机器学习在卫星遥测分析建模中的应用

Application of Machine Learning in Satellite Telemetry

摘要:

卫星在研制和应用的各个环节产生了庞大的、特征和相关性复杂的遥测数据，人工判读或使用专家策略进行分析均无法充分梳理出这些数据的相关性并进一步挖掘和应用它们的价值。利用Python强大的机器学习生态环境和算法，能够实现对卫星遥测数据的分析和建模。以星载铷钟遥测数据为样本，提出了对卫星遥测数据进行算法分析和建立模型的方法，为实现对铷钟的健康状态进行实时评估、预报、精准控制奠定了基础。除遥测数据外，该方法亦可用于分析卫星其它数据，为卫星研制和应用过程中利用机器学习技术提高智化能水平提供了参考。

Abstract:

All kinds of data are generated during satellite development, production and in-orbit flights, how to deal with it is a big problem. Due to the large amount of data, diverse data types, and complex characteristics of the correlation，it’s almost impossible to sort out the correlation and explore the value of these data by using of manual analysis or expert strategy. A method for analyzing, algorithmic evaluation and modeling of rubidium clock’s telemetry data by using Python is presented. This model can be used to predict the state of the rubidium clock, and to enable satellite to control the rubidium clock intelligently. This method can also be used to process data from other devices on satellite, and it provides a reference for the use of machine learning in the design and manufacture of satellites.