Applications of deep learning to selected aerospace systems

Long-distance space systems generate enormous amounts of bigdata. These bigdata can be used to generate intelligent that can help us better understand the behavior of space systems. There is currently no such tool for precisely understanding and predicting the behavior of aerospace systems. In this study, three different aerospace systems are analyzed to build the respective artificial intelligence (AI) models to understand and predict their space behavior using the deep learning (DL) ecosystem. We studied the pulsed plasma thruster (PPT), an electric space propulsion system; the ARTEMIS-P1 spacecraft sensor array; and the UAV battery system. Three sets of comparative analyses are carried out to assess the model accuracy. A number of tests are utilized to assess and predict the exact physical behavior. The comparison and test results show that DL-based artificial models are capable enough (> 99%) to mimic the exact system behaviors. This DL-based approach provides a novel means of understanding and predicting the real behavior of the aerospace systems.