基于序列-图像映射的航天器智能故障诊断方法

卫星组网是未来航天的发展大趋势，要保证众星在轨安全可靠稳定运行，要求单星具备高精度的在轨自主故障诊断能力。本文针对航天器控制系统故障闭环传播和数据维数高的特点，结合某航天器的地面测试数据，首先对高维耦合序列数据进行处理，实现序列到灰度图像的映射，然后采用卷积神经网络完成高精度同一故障部件的故障诊断。通过将所提方法与K邻近算法、基于主成分分析的K邻近算法等非图像化机器学习算法进行对比验证，说明了本文所提方法的有效性。

Intelligent Fault Diagnosis Method for Spacecraft Based on Sequence-Image Mapping

Satellite networking is a major trend in the future development of spaceflight, and to ensure the safe, reliable and stable operation of many satellites in orbit, a single satellite is required to have high-precision in-orbit autonomous fault diagnosis capability. In this paper, for the characteristics of closed-loop fault propagation and high data dimensionality of the spacecraft control system, combined with the ground test data of a spacecraft, we first process the high-dimensional coupled sequence data to realize the mapping from sequence to grayscale image, and then use the convolution neural network (CNN) to complete the fault diagnosis of the same faulty component with high accuracy. The effectiveness of the proposed method is illustrated by comparing and validating it with non-image-based machine learning algorithms such as the K-neighborhood algorithm and the K-neighborhood algorithm based on principal component analysis.