Deep Learning Based Fault Classification for Micro-thrusters of Drag-Free Spacecraft
摘要
This paper proposes a deep learning (DL) based fault classification framework for the micro-thrusters of drag-free spacecraft, which consists of autoencoders and a softmax classifier. Firstly, the drag-free system dynamics is established and the fault injection technique is used to obtain the dataset for training the DL networks. Subsequently, the optimal parameters of the autoencoders are selected using a layer-by-layer training approach. After that, the softmax classifier is integrated with autoencoders to form the final network structure for fault classification. Finally, the standardized dataset is used to complete the training and testing of the DL network. The results show that the proposed method can quickly and precisely classify the common fault types of micro-thrusters of drag-free spacecraft.