Building an automaton for condition monitoring in a biotechnological process

This paper describes a methodology for the design of a supervisory system applied to a biotechnological process. A discrete-event system (DES) is build by the joint participation of the process expert along with clustering and classification techniques applied to measured signals. The automaton becomes a heuristic model of the process under supervision. In the application example, a yeast batch culture, the automaton states are identified as the physiological states of the batch procedure and plant events generate transitions on the DES model, thus reflecting the dynamics of the continuous system underneath.