深度森林算法在高速飞行器智能控制中的 应用研究

To address the attitude control challenges of air-breathing high-speed vehicles operating across large airspace and wide velocity ranges, an intelligent control optimization method based on the deep forest algorithm is proposed. Firstly, a robust uniformly convergent observer is employed to accurately estimate system uncertainty disturbances, which are subsequently suppressed through a dynamic compensation mechanism. Secondly, a robust controller with preset performance constraints is designed by integrating prescribed performance control and sliding mode control algorithms. Thirdly, a strategy database of optimal performance indices is constructed using the deep forest algorithm, enabling autonomous query of optimal control parameters across all flight conditions. Meanwhile, a multi-modal training sample generation approach combining the NSGA- Ⅲ multi-objective optimization algorithm and coefficient of variation method is developed to ensure dataset diversity and comprehensive flight envelope coverage. Finally, an intelligent controller with autonomous parameter decision-making capability is designed by integrating the aforementioned control algorithms and strategy database. Through real-time flight condition-triggered database query mechanisms, optimal control parameters are matched instantaneously, demonstrating superior control accuracy, enhanced robustness, and faster dynamic response during large airspace and wide-speed-range operations. Numerical simulations validate the strong robustness of the proposed method under model parameter perturbations and complex aerodynamic disturbances.