On nonlinear system diagnosis via high-gain observers for bioreactors

We study the diagnosis problem for a bioreactor process. The solvability of the problem is seen as the observability of the variable which models the failure presence. We obtain an algebraic relation that depends on the input, output, and their time derivatives, which is satisfied by the given fault component. The present communication follows an approach of nonlinear diagnosis except that we use a high-gain nonlinear observer based on the multi-output generalized observability canonical form to estimate the fault components in lieu of the numerical differentiation observers.