Point-Correlate Adversarial Transformer for Unsupervised Multivariate Time Series Anomaly Detection

Multivariate time series anomaly detection plays a crucial role in industrial production. However, the inherent complexity and randomness of time series pose significant challenges. Furthermore, existing detection methods struggle to provide reliable explanations for outliers. To address these issues, this paper presents an unsupervised multivariate time series anomaly detection model named Point-Correlate Adversarial Transformer (PCAT). In this work, we leverage Transformer networks to capture the underlying correlations between different points in a time series and reconstruct the original sequence. By analyzing the correlation differences and reconstruction errors, we identify anomalies at the point level. Our model incorporates an adversarial structure, enabling unsupervised learning and enhancing the learning capability and robustness of the detection network. Experimental evaluations on four real-world datasets demonstrate the superiority of our approach over other state-of-the-art models in terms of detection delay and accuracy.

I. Introduction

Nowadays, the rapid development of the Industrial Internet of Things (IIoT) has promoted the traditional industry to a new stage of intelligence, which has spawned the widespread application of edge devices such as sensors and actuators [1] [2]. These devices establish network connections with industrial machines, enabling comprehensive monitoring capabilities. As a result, a large number of multivariate time series data for intelligent analysis are generated, providing real-time and effective decision-making for system applications such as industrial transportation, industrial manufacturing and intelligent medical treatment [3]. However, the abnormal fragments in these time series data often indicate abnormal behaviors in the operation of industrial machines, which can cause incalculable losses if not taken seriously.