A Data-Driven Quantitative Stability Monitoring Approach for Quadrotor Blade Damage

To quantitatively assess the maximum impact of blade damage on quadrotor flight stability, this article proposes a data-driven stability monitoring approach for fault diagnosis (FD) based on the closed-loop stability margin. A stable kernel/image representation of data-driven closed-loop systems is first developed using subspace identification methods, enabling the computation of a data-driven closed-loop stability margin. To support real-time monitoring, an online estimation algorithm is presented, leveraging sliding window techniques combined with efficient updating and downdating of lower triangular and orthogonal matrix (LQ) decomposition through Givens and hyperbolic transformations. This enables dynamic tracking of quadrotor stability under varying degrees of blade damage. Furthermore, a test statistic aligned with stability analysis is constructed to provide a rigorous decision basis for FD related to blade damage. The effectiveness of the proposed method is validated through a series of experimental tests on quadrotor blade damage scenarios.