基于模型融合的煤矿瓦斯灾害隐患识别方法
A model fusion method, which involves deploying a lightweight LSTM model underground for real-time inference and cloud-based deep analysis, solves the problem of balancing the timeliness and accuracy of early warning in coal mine gas disaster monitoring, and achieves efficient gas disaster identification and early warning.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- CHINA COAL RES INST
- Filing Date
- 2026-04-01
- Publication Date
- 2026-07-17
AI Technical Summary
In existing coal mine gas disaster monitoring, fixed threshold alarm methods are difficult to capture early warning information, the early warning timing is delayed, and the false alarm rate of single indicator alarms is high. Machine learning methods are not adaptable to the fusion of multi-source heterogeneous data and complex geological conditions.
A model fusion-based approach is adopted to deploy a lightweight LSTM time series prediction model for real-time inference at the downhole edge. A dual-threshold judgment mechanism is combined to generate high-confidence early warning results. When there are suspicious anomalies, the data is uploaded to the cloud for in-depth analysis. The Transformer model and gas disaster knowledge graph are used for in-depth analysis.
It achieves a balance between real-time monitoring and early warning in coal mines and high accuracy in identifying complex disasters, reduces system response time and improves the accuracy of early warnings, and provides high-confidence rapid response and in-depth analysis of suspicious anomalies.
Smart Images
Figure CN122416637A_ABST