Attitude control of variable swept-wing aircraft: A novel composite control strategy

Abstract

To address the complex issues of discontinuous disturbances, parameter uncertainties, actuator deadzone, and saturation nonlinearity in Variable Swept-Wing Near Space Vehicles (NSV), an attitude controller combining reinforcement learning and adaptive switching sliding mode control is proposed, along with an adaptive threshold event-triggered mechanism to reduce the actuator executing frequency. Firstly, the motion characteristics of the Variable Swept-Wing NSV across the full range of operating modes are modeled as a nonlinear switched system. Secondly, a nonlinear switched disturbance observer is employed to estimate the composite disturbances caused by discontinuous disturbances and parameter uncertainties. By introducing a deadzone right inverse function and designing an auxiliary system, the composite nonlinearity of the actuator are effectively addressed. An adaptive multi-modal switching sliding mode controller is then proposed based on the backstepping method to achieve basic control. Subsequently, considering the higher dimensionality of aerodynamic control surfaces and the increased complexity of aerodynamic characteristics in the subsonic mode, which imposes stricter control requirements, a reinforcement learning-based controller is designed. Leveraging the self-learning and optimization capabilities of reinforcement learning, which does not rely on an accurate model, the controller achieves end-to-end control of the horizontal canard. Finally, an event-triggered mechanism with an adaptively varying threshold is also developed. The multi-Lyapunov stability theory and the average dwell-time theory are employed to guarantee the stability of the closed-loop nonlinear switched system while excluding the undesired Zeno behavior. Simulations and comparative experiments demonstrate that the proposed method achieves superior tracking accuracy and control performance, while the adaptive threshold event-triggered mechanism effectively reduces data transmission.