Gradient-Based Performance Optimization for Flight Control System With Real-Time Data

A flight control system acts as a core subsystem in aircraft for the purpose of attitude control and trajectory tracking. Due to the inaccurate system modeling and the internal/external disturbances (e.g., wind disturbance, aerodynamic parameter change, and load perturbation), the performance of controller designed for the typical operating points may be not optimal for the full flight envelope. This article aims to address the control performance optimization problem of flight control system in the presence of an incompletely known aircraft model and real-time flight data. First, the longitudinal aircraft dynamic model is formulated, followed by the establishment of two typical disturbance models, including periodic and constant disturbances. Subsequently, a real-time optimization framework for enhancing flight control performance is introduced, with the aid of residual-driven realization of Youla parameterized controller. Moreover, the gradient-based performance optimization strategy is proposed to mitigate the performance degradation induced by disturbances, using the online monitored actuator and sensor data of flight control system. By developing tools from optimization theory, the convergence and optimality of gradient-based performance optimization method are analyzed comprehensively. Finally, the proposed methods are testified on the model of an aircraft.