Global observability and detectability analysis for a class of nonlinear models of biological processes with bad inputs

This paper proposes a new general approach to the global analysis of observability and detectability for nonlinear systems. Based on the definition of indistinguishability it is possible to derive the dynamics of the non-observable part of the system and thus to study its stability properties using methods of nonlinear systems theory. The method is first introduced in general and then applied to a class of nonlinear models for biological processes as e.g. in waste water treatment. Finally detectability conditions for the reactor model are deduced.