基于神经网络与自适应控制的系统重构方法

针对可重构飞行控制系统提出一种基于神经网络的抗扰动方法,其核心是运用神经网络的函数逼近能力去抑制非线性形式且动态特征未知的扰动。同时提出一种基于输入误差的自适应模型跟踪控制算法,可以保证系统在发生故障或局部损坏时的安全性和稳定性。将神经网络与自适应控制结合,从而降低扰动对故障检测和系统重构的影响,进而有效的降低故障误报率,提高系统的安全性。通过一个具体航空系统的仿真结果证实了该方法的有效性。

A Method of System Reconfiguration Based on Neural Network and Adaptive Control

A new method of disturbance rejection by using neural network is proposed in this paper.Emphasis is placed on utilization of neural networks(NN) to reject disturbance in nonlinear form when the characteristic of disturbance is unknown.Adaptive algorithm based on direct input error is developed for reconfiguration control system,which can guarantee the safety and stability of the aircraft system under conditions of fault or local damage.Combination of NN and adaptive control can effectively decrease the disturbance impact on fault detection and reconfiguration of flight system.Simulation results prove the performance of proposed algorithm.