Model-based sensor fault diagnosis for the Stuttgart SmartShell

The present paper proposes a concept for model-based sensor fault diagnosis and its application to the reference system Stuttgart SmartShell. The approach is based on the well-known Dedicated Observer Scheme and utilizes observable subsystems for residual generation. Particular attention is paid to the residual evaluation in order to account for the effects of noise, model uncertainties and external disturbances. The approach is evaluated in simulation and experiment. The results indicate, that the proposed concept is well suited for the isolation of additive sensor faults despite the presence of moderate disturbances.