Model predictive control for production scheduling problem in flexible manufacturing system

Production scheduling in flexible manufacturing systems aims at obtaining an operational decision such as machine processing sequence, to achieve a production goal like minimum cost or makespan (i.e. total processing time), given the available resources. This dissertation presents a method to solve such problems based on Petri Net (PN) and Model Predictive Control (MPC). After reviewing the existing methods in the literature, a method to simplify the PN model is proposed, resulting in a more compact PN model with reduced dimensions. Next, there are three model structures provided, each with two types of numerical expressions in the form of state space model. The three model structures are PN with unit production time, PN with non-unit production time, and modular PN. PN with unit production time refers to cases where the production time for each unit is the same, while PN with non-unit production time refers to cases where the production time for each unit may be different, and modular PN uses a systematic way to organise the problem information. The two numerical expressions are model one for simplicity and model two for dimension reduction. Then, based on the modelling method, MPC is capable of working out most kinds of production scheduling problems in the flexible manufacturing system. Finally, a fundamental scheduling example and a real production case are simulated to validate the sufficiency and merits of the proposed framework.Master of Science (Computer Control and Automation