Modeling Automotive Assembly Lines with Generalized Stochastic Petri Nets and Markov Decision Processes with Imprecise Probabilities

This paper proposes a methodology for automotive manufacturing lines scheduling. This methodology is based on generalized stochastic Petri Nets and Markov decision processes with imprecise probabilities. The usual generalized stochastic Petri Nets is extended by allowing imprecision about probabilities to be explicitly represented and by human task time graph of different products to be attached. Once the system is modeled using this tool and its extensions, we translate the resulting models into Markov decision processes with imprecise probabilities, in order to compute optimal policies that will result in the line scheduling. This paper introduces an algorithm that performs this translation.