Failure prognosis and remaining useful life prediction of control moment gyroscopes onboard satellites

Abstract

• A data-driven method for failure prognosis and RUL prediction of CMGs is proposed. • The proposed method reduces the need for historical data by 93.75%. • The proposed method uses General path model and Bayesian updating technique. • The proposed method has 96.25% accuracy when 30% of data is available for online prediction. Developing a Diagnosis, Prognosis, and Health Monitoring (DPHM) framework for a small satellite is a challenging task due to the limited availability of onboard health monitoring sensors and computational budget. This paper deals with the problem of developing a DPHM framework for a satellite attitude actuator system that uses single gimballed Control Moment Gyros (CMGs) in a pyramid configuration using only the attitude measurement data to eliminate the need for subsystem level sensor data acquisition. A data-driven model is used to mimic the nominal plant dynamics and fault is induced in the spin motor of the CMG of the satellite dynamics to generate run to failure data of the attitude control system. The general path model is applied to capture the prognosis of the system and using the parameters from the general path model as apriori information, the Bayesian updating technique is used to predict the remaining useful life of a system in real-time. The algorithm performs with 96.25% accuracy when 30% of data is available for online prediction.