Detecting anomalies in multivariate time series (MTS) data is essential for ensuring stability in critical domains such as industrial monitoring, healthcare, and financial transactions. Traditional methods often fail to capture complex spatiotemporal patterns and are sensitive to noise interference. This study proposes an innovative Spatio-Temporal Attention with Decomposition (STAD) model that addresses these limitations. STAD effectively captures global spatial structures and local temporal fluctuations through an independent-channel, patch-based attention mechanism and a trend-residual decomposition approach. A dynamic sample selection loss function is introduced to enhance model robustness by dynamically adjusting training samples and minimizing the impact of anomalous data during training. An innovative information entropy-based scoring method effectively filters out noise and redundant information by identifying and selecting the most relevant features for anomaly detection. Experiments conducted on benchmark datasets, including Server Machine Dataset (SMD), Secure Water Treatment dataset (SWaT), and Mars Science Laboratory Rover dataset (MSL) show that STAD achieves state-of-the-art performance.
AudibertJMichiardiPGuyardF, et al. (2020) USAD: Unsupervised anomaly detection on multivariate time series. In: Proceedings of the 26th ACM SIGKDD international conference on knowledge discovery & data mining, Virtual Event, 6–10 July, pp. 3395–3404. New York: Association for Computing Machinery.
2.
BengioYSimardPFrasconiP (1994) Learning long-term dependencies with gradient descent is difficult. IEEE Transactions on Neural Networks5(2): 157–166.
3.
BronsteinM-MBrunaJLeCunY, et al. (2017) Geometric deep learning: Going beyond Euclidean data. IEEE Signal Processing Magazine34(4): 18–42.
4.
DaiEChenJ (2022) Graph-augmented normalizing flows for anomaly detection of multiple time series. arXiv [preprint] arXiv: 2202.07857.
5.
DanesePKalchschmidtM (2011) The role of the forecasting process in improving forecast accuracy and operational performance. International Journal of Production Economics131(1): 204–214.
6.
DengAHooiB (2021) Graph neural network-based anomaly detection in multivariate time series. Proceedings of the AAAI Conference on Artificial Intelligence35(5): 4027–4035.
7.
EsterMKriegelH-PSanderJ, et al. (1996) A density-based algorithm for discovering clusters in large spatial databases with noise. Knowledge Discovery in Databases96(34): 226–231.
8.
GravesA (2012) Long short-term memory. In: GravesA (ed.) Supervised Sequence Labelling with Recurrent Neural Networks. Berlin: Springer, pp. 37–45.
9.
HeSGuoQFengK, et al. (2026) Multivariate time series anomaly detection using learnable spatial-temporal graph ordinary differential equations network. IEEE Transactions on Dependable and Secure Computing23: 3723–3737.
KwaoVRaptisIWorknehH (2024) Applied fault diagnosis with distributed monitoring systems: Bridging theory with practice. Transactions of the Institute of Measurement and Control. Epub ahead of print 17 December. DOI: 10.1177/01423312241296934.
12.
LachureJDoriyaR (2024) Intelligent sensor data analysis through hybrid deep hierarchical clustering for anomaly detection. Transactions of the Institute of Measurement and Control. Epub ahead of print 31 December. DOI: 10.1177/01423312241299859.
13.
LeCunYBengioYHintonG (2015) Deep learning. Nature521(7553): 436–444.
14.
LeeJKimS-B (2022) Uncertainty-aware hierarchical segment-channel attention mechanism for reliable and interpretable multichannel signal classification. Neural Networks150: 68–86.
15.
LiRLiuZZhuX, et al. (2025) Detecting multivariate time series anomalies with cascade decomposition consistency. IEEE Transactions on Instrumentation and Measurement74: 1–14.
16.
LinSClarkRBirkeR, et al. (2020) Anomaly detection for time series using VAE-LSTM hybrid model. In: ICASSP 2020-2020 IEEE international conference on acoustics, speech and signal processing (ICASSP), Barcelona, 4–8 May, pp. 4322–4326. New York: IEEE.
17.
LiuSZhouBDingQ, et al. (2022) Time series anomaly detection with adversarial reconstruction networks. IEEE Transactions on Knowledge and Data Engineering35(4): 4293–4306.
