Optimization of Discrete Event Systems Using Extended Finite Automata and Mixed-Integer Nonlinear Programming

This paper presents a concept for converting a discrete event model, modeled with Extended Finite Automata (EFA), to mixed-integer linear constraints. The conversion handles the structure of modular EFAs, synchronization of EFAs using shared events and EFA execution order due to logical transition conditions. The paper also presents methods to reduce the number of variables and constraints by automatically analyzing the EFA model and the resulting problem formulation. An example of this is the special case of transition conditions used to model mutual exclusion of shared resources, where the conversion results in a significantly reduced problem formulation. The objective function is then built by summarizing weighted state cost functions and the result is a Mixed-Integer Nonlinear Programming problem. The main contribution of this paper is hence the combination of the simplicity in modeling a system with EFAs and an efficient formulation of the optimization problem that can be solved by standard optimization software.