Simulation degradation datasets for health prognosis of a control moment gyroscope's flywheel system

A Control moment gyroscope (CMG) is an attitude control actuator widely used in spacecraft. As a consequence of the continuous high-speed rotation to output stable angular momentum, CMGs are at a higher risk of degradation failure, which could lead to a loss of spacecraft control. However, implementing health prognosis for CMGs is challenging due to the high cost associated with conducting comprehensive lifetime tests. Moreover, environment differences between space and ground make data generated with simulation in the space environment more valuable than data from ground experiments for health prognosis. Thus, we develop a simulation model for the flywheel system of a CMG to generate degradation data, aiming to advance deep learning-based health prognosis in aerospace applications. The datasets focus on the brushless DC motor of the flywheel system, encompassing the bus current, bus voltage, rotor speed time window from CMGs under three different operational conditions. A data-approximate model is built and two parameters are set to change over time to simulate the mechanical-loss-related and electromechanical-conversion-related degradation. When compared to a dataset from a real CMG, the simulation dataset demonstrates similar dynamic characteristics and data patterns. The dataset can support the development and validation of prognostics methods such as remaining useful life(RUL) prediction, degradation trend estimation, and cross-condition transfer learning for CMG flywheel systems.