Data-Driven Fault Detection and Isolation for Quadrotor Using Sparse Identification of Nonlinear Dynamics and Thau Observer

This paper suggests data-driven fault detection and isolation for a quadrotor system using sparse identification of nonlinear dynamics (SINDy) and Thau observer. We propose a novel fault detection method to solve the challenge of a quadrotor with unknown dynamic effect and parameter uncertainty. The SINDy can discover the governing equations of target systems with low data assuming that few functions have the dominant characteristic of the system. Using these properties, system model identification is performed to obtain a nonlinear term that is needed to apply a Thau observer for a quadrotor system. First, the SINDy model is derived by considering a gyroscopic effect and an aerodynamic effect. Second, a SINDy-based Thau observer is proposed to generate a residual that can be used to determine a faulty actuator. Finally, results of the simulation demonstrate that the suggested observer outperforms the Thau observer in detecting faults, even in the presence of uncertain parameters and unknown model effects.