遥感卫星任务智能决策的机器学习方法研究

Research on Machine Learning Method for Intelligent Decision-making of Remote Sensing Satellite Mission

基于遥感卫星任务决策的特点,研究如何采用机器学习方法对执行任务时产生的大量动作、指令和遥测数据进行分析和训练。为了给遥感卫星任务建立机器学习方法,探索机器学习辅助遥感卫星任务智能决策的可行性,并探讨机器学习模型对卫星任务数据的适应性和处理效率。借鉴地面相关人工智能系统成熟的机器学习架构,研究建立遥感卫星任务相关智能决策的机器学习方法,并给出了机器学习的样例。研究结果表明:机器学习方法的适应性很强,初步实现了遥感卫星自主任务决策,并达到一定的准确率,对卫星任务智能决策技术进行了有益探索。

Based on the characteristics of remote sensing satellite mission decision-making,the machine learning method is studied to analyze and train a large amount of actions,commands and telemetry data generated during mission execution.In order to establish a machine learning method for remote sensing satellite mission,the feasibility of machine learning assisted intelligent mission decision-making of remote sensing satellite is explored,and the adaptability and processing efficiency of machine learning methods for satellite mission data are also discussed.The mature machine learning framework of ground-related artificial intelligence systems is used for reference to research the establishment of machine learning method for intelligent decision-making of remote sensing satellite mission.An example of machine learning for remote sensing satellite mission is provided in this paper.The research result shows that machine learning methods have strong adaptability and have initially achieved satellite intelligent mission decision-making with certain level of accuracy,which is a beneficial exploration for intelligent decision-making technology of satellite mission.