Prognostic for on-orbit satellite momentum wheel based on the similitude method

Momentum wheel is the main execution unit for the attitude control of three-axis stabilized satellite. On the long-life satellites, momentum wheels are often required working stably and reliably for several years or even ten years. It is necessary to evaluate the remaining life of on-orbit satellite momentum wheel. The failure mechanism of momentum wheel is very complex and it is difficult to predict its long-range performance with trend extrapolation method. Data-driven prognostic is combined with similitude method to predict the remaining life of momentum wheel. The causes of the momentum wheel failure are analyzed to select key parameters to indicate life characteristics of momentum wheel. The lifetime performance data of some other momentum wheels failed are used to train neural network models to describe the degradation process of similar momentum wheels, which is used for predicting momentum wheel on-orbit. In the case study, this method showed a good performance. It is more applicable than trend extrapolation and has much better performance.