一种卫星遥测参数趋势异常的自动识别方法  

An Autonomous Identification Method of Telemetry Parameters Trend Anomaly

作　　者：王佳伟[1] 秦巍[1] 邵坤[1] 刘超 WANG Jiawei;QIN Wei;SHAO Kun;LIU Chao(Beijing Institute of Spacecraft System Engineering,Beijing 100094,China)

机构地区：[1]北京空间飞行器总体设计部,北京100094

出　　处：《航天器工程》2021年第4期62-68,共7页Spacecraft Engineering

摘　　要：针对卫星遥测参数变化趋势人工监视准确度低、成本高的问题,在基于小波的频域转换方法基础上,提出了一种基于神经网络的趋势异常的自动学习和判别方法,即通过建立遥测参数趋势变化的频域特征向量模型,利用多层前馈(BP)神经网络方法,实现遥测趋势变化的自动判别和定位,能够有效提高卫星在轨监视的精细化程度和准确度。通过某卫星红外敏感器和太阳电池阵电流遥测趋势识别对方法的有效性进行了验证,结果表明:该方法能够满足自动监视要求,可为卫星在轨管理提供参考。Regarding to the low accuracy and high cost of manual monitoring to the changing trend of satellite telemetry parameters,this paper proposes an autonomous learning and distinguishing method of trend anomaly based on neural network.By establishing frequency-domain eigenvector model,autonomous distinguishing and positioning of telemetry changing trend is fulfilled using BP neural network method.This method can effectively improve the precision and accuracy of satellite on-orbit monitoring.In this paper,the effectiveness of the method is verified through satellite infrared sensor and solar array current telemetry trend recognition.The results show that the method can meet the requirements of autonomous monitoring,providing reference for satellite on-orbit management.

关 键 词：卫星遥测参数 频域分析 神经网络 自动识别 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术] V557[自动化与计算机技术—计算机科学与技术]