An adaptive filter based fault detector applied to dynamic positioning system

The paper introduces a fault detection method based on adaptive filtering of the sensor/actuator signals and shows how it has been applied in the simulated dynamic positioning system of a ship. Detection of faults is done by observing changes of statistical parameters in the innovation signals of the filter. Localization is based on a rule logic which takes into account the specific features of the physical system. Faults both in the sensor and actuator components can be detected and localized. The actuators form the thruster system of the ship which maintain controlability in the case one of the thruster faults. It is also demonstrated in the paper how post fault situation can be handled by making a proper change in the control policy.