Gautam Biswas, Hamed Khorasgani, Gerald Stanje, Abhishek Dubey, Somnath Deb, Sudipto Ghoshal
International Journal of Prognostics and Health Management 2016
Spacecraft operations involve complex mode transitions and potential anomalies difficult to characterize a priori. This approach innovates by combining unsupervised learning to discover operational patterns with expert consultation to validate findings, enabling detection of new anomalies while avoiding false positives from specialized operational scenarios previously unknown to the detection system.
This paper discusses mode and anomaly detection approaches for spacecraft telemetry combining unsupervised learning with expert knowledge. The method applies data-driven techniques to identify temporal patterns representing operational modes, distinguishes between specialized and anomalous modes using hybrid approaches. The framework enables automated detection while accounting for temporal dynamics in telemetry signatures.
The framework successfully demonstrates mode classification and anomaly identification in spacecraft power system data through hierarchical clustering and feature analysis. Temporal pattern matching reveals distinct operational signatures for different mission phases. Expert validation confirms the approach effectively separates nominal specialized modes from true anomalous behaviors in long-duration space missions.
@article{Biswas2016a,
author = {Biswas, Gautam and Khorasgani, Hamed and Stanje, Gerald and Dubey, Abhishek and Deb, Somnath and Ghoshal, Sudipto},
journal = {International Journal of Prognostics and Health Management},
title = {An approach to mode and anomaly detection with spacecraft telemetry data},
year = {2016},
abstract = {This paper discusses a mixed method that combines unsupervised learning methods and human expert input for analyzing telemetry data from long-duration robotic space missions. Our goal is to develop more automated methods detecting anomalies in time series data. Once anomalies are identified using unsupervised learning methods we use feature selection methods followed by expert input to derive the knowledge required for building on-line detectors. These detectors can be used in later phases of the current mission and in future missions for improving operations and overall safety of the mission. Whereas the primary focus in this paper is on developing data-driven anomaly detection methods, we also present a computational platform for data mining and analytics that can operate on historical data offline, as well as incoming telemetry data on-line.},
contribution = {colab},
file = {:Biswas2016a-An_approach_to_mode_and_anomaly_detection_with_spacecraft_telemetry_data.pdf:PDF},
keywords = {mode detection, anomaly detection, spacecraft systems, unsupervised learning, temporal analysis, hybrid methods, health monitoring},
tag = {a14cps}
}