109/IECON.2016.7794053

Reactive production scheduling involves searching methodologies in a wide solution space. A criterion often used in the analysis of results for this type of problem is makespan. Many artificial intelligence techniques are being used in an attempt to minimize the makespan. The use of global and local search hybridization has been well referenced. Usually the each global search for a solution applies some local searches in the same solution, trying to find a better solution in the neighborhood. These local search operations have used neighborhood operators. This paper has two research objectives: investigation of the better neighborhood operator and the better ratio local/global. This proposal implements a hybrid Genetic Algorithm (GA) with Variable Neighborhood Search (VNS). The proposal was validated in three scenarios of Job Shop Scheduling (JSS) coming from Taillard base and using three neighborhood structures.