Neural observer-based approach to fault-tolerant control of a three-tank system

Abstract

The simulated laboratory set-up Three-Tank System is investigated from the standpoint of fault-tolerant control. Different neural observer schemes are used for symptom generation. Symptom evaluation is based on neural classifiers. The output of the classification stage is provided to a supervisory unit. This switches on the appropriate remedy action in case of a process fault. For actuator and component faults, the reconfiguration strategies are based on the redundant hardware and optimised control laws. The reconfiguration in case of sensor faults is based on neural observers. The outputs of the latter replace the faulty sensor(s).