基于神经网络观测器的飞控传感器在线故障诊断

为提高无人机飞行安全可靠性,针对飞行控制系统中常出现的传感器故障以及非线性气动力模型参数难以确定的问题,提出了基于BP神经网络观测器估计的故障诊断方法;引用LM改进算法对网络参数进行调整,构造了神经网络观测器模型逼近非线性系统,并运用于飞行控制系统进行在线数字仿真,对垂直陀螺输出卡死故障、恒偏差故障和恒增益故障分别进行仿真分析;仿真结果表明,所设计神经网络观测器可以有效估计系统输出,在线诊断传感器故障。

Flight Control System Sensor Online Fault Diagnosis Based on Neural Network Observer

In order to improve the safety and reliability of UAV,aiming at the common sensor fault in flight control system and parameter uncertainty for nonlinear aerodynamic model,a method for online fault diagnosis based on BP neural network is proposed.The LM algorithm is cited to adjust the network parameters,and a neural network observer is constructed to approximate nonlinear model,which applies to flight control simulation,and analyzed the stuck fault,constant deviation fault and constant gain fault of vertical gyro.Simulation result shows that,the sensor output is effectively estimated by the neural network observer,and fault diagnosis is implemented online.