Adaptive Monitoring and Accommodation of Nonlinear Actuator Faults in Positive Real Infinite Dimensional Systems

• Published in: IEEE transactions on automatic control Vol. 52; no. 12; pp. 2332 - 2338
• Main Authors: Demetriou, M.A., Ito, K., Smith, R.C.
• Format: Journal Article
• Language: English
• Published: New York, NY IEEE 01.12.2007; Institute of Electrical and Electronics Engineers; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Accommodation; Actuator faults; Actuators; Adaptive signal detection; Applied sciences; Automatic control; Computer science; control theory; systems; Control theory. Systems; distributed parameter systems; Exact sciences and technology; fault accommodation; Fault detection; Faults; Hydraulic actuators; Indium tin oxide; Inverse; Modelling and identification; Monitoring; nonlinear actuator dynamics; Nonlinear dynamical systems; Nonlinearity; Observers; Reconfiguration; Robustness; Signal generators; Stability; System theory; Positive system; Modeling; Time varying system; fault accommodation; Unmodeled dynamics; Surveillance; Alarm; Actuator faults; nonlinear actuator dynamics; distributed parameter systems; Infinite dimension; Adaptive observer; Distributed parameter system; Monitoring; Actuator; Reconfiguration
• ISSN: 0018-9286; 1558-2523
• DOI: 10.1109/TAC.2007.910694

We consider a class of positive real infinite dimensional systems which are subjected to incipient actuator faults. The actuator fault is modeled as a time varying transition from an initial (linear or even nonlinear) map into another unknown nonlinear map at the onset of the fault occurrence. An infinite dimensional adaptive detection observer is utilized to generate a residual signal in order to detect the fault occurrence and to assist in the fault accommodation. This is done via an automated control reconfiguration which utilizes information on the new actuator map and adjusts the controller via a right inverse of the new actuator map. A robust modification is utilized in order to avoid false alarms caused by unmodeled dynamics. An example is included to illustrate the applicability of the proposed detection scheme.