DCFF-MTAD: A Multivariate Time-Series Anomaly Detection Model Based on Dual-Channel Feature Fusion

• DOI: 10.3390/s23083910
• Publication Date: 2023-04-12T06:57:08Z
• Authors: Zheng Xu; Yumeng Yang; Xinwen Gao; Min Hu
• Keywords: Feature (linguistics)

The detection of anomalies in multivariate time-series data is becoming increasingly important the automated and continuous monitoring complex systems devices due to rapid increase volume dimension. To address this challenge, we present a anomaly model based on dual-channel feature extraction module. module focuses spatial time features using short-time Fourier transform (STFT) graph attention network, respectively. two are then fused significantly improve model’s performance. In addition, incorporates Huber loss function enhance its robustness. A comparative study proposed with existing state-of-the-art ones was presented prove effectiveness three public datasets. Furthermore, by shield tunneling applications, verify practicality model.