飞控系统传感器故障诊断的在线方法研究

Research for Online Sensor Fault Diagnosis for Flight Control System

传感器是飞行控制系统当中的一个重要组成部分,在系统中往往利用传感器的各个输出来建立飞机的动态状态;因此,实时准确的对传感器进行故障检测和识别可有效地提高系统的安全可靠性;提出一种带有可变遗忘因子的BP神经网络在线递推学习算法,应用改进的算法对飞行控制系统的传感器故障进行实时在线的检测和识别,且利用神经网络的输出对系统进行重构;仿真结果表明提出的方法可准确的对传感器的故障进行故障诊断和容错控制。

Sensor is an important component in the flight control system, the outputs of the sensors are used to establish the dynamic state of the aircraft, so the system security and reliability can be enhanced effectively by utilizing the accurate and real--time sensor failure detection and identification. An extended BP neural network is proposed in this paper, which uses online recursive algorithm with forgetting factor （RFF） to train the weights. The improved algorithm is applied to identify the real--time sensor fault in the flight control system, and using the outputs of the neural networks reconfigure the system. The simulation results show that the sensor fault ean be identified accurately and reconfigured by using the improved method.