18.
LiuYHuTZhangH, et al. (2023) ITransformer: Inverted transformers are effective for time series forecasting. arXiv [preprint] arXiv: 2310.06625.
19.
NieYNguyenN-HSinthongP, et al. (2022) A time series is worth 64 words: Long-term forecasting with transformers. arXiv [preprint] arXiv: 2211.14730.
20.
PatersonRMacLeodDThetfordD, et al. (2006) Prediction of in-hospital mortality and length of stay using an early warning scoring system: Clinical audit. Clinical Medicine6(3): 281–284.
21.
PaulauskasNBaskysA (2019) Application of histogram-based outlier scores to detect computer network anomalies. Electronics8(11): 1251.
22.
PeiaORoszbachK (2015) Finance and growth: Time series evidence on causality. Journal of Financial Stability19: 105–118.
23.
RuffLVandermeulenRGoernitzN, et al. (2018) Deep one-class classification. International Conference on Machine Learning80: 4393–4402.
24.
SchmidlSWenigPPapenbrockT (2022) Anomaly detection in time series: A comprehensive evaluation. Proceedings of the VLDB Endowment15(9): 1779–1797.
25.
SchölkopfBPlattJ-CShawe-TaylorJ, et al. (2001) Estimating the support of a high-dimensional distribution. Neural Computation13(7): 1443–1471.
26.
ShannonC-E (1948) A mathematical theory of communication. The Bell System Technical Journal27(3): 379–423.
27.
ShenLWeiYWangY, et al. (2024) AFMF: Time series anomaly detection framework with modified forecasting. Knowledge-Based Systems296: 111912.
28.
ShiYWangBYuY, et al. (2023) Robust anomaly detection for multivariate time series through temporal GCNs and attention-based VAE. Knowledge-Based Systems275: 110725.
TangXZhangYLiY (2025) Augmented switching linear dynamic latent variable model based for multimode process fault diagnosis. Transactions of the Institute of Measurement and Control. Epub ahead of print 11 July. DOI: 10.1177/01423312241295796.
31.
TaxD-MDuinR-P (2004) Support vector data description. Machine Learning54: 45–66.
32.
TorresJ-FHadjoutDSebaaA, et al. (2021) Deep learning for time series forecasting: A survey. Big Data9(1): 3–21.
33.
VaswaniAShazeerNParmarN, et al. (2017) Attention is all you need. Advances in Neural Information Processing Systems30: 5998–6008.
34.
WangCZhuangZQiQ, et al. (2023) Drift doesn’t matter: Dynamic decomposition with diffusion reconstruction for unstable multivariate time series anomaly detection. Advances in Neural Information Processing Systems36: 10758–10774.
35.
XuHWangYJianS, et al. (2024) Calibrated one-class classification for unsupervised time series anomaly detection. IEEE Transactions on Knowledge and Data Engineering36(11): 5723–5736.
36.
Zamanzadeh DarbanZWebbG-IPanS, et al. (2024) Deep learning for time series anomaly detection: A survey. ACM Computing Surveys57(1): 1–42.
37.
ZengAChenMZhangL, et al. (2023) Are transformers effective for time series forecasting?Proceedings of the AAAI Conference on Artificial Intelligence37(9): 11121–11128.
38.
ZhaoBXingHWangX, et al. (2023) Rethinking attention mechanism in time series classification. Information Sciences627: 97–114.
39.
ZhaoHWangYDuanJ, et al. (2020) Multivariate time-series anomaly detection via graph attention network. In: 2020 IEEE international conference on data mining (ICDM), Sorrento, 17–20 November, pp. 841–850. New York: IEEE.
40.
ZhouJCuiGHuS, et al. (2020) Graph neural networks: A review of methods and applications. AI Open1: 57–81.
41.
ZhouQHeSLiuH, et al. (2024) Label-free multivariate time series anomaly detection. IEEE Transactions on Knowledge and Data Engineering36(7): 3166–3179.
42.
ZhuSLiYXuK, et al. (2025) Partially-supervised graph derivation network with meta-learning for time series anomaly detection. IEEE Internet of Things Journal12: 25472–25486.
