A baseline determination of flywheel maneuver performance based on Transformer

As a prerequisite for satellite component health assessment, the performance baseline has important reference significance and engineering value for ensuring safe and reliable satellite operation and mission success, and at the same time, it can provide scientific basis and theoretical and methodological support for the reasonable classification and determination of component health states. This paper takes the flywheel in attitude maneuvering mode as the research object, and constructs a multi-domain performance feature indicator set by analyzing the working mechanism of the flywheel in the maneuvering mode and selecting suitable health feature parameters. Secondly, a performance baseline establishment method based on the Transformer neural network model is proposed, and the hidden mapping relationship between the health feature parameters and the dynamic performance indicators is explored by using multi-head attention mechanism. Finally, the validity of the baseline model of the flywheel in maneuvering mode mentioned in this paper is verified by building a simulation platform.