Analytical and knowledge based approaches for a bioprocess supervision

The paper deals with supervision of biotechnological processes. Two main tasks are considered in this supervision scheme. The first one deals with Fault Detection and Isolation (FDI). The objective is to detect changes in the process dynamics using a model-based method. The second task concerns process physiological state recognition. A behavioral model is built using the expert knowledge along with clustering techniques applied to the measured data. The behavioral model under automata form is an heuristic model of the process under supervision.