基于深度神经网络与状态预测器的 无人飞行器自适应控制

In swarm flight scenarios, the widespread presence of unstructured uncertainties can adversely affect the control performance of unmanned aerial vehicles and even pose flight safety risks. An adaptive control method based on deep neural networks and state predictors was proposed to achieve satisfactory trajectory tracking performance in the presence of unstructured uncertainty. This method leverages the feature extraction capabilities of deep neural networks to design feature vectors for unstructured uncertainties,thereby enhancing the uncertainty estimation capability of the control system. The adaptive law was derived based on the nonsmooth Lyapunov stability theory to ensure the stability of deep neural network applications in the control system. Uncertainty was compensated according to the estimated value obtained, resulting in improved trajectory tracking and attitude control performance. Finally, numerical simulations demonstrate that the proposed method improves the trajectory tracking accuracy of unmanned aerial vehicles under the influence of unstructured uncertainties, ensuring the stability and safety of unmanned aerial vehicle swarm flight.