摘要： 在多个固定翼无人机姿态主从式一致性控制过程中，给出单个固定翼无人机在理想情况下的姿态动力学模型，即名义模型。考虑到无人机在实际运行过程中存在的外部干扰、状态测量误差、控制器微小故障以及无人机实际模型与名义模型之间的偏移，提出一种基于观测器和神经网络的故障检测方法，以实时检测出无人机中存在的故障、模型不确定以及干扰情况。基于无人机名义模型和检测出的故障及干扰，设计主从式多无人机姿态一致性控制器，以实现多无人机姿态的一致性准确跟踪。仿真结果表明，在外部干扰、状态测量误差与控制器微小故障下，与基于神经网络的直接姿态一致性控制器相比，该控制器能够使得无人机的姿态运动状态更接近于期望状态。

关键词: 状态测量误差, 故障检测, 观测器, 神经网络, 主从式一致性控制器

Abstract: In the process of master-slave consistency attitude control of multiple fixed wing Unmanned Aerial Vehicles (UAVs),the attitude dynamics model of a single fixed wing UAV under ideal condition is given,which is also called the nominal model.Considering the external disturbance,state measurement errors,micro faults of the controller and the deviation between the actual model and nominal model of UAV,this paper proposes a fault detection method based on the observer and neural network to detect the faults,model uncertainty and disturbance of UAV in real time.Based on the nominal model of UAV and the detected faults and disturbances,a master-slave attitude consistency controller of UAV is designed to achieve accurate attitude consistency tracking of multiple UAVs.The simulation results show that compared with the direct attitude consistency controller based on neural network,this controller can make the UAV attitude motion state closer to the desired state under the external interference,state measurement errors and minor faults of the controller.

Key words: state measurement error, fault detection, observer, neural network, master-slave consistency controller

中图分类号: 

• TP273