A stochastic fault tolerant control architecture for morphing hexarotors under motor failure

Creative Commons License

Köse O., Oktay T.

JOURNAL OF THE BRAZILIAN SOCIETY OF MECHANICAL SCIENCES AND ENGINEERING, cilt.48, sa.1, ss.934-955, 2026 (SCI-Expanded, Scopus)

Özet

Motor failures typically result in inevitable crashes for unmanned aerial vehicles (UAVs). The main contribution of this

study to the literature is that it presents a novel Fault-Tolerant Control (FTC) architecture that reconfigures the system

after a motor failure not only at the controller level but also at the physical level through a differential morphing strategy. The developed hybrid approach, while estimating the continuously varying moments of inertia resulting from active

morphing in real time using a Deep Artificial Neural Network (DNN) trained on noisy data, simultaneously optimizes

both the independent arm lengths and the proportional-integral-derivative (PID) gains under a single cost function using

the Simultaneous Perturbation Stochastic Approximation (SPSA) algorithm. SPSA exhibits a high adaptation speed by

optimizing these 12 critical parameters, which determine the system's stability, in just 2.9560 s. Simulation results show

that the proposed method achieves a 49.95% improvement in the total cost function and reduces the rise time in the lateral

flight mode, which is most affected by the fault, by 90.36%. This study provides concrete evidence in a simulation environment that integrating physical morphing capability with AI-assisted stochastic optimization enables UAVs to maintain

their stable equilibrium even in severe fault scenarios.