A Two Phase Optimisation Strategy for Deds: Application to a Manufacturing System

Model-based optimisation can make the design of complex DEDS more efficient. The process of optimising a manufacturing system is considered in this paper as an application, where the main problem lies in the computational effort required for a series of long simulation runs. We propose a two-phase optimisation method starting with a fast preoptimisation. This first step is done by computing rough approximations based on interpolation of upper and lower bounds of the performance indexes, instead of long run simulations (or analytical computations). Petri nets are used for the modelling of these systems, enabling the application of linear programming techniques for the bounding analysis in polynomial time on the size of the model. The second phase is a post-optimisation, in which every evaluation is conducted by means of simulations. A comparison of the achieved results and computation times with the ones obtained by standard techniques shows the usefulness of the proposed approach.