Fault Diagnosis Method for Micro-nano Satellite Thruster Based on VAE-DRSN
Aiming at the problem of fault diagnosis for micro-nano satellite thrusters,a data-driv-en propulsion system fault diagnosis method based on variational autoencoder-deep residual shrinkage network(VAE-DRSN)is proposed in this paper.This method uses a variational au-toencoder to extract the features from attitude data and controller output,and the extracted fea-tures are classified through a deep residual shrinkage neural network.It can accurately detect,di-agnose and locate the stuck on/off and make efficiency reduction faults of thrusters online,with-out the need for satellite thruster model and dynamic model,and without the need for separate hardware measurement mechanism.The numerical simulation results show that the accuracy of this method for single nozzle fault detection can reach over 99%,and it has good ability for thruster fault location and diagnosis.
国家科技期刊平台
NSTL
万方数据
