A two phase optimization method for Petri net models of manufacturing systems

Optimization is a key issue in the design of large manufacturing systems. An adequate modeling formalism to express the intricate interleaving of competition and cooperation relationships is needed first. Moreover, robust and efficient optimization techniques are necessary. This paper presents an integrated tool for the automated optimization of DEDS, with application to manufacturing systems. After a very quick overview of optimization problems in manufacturing systems, it presents the integration of two existing tools for the modeling and evaluation with Petri nets and a general-purpose optimization package based on simulated annealing. The consideration of a cache and a two phase technique for optimization allows us to speed-up the optimization by a factor of about 35. During the first preoptimization phase, a rough approximation of the optimal parameter set is computed based on performance bounds. Two application examples show the benefits of the proposed technique.