Machine Learning-based vs Deep Learning-based Anomaly Detection in Multivariate Time Series for Spacecraft Attitude Sensors

TL;DR

This paper compares machine learning and deep learning methods for detecting anomalies in multivariate time series data from spacecraft attitude sensors, highlighting their performance, interpretability, and generalization capabilities.

Contribution

It characterizes and compares ML-based and DL-based approaches for stuck value detection in spacecraft sensor data, providing insights into their effectiveness and applicability.

Findings

Deep learning approaches show higher accuracy in anomaly detection.

Machine learning methods offer better interpretability.

Performance varies across different scenarios.

Abstract

In the framework of Failure Detection, Isolation and Recovery (FDIR) on spacecraft, new AI-based approaches are emerging in the state of the art to overcome the limitations commonly imposed by traditional threshold checking. The present research aims at characterizing two different approaches to the problem of stuck values detection in multivariate time series coming from spacecraft attitude sensors. The analysis reveals the performance differences in the two approaches, while commenting on their interpretability and generalization to different scenarios.