A multi-agents approach to solve job shop scheduling problems using meta-heuristics

Abstract

This paper presents a multi-agents approach to solve job shop scheduling problem using meta-heuristics. The job shop scheduling problem (JSP) is a traditional problem largely exploited in the operational research area through the academic research but belonging also to the domain of practical situations, especially in industrial companies. Several approaches have been used to investigate this problem that is very complex when addressing real industrial cases. The JSP is a NP-hard problem and no exact solution can be obtained in polynomial time. Meta-heuristics approaches when solving scheduling problems have proven to be very effective and useful in practical situations. Among them, Tabu Search (TS) and Genetic Algorithms (GA) have been used to solve optimization problems with success. The main reason is that these algorithms have good performance in terms of solution quality and execution time, when compared with optimization or simple heuristics techniques respectively. In this sense, the multi-agent approach proposed in this paper combining these algorithms brings new perspective to solve this kind of problem. In this paper a multi-agent approach based on A-Team that combines TS and GA and other specific agents is presented to solve the JSP. Results for benchmark problems from the literature are presented aiming to demonstrate the applicability of the proposal.