Enhancing Aircraft Fault-Tolerant Control: Hybrid Fusion of Fast Convergence Super-Twisting Observer and Adaptive Neural Network Approaches

Robust performance in flight control systems requires minimizing the impact of nonlinear dynamics, sensor faults, signal loss, and external disturbances. To address these challenges, conventional architectures rely on high-reliability systems supported by human intervention to detect and manage errors that can critically impair aircraft operation. Fault-tolerant control strategies integrate multiple layers of estimation and compensation to manage various fault types. This study proposes a three-stage methodology to mitigate sensor faults in the presence of disturbances. In the first stage, a fast-converging super-twisting sliding mode observer is employed to eliminate disturbances and noise from the outputs of angular velocity sensors. The second stage applies an adaptive neural observer for fault detection and isolation (FDI). In the third stage, a control system based on the backstepping method generates smooth control commands to compensate for the identified faults. The control strategy dynamically adapts based on the fault type and location-whether in sensors, actuators, or other components-ensuring consistent fault mitigation. Simulation results using a nonlinear six-degree-of-freedom F-18 aircraft model validate the effectiveness of the proposed algorithm in simultaneously detecting and compensating for disturbances and faults.