|Tipo di tesi||Tesi di laurea magistrale|
|Titolo||Modellazione e risoluzione di un problema reale di schedulazione su forni|
|Titolo in inglese||Modeling and solving a real-world furnace scheduling problem|
|Struttura||Dipartimento di Scienze e Metodi dell'Ingegneria|
|Corso di studi||INGEGNERIA GESTIONALE (D.M.270/04)|
|Data inizio appello||2021-04-20|
|Disponibilità||Accessibile via web (tutti i file della tesi sono accessibili)|
Questa tesi presenta i risultati del lavoro svolto con il FZI Forschungszentrum Informatik (Centro di Ricerca per l’Informatica) di Karlsruhe, riguardante la formulazione e la risoluzione di un problema di schedulazione di un'azienda tedesca nel settore dell'automazione.
This thesis presents the results of the work carried out with the FZI Forschungszentrum Informatik (i.e. Research Centre for Information Technology) of Karlsruhe, concerning the formulation and resolution of a scheduling problem of a German company in the automation industry. The problem addressed is known in the literature as Unrelated Parallel Machine Scheduling (UPMS) and the aim of this study is to provide a tool to solve the problem of scheduling gear parts to be hardened in furnaces, with the objective of minimizing the total completion time. This problem is tackled with the development of a Mixed-Integer Linear Programming (MILP) model of the real-world situation. Given the large computation time required by a Branch-and-Bound-based solver to find optimal solutions to the model as the number of parts increases, a heuristic search approach is also presented. In particular, several neighbourhood operators are proposed and evaluated with the aim of finding the most promising one. Both, the mathematical model and the heuristic algorithm, are implemented in Java and the model is solved with IBM ILOG CPLEX Optimizer, a tool for solving mixed-integer optimization problems. The mathematical model and the heuristic approach are tested with instances generated according to the real-world production process. Their computational results are compared to assess the performance of the heuristic approach with respect to the model and in addition to evaluate which of the different proposed operators gives better results. The exact solving process of the mathematical formulation needs to be improved to meet real-world requirements in terms of runtime, but two operators show promising results with regard to near-optimal heuristic solutions.