基于多模态时空网络的瓦斯地质异常预警方法及电子设备
By employing a multimodal spatiotemporal network approach, combined with an underground multi-source sensing network and a deep learning model, the dynamic response and geological error calibration of the gas early warning system to mining disturbances were addressed, achieving highly sensitive identification and adaptive early warning of gas anomalies.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- CCTEG CHINA COAL RES INST
- Filing Date
- 2026-03-06
- Publication Date
- 2026-07-17
AI Technical Summary
Existing gas early warning technologies fail to effectively integrate the dynamic disturbance mechanism of mining activities on geological structures, lack the integration of geophysical spatial constraints into data-driven models, and cannot adaptively calibrate initial geological errors, leading to false alarms or missed alarms.
A multimodal spatiotemporal network approach is adopted, which collects data through a downhole multi-source sensing network, eliminates heterogeneous data errors using a spatiotemporal alignment formula, constructs a topology map structure and calculates the fault influence matrix, and generates gas anomaly trend prediction values by combining deep neural networks and LSTM models. Geological parameters are then updated through a reverse correction formula to achieve adaptive calibration.
It improves the sensitivity to gas anomalies induced by mining, identifies hidden disaster-causing factors, and ensures that the early warning system is dynamically adjusted with the production process, thus improving accuracy and adaptability.
Smart Images
Figure CN122407282A_ABST