Monitoring nonlinear dynamical systems: a combined input-output and local approach

The problem of fault detection and isolation (FDI) for nonlinear dynamical systems in state-space form is addressed through a statistical approach. Standard detection tools turn out to be computationally costly, basically because they require an estimate of the full state. As an alternative, we propose a new approach based on both an input-output representation of the system and a local method to derive the detector. Simulation results on a microbial growth process are provided, which prove the relevance of the detection method.