Fault Detection Algorithm of INS of Parametrically Uncertain Unmanned Vessel

Dmitry Bazylev, Alexey A. Margun, Dmitrii Dobriborsci

Abstract: This paper is addressed to the problem of fault detection and localization of an inertial navigation system (INS) of an unmanned surface vessel (USV) with unknown parameters. The INS under consideration includes angular and linear velocity measurement units. The proposed approach is based on full-order Luenberger observers that are used to syn-thesize directional generators of residual signals. To estimate the unknown parameters of the USV a dynamic regressor extension and mixing (DREM) technique is applied. Developed algorithm ensures finite-time convergence of parameter estimation errors to zero under some reasonable assumptions. Simulation results demonstrate the efficiency and performance of the developed algorithm for correct fault detection of each sensor.