Diagnostic goal driven modelling and simulation of multiscale process systems

Most industrial and manufacturing processes are inherently multiscale systems. The systematic way of constructing multiscale models, however, is still an open research area. It is well known [1] that a process model is jointly determined by the process system and the modelling goal. Little is known, however, about a systematic way of taking the effect of the modelling goal into account during the modelling process [2, 3]. The aim of this paper is to propose a modelling goal-driven methodology for developing multiscale process models and show how this model can be used for diagnosis. Because of the novelty of the field of multiscale modelling , we restrict our attention to the case when the modelling goal is to obtain a model that is capable of performing prediction-based diagnosis. Diagnosis requires the use of a dynamic process model, where both uncertainty and empirical information can be described. Therefore, coloured Petri nets (CPNs) [4] have been selected as a unifying modelling tool for describing hierarchically integrated multiscale models.