Projection operations for failure detection and isolation

The problem of failure detection and isolation in linear systems is discussed. Typically the solutions to this problem have included a statistical test being performed on the measurement residual generated by a Kalman filter. These solutions are usually based on one or more of the assumptions: the type of failure is known, or the failure location (in sensor or state) is known, so that the failure can be estimated. In this paper a technique is developed which is based on a set of projection operators which, under appropriate conditions, can both detect and isolate the failures when the type, location, time of occurrence and magnitude of the failure are not known a priori.

[1]  C. Chang,et al.  A recursive generalized likelihood ratio test algorithm for detecting sudden changes in linear, discrete systems , 1978, 1978 IEEE Conference on Decision and Control including the 17th Symposium on Adaptive Processes.

[2]  R. Clark,et al.  Sensor Fault Detection in a System with Random Disturbances , 1980, IEEE Transactions on Aerospace and Electronic Systems.

[3]  On detecting sudden changes (failures) of biases in linear dynamic systems , 1980, 1980 19th IEEE Conference on Decision and Control including the Symposium on Adaptive Processes.