Knowledge based modular networks for process modelling and control

This payer addresses methods of knowledge engineering for chemical and bioc hemical process modelling and control. The concept of knowledge Based Modul ar (KBM) Networks is presented. KBM networks represent a method of expressi ng and combining different types of knowledge usually available for modelli ng chemical and biochemical processes: mechanistic, heuristic and knowledge hidden in process data records, The Expectation Maximisation (EM) algorith m is employed to optimally combine the modules within the KBM network. The concepts are illustrated with an application to a baker's yeast production process. The results show that it is possible to obtain more accurate proce ss description when all available sources of knowledge are incorporated in the process model. (C) 2001 Elsevier Science Ltd. All rights reserved.