Output observer for fault detection in linear systems

Fault detection (FD) is the process of monitoring a system to identify any malfunction occurring in the system. In model based fault detection, a state estimator or observer is usually designed in order to provide an estimated output which is compared with the measured output and evaluated with a decision making algorithm to identify a fault. In this paper, a new approach for fault detection is proposed based on the design of an output observer, instead of a state observer, using an input/output system representation. The proposed output observer is employed to estimate the output of the system only in order to generate a set of residuals that are indicative of the presence of a fault. The convergence of the observer is proven for any initial condition and a fault detectability condition is set. This FD scheme is validated by simulating a model of a three wheels robot.