Comparison of ant colony optimization with simulated annealing for
Transcription
Comparison of ant colony optimization with simulated annealing for
ArcelorMittal Gent Vlak-Koolstofstaal Comparison of ant colony optimization with simulated annealing for line scheduling of finishing lines Master thesis – Academic year 2015 – 2016 Brief description ArcelorMittal Gent (formerly known as Sidmar) is a steel plant situated in the port of Ghent. It produces flat steel products, used amongst others in the automotive industry. Starting from raw materials, the production process consists of a chain of treatments and manipulations, performed by different installations. Each installation has an inventory available of about 1000 steel coils. Starting from this inventory, a planner has to prepare a feasible schedule of approximately 200 coils, which he must select and sequence taking into account the urgencies of the coils, as well as a set of constraints determined by the installation. Due to the size of the problem, mathematical models are needed to find a way through the cost landscape of the solution space, as quickly as possible, in search of the optimal solution. Recently, a new heuristic was implemented for a couple of installations, which performs both selection and sequencing of the coils in one step. Previously, it was only possible to optimize the sequence starting from a manual selection of material. The new heuristic is based on iterated local search and simulated annealing, but these techniques pose some drawbacks. The result very much depends on certain parameters that need to be determined over and over again, when a constraint or cost changes. In this master thesis, you will apply an alternative technique to this selection and sequencing problem, being ant colony optimization. Ant colony optimization is well described in literature, but has never been tested for such problem. You will get the opportunity to get hands-on experience in a state-of-the-art steel plant, work together with industry specialists in optimizing techniques and contribute to technological innovation. Additional info Number of students = 1 The master thesis will be conducted at the Systems & Models department, a team of 30+ young engineers, whose mission consists of optimizing all kinds of production processes and logistic problems by means of mathematical techniques. For additional info, contact the thesis supervisor, Ward Bijttebier. v v 09/347.31.81 ward.bijttebier@arcelormittal.com www.arcelormittal.com/gent Correspondentieadres: ArcelorMittal Gent John Kennedylaan 51 B - 9042 Gent T +32 (0)9 347 31 11 F +32 (0)9 347 49 07 www.arcelormittal.com/gent Zeteladres: ArcelorMittal Belgium NV Keizerinlaan 66 B - 1000 Brussel BTW BE 0400.106.291 - RPR Brussel