The first experience in application of artificial neural networks to a study of the dynamic structure of a selected region of the near-Earth orbital space is described. An analysis of time series describing the evolution of the resonant characteristics of the dynamic structure of the region is usually performed manually. However, a study of the dynamic structure of a large region of the orbital space requires consideration of several tens of thousands of such time series. As an alternative approach, technologies of deep learning can be used, namely, design of architecture of one-dimensional convolutional neural network for supervised learning.

中文翻译：

---

人工神经网络在近地轨道空间动力结构分析中的应用

描述了将人工神经网络应用于研究近地轨道空间选定区域的动态结构的第一次经验。描述区域动态结构共振特性演变的时间序列分析通常是手动执行的。然而，研究轨道空间大区域的动态结构需要考虑数万个这样的时间序列。作为替代方法，可以使用深度学习技术，即用于监督学习的一维卷积神经网络架构设计。