Observer-based actuator fault detection for chemical batch reactors: a comparison between nonlinear adaptive and H_infty-based approaches

This paper deals with actuator fault diagnosis for chemical batch reactors. Two different nonlinear model-based approaches have been developed in order to detect incipient and abrupt faults. Two residual generators were built by using an adaptive observer and a H-infty approach. Both approaches are well used for sensor fault detection due to their capability to estimate uncertain parameters and to guarantee robustness against modeling errors and disturbances. Here, the approaches have been extended to achieve actuator fault detection and a comparison of both techniques has been done by simulating the whole system in-design and off-design operating conditions.