In spacecraft health management a large number of time series is acquired and used for on-board units surveillance and for historical data analysis. The early detection of abnormal behaviors in telemetry data can prevent failures in the spacecraft equipment. In this paper we present an advanced monitoring system that was carried out in partnership with Thales Alenia Space Italia S.p.A, a leading industry in the field of spacecraft manufacturing. In particular, we developed an anomaly detection algorithm based on Generative Adversarial Networks, that thanks to their ability to model arbitrary distributions in high dimensional spaces, allow to capture complex anomalies avoiding the burden of hand crafted feature extraction. We applied this method to detect anomalies in telemetry data collected from a simulator of a Low Earth Orbit satellite. One of the strengths of the proposed approach is that it does not require any previous knowledge on the signal. This is particular useful in the context of anomaly detection where we do not have a model of the anomaly. Hence the only assumption we made is that an anomaly is a pattern that lives in a lower probability region of the data space.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.