TFCdetector: Multivariable time series anomaly detection with time-frequency constraints
- 1College of Physics and Electronic Information Engineering, Qinghai Normal University, Xining, China.
- 2College of Computer Technology and Application, Qinghai University, Xining, China.
- 3College of Computer Science and Technology, Xinjiang University, Urumqi, China.
Abstract
The multivariate time series collected from numerous sensors exhibit complex interwoven relationships, arising from both the coupling among different components and the temporal dependencies of fluctuations within each component. Disentangling these relationships remains a persistent challenge for the time series community. Existing reconstruction-based methods attempt to unravel the intertwined relationships through sequence decomposition and multi-scale modeling. However, the decomposed components remain mixtures of complex latent factors. Moreover, most methods overlook anomalies present in the training set, which inadvertently causes the model to develop a certain affinity toward anomalous patterns. To address these limitations, we propose TFCdetector. Motivated by the intrinsic properties of anomalies, TFCdetector leverages time-frequency domain analysis to explicitly suppress the model’s representation learning of anomalies across diverse granularities and components. To further constrain the model’s learning of anomalies, we propose an energy-correlation constraint loss that operates on both segment-level energy and sequence-level correlation. Notably, the adaptively updated memory storage block enhances the expression of normal patterns by leveraging the affinity between normal patterns in the input sequence and the memory storage block. Furthermore, the meticulously designed twin frequency attention enables the model to dynamically select local or global frequency information beneficial for the task. Extensive experiments on 1 synthetic dataset and 7 real-world datasets show that TFCdetector achieves competitive results in multivariate time series detection tasks.The implementation code is available at https://github.com/FanJY3/TFCdetector.
Related
- Learn the interactions: Weakly supervised video anomaly detection with human-object interactions
- Dual-domain self-supervised feature alignment via spectral–spatial representation learning for deepfake anomaly detection
- Comparison of anomaly detection and supervised learning for acute myeloid leukemia blast identification: A proof-of-concept study
- A cascaded U-nets model based on hybrid attention for video anomaly detection
- Anomaly detection uncovers emerging novelty in coral assemblages of the Great Barrier Reef
BibTeX
@article{fan2026tfcdetector,
title = {{TFCdetector: Multivariable time series anomaly detection with time-frequency constraints}},
author = {Jinyu Fan and Jianfei Zhao and Jun Ma and Yuxi Cai},
journal = {PLOS One},
year = {2026},
month = oct,
doi = {10.1371/journal.pone.0361145},
url = {https://doi.org/10.1371/journal.pone.0361145},
note = {CC BY},
}
RIS
TY - JOUR
TI - TFCdetector: Multivariable time series anomaly detection with time-frequency constraints
AU - Jinyu Fan
AU - Jianfei Zhao
AU - Jun Ma
AU - Yuxi Cai
PY - 2026
DA - 2026/10/28
JO - PLOS One
DO - 10.1371/journal.pone.0361145
UR - https://doi.org/10.1371/journal.pone.0361145
AB - The multivariate time series collected from numerous sensors exhibit complex interwoven relationships, arising from both the coupling among different components and the temporal dependencies of fluctuations within each component. Disentangling these relationships remains a persistent challenge for the time series community. Existing reconstruction-based methods attempt to unravel the intertwined relationships through sequence decomposition and multi-scale modeling. However, the decomposed components remain mixtures of complex latent factors. Moreover, most methods overlook anomalies present in the training set, which inadvertently causes the model to develop a certain affinity toward anomalous patterns. To address these limitations, we propose TFCdetector. Motivated by the intrinsic properties of anomalies, TFCdetector leverages time-frequency domain analysis to explicitly suppress the model’s representation learning of anomalies across diverse granularities and components. To further constrain the model’s learning of anomalies, we propose an energy-correlation constraint loss that operates on both segment-level energy and sequence-level correlation. Notably, the adaptively updated memory storage block enhances the expression of normal patterns by leveraging the affinity between normal patterns in the input sequence and the memory storage block. Furthermore, the meticulously designed twin frequency attention enables the model to dynamically select local or global frequency information beneficial for the task. Extensive experiments on 1 synthetic dataset and 7 real-world datasets show that TFCdetector achieves competitive results in multivariate time series detection tasks.The implementation code is available at https://github.com/FanJY3/TFCdetector.
SN - 1932-6203
ER -