Anomaly Detection Scheme Using Global and Local Features in Time Series Data

In the industrial field, a variety of time series data is continuously generated. Analyzing this data is essential for monitoring processes within the industrial environment. One of the key methods of time series analysis is anomaly detection. Anomalies in time series data can indicate various situations, such as facility failures or sensor errors. In this paper, we propose a method that considers both global and local features to detect time series anomalies. Using these features, the proposed scheme classifies time series data into multiple clusters. A deep learning model, such as a Transformer, is then trained for each cluster to perform time series prediction. Finally, the proposed scheme utilizes an ensemble approach for robust anomaly detection.