摘要
为了有效地利用卫星下传的海量遥测数据,在测试过程中对卫星进行实时的故障诊断,提出了一种基于BP神经网络的卫星故障诊断方法;该方法包括离线自主学习和实时在线故障诊断两部分;离线自主学习部分基于历史数据库和更新样本进行自主学习,学习获得神经网络模型存储于知识库;实时在线故障诊断部分依据相应的神经网络模型,对遥测数据进行实时在线的诊断;为了验证基于BP神经网络的卫星故障诊断方法的有效性和优越性,以现有型号三轴稳定近地卫星控制分系统为实验对象,利用该方法对具有代表性的红外地球敏感器和动量轮的相关遥测数据进行分析;通过将该方法的实验结果与基于Kalman滤波的方法的实验结果进行对比分析,表明该方法能够有效地对卫星的故障进行诊断。
In order to effectively use of the massive remote sensing data transmitted from satellites,in the process of testing for satellite real-time fault diagnosis,a method of satellite fault diagnosis based on BP neural network is proposed.The method includes offline autonomous learning and the real-time online fault diagnosis.The offline autonomous learning part automatically learns based on historical database and the updated samples,learning for neural network model is stored in the knowledge base.The real-time on-line fault diagnosis part is for the diagnosis of remote sensing data in real time online,based on the corresponding neural network model.To verify the method of satellite fault diagnosis based on BP neural network is effective and superior,with the control subsystem of the three axis stabilized near earth satellite as experimental object,the method is used to analysis the typical remote sensing data of infrared earth sensors and momentum wheel.By the experimental results analysis of this method and the method based on Kalman filtering,the experimental results show that the method is effective to satellite fault diagnosis.
出处
《计算机测量与控制》
2016年第5期63-66,共4页
Computer Measurement &Control
关键词
卫星
BP神经网络
故障诊断
satellite
BP neural network
fault diagnosis
