A Review of Anomaly Detection Methods for Multivariate Time Series Based on Time-Domain and Frequency-Domain Perspectives
Journal of Big Data and Computing · 2026-03 · 4 (1) · 167-180
DOI: 10.62517/jbdc.202601123 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Big Data and Computing, 2026-03. Verfasst von Jiaming Xu und 2 weiteren Autor:innen. Die Arbeit nennt 35 Literaturangaben.
Abstract (Originalsprache)
With the increasing proliferation of sensor fusion, industrial control systems (ICS), and internet services, multivariate time-series data has become ubiquitous in fields such as intelligent manufacturing, financial monitoring, network security, and transportation. Multivariate Time-Series Anomaly Detection (MTSAD) aims to identify patterns that deviate significantly from normal behavior within high-dimensional dynamic metrics, thereby providing critical support for system monitoring and early warning. Compared with static data and univariate sequences, multivariate time series are characterized by temporal dependencies, complex inter-variable interactions, and dynamic distribution shifts, posing strictly higher requirements for anomaly detection. This paper first introduces the definitions of time-series anomaly detection and categorizes common anomaly types. Secondly, it classifies existing methods from both time-domain and frequency-domain perspectives, providing a comprehensive analysis of their advantages, limitations, and application scenarios. Finally, the paper explores key research directions for the future design of anomaly detection methods, offering a reference for both theoretical and applied research in this domain.
Zitieren
APA
Xu, J., Liu, H., & Xu, J. (2026). A Review of Anomaly Detection Methods for Multivariate Time Series Based on Time-Domain and Frequency-Domain Perspectives. Journal of Big Data and Computing, 4(1), 167-180. https://doi.org/10.62517/jbdc.202601123
Vancouver
Xu J, Liu H, Xu J. A Review of Anomaly Detection Methods for Multivariate Time Series Based on Time-Domain and Frequency-Domain Perspectives. Journal of Big Data and Computing. 2026;4(1):167-180. doi:10.62517/jbdc.202601123
BibTeX
@article{Xu2026,
author = {Xu, Jiaming and Liu, Hanqi and Xu, Jun},
title = {A Review of Anomaly Detection Methods for Multivariate Time Series Based on Time-Domain and Frequency-Domain Perspectives},
journal = {Journal of Big Data and Computing},
year = {2026},
volume = {4},
number = {1},
pages = {167--180},
doi = {10.62517/jbdc.202601123}
}
EUVIMED