Data processing method based on telemetry data, computer device and storage medium
By extracting telemetry data features through a parallel architecture of graph convolutional networks and bidirectional long short-term memory networks, and combining it with a dynamic threshold evaluator of variational autoencoders, the problem of capturing the nonlinear coupling law of telemetry parameters in complex systems is solved, the false alarm and false alarm rates are reduced, and adaptive telemetry data monitoring is achieved.
Patent Information
- Application Number
- CN202610864429.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-06-16
- Publication Date
- 2026-07-17
AI Technical Summary
Existing technologies cannot effectively capture the nonlinear coupling patterns between various telemetry parameters in complex systems, resulting in a high rate of missed reports of latent faults. Furthermore, the traditional static threshold discrimination method is prone to causing a high frequency of false alarms when switching operating conditions.
A parallel architecture of graph convolutional network and bidirectional long short-term memory network is used to extract spatial topological correlation features and long-range time-axis dependency features of telemetry data. Combined with a dynamic threshold evaluator of variational autoencoder, anomaly scores are generated to achieve adaptive alarm.
It effectively reduces the false alarm rate of latent faults, improves the monitoring accuracy of complex systems, and reduces the false alarm frequency through a dynamic threshold mechanism, adapting to parameter baseline drift at different operating stages.
Smart Images
Figure CN122412923A_ABST