A time series anomaly detection method and system based on multi-scale spatio-temporal modeling

By employing a multi-scale spatiotemporal modeling approach, this method utilizes one-dimensional convolution and a spatiotemporal encoder to generate multi-scale sub-time series, enabling cross-scale information exchange and decoding reconstruction. This addresses the issues of insufficient cross-scale interaction and neglect of spatiotemporal dependencies in existing methods, achieving highly accurate anomaly detection.

CN122365232APending Publication Date: 2026-07-10NANJING UNIV OF AERONAUTICS & ASTRONAUTICS
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
NANJING UNIV OF AERONAUTICS & ASTRONAUTICS
Filing Date
2026-03-30
Publication Date
2026-07-10

AI Technical Summary

Technical Problem

Existing multi-scale time series anomaly detection methods suffer from insufficient cross-timescale interaction and lack joint modeling of spatiotemporal dependencies in evolution, resulting in inadequate anomaly detection accuracy.

Method used

A multi-scale spatiotemporal modeling approach is adopted, which generates sub-time series of different scales through one-dimensional convolution, extracts features using a spatiotemporal encoder, and combines a cross-scale hybrid expert mechanism for information exchange and decoding reconstruction. The reconstruction error is then calculated to generate anomaly scores.

Benefits of technology

It improves the accuracy of anomaly detection, effectively capturing subtle and scale-sensitive anomalies, thereby enhancing system security and operational efficiency.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122365232A_ABST
    Figure CN122365232A_ABST
Patent Text Reader

Abstract

This invention discloses a time series anomaly detection method and system based on multi-scale spatiotemporal modeling, comprising: extracting multivariate time series from business data; generating multiple sub-time series at different scales using one-dimensional convolution; independently encoding each scale sub-time series using a scale-independent spatiotemporal encoder to obtain spatiotemporal enhanced features of the multi-scale time series; employing a cross-scale hybrid expert mechanism to achieve information exchange between scales, obtaining a multi-scale sequence representation after scale interaction; integrating the interaction representations of each scale and performing decoding and reconstruction; generating anomaly scores by calculating the reconstruction error between the original input and the reconstructed sequence, and obtaining anomaly detection results. This method decomposes complex time series into multiple scales, with each scale collaboratively modeling spatiotemporal dependencies, which helps to more accurately discover and verify anomalous signals, improving the accuracy and reliability of detection.
Need to check novelty before this filing date? Find Prior Art