Solving the job shop scheduling problem using evolutionary algorithms

Abstract

Scheduling has received considerable importance in planning processes over the years, while several models have been developed that aim to accurately represent this type of problem. In this dissertation, we consider the Job Shop Scheduling Problem (JSSP), which is one of the most intractable combinatorial optimisation problems considered so far in the area of Operations Research. It is well known that, during the last decades, substantial progress has been done in this problem’s solution approaches. Naturally-inspired evolutionary algorithms are common solution approaches for solving combinatorial optimisation problems such as the JSSP. In this dissertation, we will study in detail the most important features of the JSSP and we will extensively discuss its mathematical formulation. Since exact solution methods fail to provide good-enough solutions in a reasonable amount of time, our aim in this study is to test the performance of several metaheuristic techniques for the solution of the JSSP. More specifically, three evolutionary algorithms, namely the Genetic Algorithm, the Particle Swarm Optimisation algorithm, and the Firefly Algorithm, have been thoroughly discussed and applied to obtain a solution to the JSSP. The performance of these methods is evaluated on a set of synthetic data that has been generated for the aims of this study. Comparisons with the results obtained from the application of an exact method, namely the Branch-and-Cut algorithm, are also provided.