On-line Fault Detection, Diagnosis, Isolation and Accommodation of Dynamical Systems with Actuator Failures

Abstract

A general framework for the on-line detection, diagnosis, isolation and accommodation for a class of actuator faults in dynamical systems is developed. This framework, which addresses both abrupt and incipient faults, utilizes a nonlinear adaptive detection observer along with an on-line approximation scheme for failure assessment. The actuator failure is modeled as an additive perturbation of the actuator signal (actuator gain) that may be described by a nonlinear function of both the input and output signals. Robust adaptive schemes are introduced to account for modeling errors that affect the diagnosis process and may cause false alarms. Numerical studies using a neural network-based on-line approximator are presented to demonstrate the applicability of the proposed automated diagnostic scheme.