Satellite actuator fault detection method based on migration component analysis

Abstract

The invention discloses a satellite actuator fault detection method based on migration component analysis, and relates to the field of satellite attitude control system fault detection, and the method comprises the steps: obtaining a source domain neural network model based on normal sample data training of a source domain system; utilizing the normal sample data of a target domain system to adjust network parameters to obtain a target domain neural network model, utilizing the network model to obtain source domain and target domain data residual errors, after residual error features are extracted, adopting a migration component analysis method, so the distribution distance of source domain data features and target domain data features is reduced, and a source domain data feature training classifier can be used for diagnosis of actual satellite data samples. According to the invention, the fault detection of a satellite flywheel is realized by using nominal model simulation data and a migration component analysis method aiming at the situation that a satellite has no fault samples, and the reliability and ground monitoring capability of a satellite attitude control system are improved.

Description

technical field [0001] The invention relates to the field of fault detection of satellite attitude control systems, in particular to a fault detection method for satellite actuators based on migration component analysis. Background technique [0002] With the continuous development of aerospace technology, the complexity of space missions continues to increase. The spacecraft attitude control system is one of the most important subsystems of the spacecraft, so it is of great significance to study its fault detection method. The flywheel is one of the most commonly used actuators in spacecraft, and its reliability is an important condition to ensure the normal operation of satellites. Conventional data-driven satellite actuator fault detection requires a large number of fault samples, but the actual satellite does not have fault samples with labels, so it is not applicable. It is urgent to propose a method that does not depend on the fault samples of the actual satellite and...

AI Technical Summary

Problems solved by technology

Conventional data-driven satellite actuator fault detection requires a large number of fault samples, but the actual satellite does not have fault samples with labels, so it is not applicable. It is urgent to propose a method that does not depend on the fault samples of the actual satellite and can realize satellite execution. fault detection of the controller, improve the reliability of the attitude control subsystem, and ensure the reliable execution of satellite missions

Images