Aircraft sensor fault detection based on unknown input observers

Abstract

Purpose - The purpose of this paper is to generate residuals which can be used to detect fault and isolate on a vertical takeoff and landing (VIOL) aircraft dynamic model.

Design/methodology/approach - In the proposed approach, a generalized observer scheme method based on an unknown input observer is used for residual generation and applied to detect and isolate a faulty sensor. A bank of robust unknown input observers estimates the state variables of the system and gathers necessary information for fault detection and isolation purposes.

Findings - A sinus signal is considered as a non-linear disturbance in simulations. A failure simulation was prepared in different times. In this situation an unknown input observer should be designed which could predict the states of the system against the disturbances or unknown inputs. In the real world, there exist unknown inputs such as system non-linearities, noise and disturbances. The paper shows that the system based on UIO is robust for unknown inputs mentioned above.

Originality/value - It is simulated on a VTOL dynamic model using MATLAB/Simulink. Any single sensor fault could be detected and isolated correctly. This kind of observer is also robust and flexible.