基于自适应神经网络的飞行器主动容错控制

针对飞行器非线性系统执行器故障,利用RBF神经网络和自适应控制律,提出了基于自适应神经网络的故障重构和容错控制方法。设计了自适应神经网络观测器,利用神经网络逼近故障,引入调节因子,设计自适应律以在线调整神经网络权重向量和中心向量。构造自适应神经网络控制器,结合神经网络设计补偿控制输入。利用Lyapunov稳定性定理证明了所提方法可以实现系统渐近稳定。仿真实验结果验证了所提的方法对故障系统具有良好的观测性能、控制精度和响应速度。

Active Fault Tolerant Control of Aircraft Based on Adaptive Neural Networks

For linear systems with nonlinear terms and faults, a fault diagnosis and fault-tolerant control method based on adaptive neural networks is proposed by using RBF neural networks and adaptive laws. First, an adaptive neural network observer is designed to approximate the fault, and an adaptive law is designed to adjust the neural network weight vector and center vector online with a regulator introduced. Second, the adaptive neural network controller was designed, and the compensating control input is designed in conjunction with the neural network. Then, the proposed method is proved to achieve asymptotic stability of the system using the Lyapunov stability theorem. Finally, the simulation experimental results verify that the proposed method has good observation performance, control accuracy, and response speed for the faulty system.