基于数据驱动的空间目标智能识别

Space Objects Intelligent Identification Based on the Data-Driven Method

随着太空领域的不断发展,空间目标数量不断增加,空间态势日益复杂,亟需改进空间目标识别技术以提高空间态势感知能力.设计了基于数据驱动的空间目标智能识别体系架构,分别使用梯度提升决策树和卷积神经网络两种机器学习算法,利用海量空间目标特性数据,基于数据驱动方法构建空间目标智能识别模型.实验结果显示,两种模型的识别准确率均达到90%以上,能够为空间目标智能识别提供有效解决方案.

With the continuous development of the space exploration,the number of space objects is increasing and the space situation is becoming complex.It is imperative to improve the space objects identification technology to enhance the space situational awareness.The space objects intelligent identification architecture based on data-driven is designed,and two kinds of machine learning algorithms,Gradient Boosting Decision Tree(GBDT) and Convolutional Neural Network(CNN),are utilized respectively to construct identification models by utilizing massive characteristics data.The experimental results show that the recognition accuracy of the two models reaches more than 90 %,which can provide an e?ective solution for intelligent identification of space objects.