Fault Detection: A Control Philosophy Approach to Design via Constrained Optimisation

Abstract As modern control brings systems nearer their performance limits, the dependence of the system on the controller increases- not only in terms of performance, but also in the avoidance of catastrophic failure. Prevention and detection of controller system failure thus become important issues. An obvious solution is to rely on hardware redundancy in the duplication of both actuators and sensors. A more economic approach, however, may be to exploit the analytical redundancy inherent in the system, in the use of fault detection methods. We present several methods of fault detection filter design by extending tools from the theoretical control literature. The resultant designs are extremely simple to implement.