Abstract This paper presents the theory of and design algorithms for a method detecting changes in the system transfer matrix (A) and distinguishing these from actuator failure/change in a dynamic system. This method utilizes optimal input to the failed/changed system to isolate hypothesized changes in A from other changes/failures. The change in the A matrix (i.e., ΔA) is related to hypothesized system failures or changes in operating point for a linearized nonlinear system. Once a failure is detected (i.e., via state error residuals from a state estimator), the input forces the system into the null space of A4 after which isolation is straightforward. This paper presents two theorems; one for the existence of solution for the optimal input and for the design freedom and the other for the actual solution in terms of feedback gain and user input freedom. In addition, design algorithms for the method are derived from the two thporems and from two additional lemmas. A specific example is given in which changes in A are detected and isolated from actuator failures for an advanced fighter aircraft.
[1]
George C. Verghese,et al.
Optimally robust redundancy relations for failure detection in uncertain systems
,
1986,
Autom..
[2]
A. Willsky,et al.
Analytical redundancy and the design of robust failure detection systems
,
1984
.
[3]
L. Daniel Metz,et al.
Controllability and Stability Aspects of Actively Controlled 4WS Vehicles
,
1989
.
[4]
Robert L. Kosut,et al.
Robust Fault Detection: The Effect of Model Error
,
1984,
1984 American Control Conference.
[5]
P. Frank,et al.
Fault detection via optimally robust detection filters
,
1989,
Proceedings of the 28th IEEE Conference on Decision and Control,.
[6]
M. M. Akhter,et al.
Effect of model uncertainty on failure detection: the threshold selector
,
1988
.
[7]
W. Ge,et al.
Detection of faulty components via robust observation
,
1988
.