一种基于机理约束与动态更新的高炉内壁温度预测方法
By combining the heat conduction mechanism model with multicenter nonnegative matrix decomposition and LightGBM regression prediction model, a dynamic update method for predicting blast furnace inner wall temperature was developed. This solved the problems of computational efficiency and stability in predicting blast furnace wall temperature, and achieved high-precision online prediction and parameter drift adaptation.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- NORTHEASTERN UNIV CHINA
- Filing Date
- 2026-06-18
- Publication Date
- 2026-07-17
AI Technical Summary
Existing methods for predicting blast furnace wall temperature are insufficient in terms of computational efficiency and stability, making it difficult to meet the needs of rapid online prediction in industrial settings. Furthermore, drift in operating parameters leads to a decrease in prediction accuracy.
By combining the furnace wall heat conduction mechanism model with multicenter nonnegative matrix decomposition and LightGBM regression prediction model, and adapting to the drift of operating parameters through a dynamic update mechanism, a method for predicting the blast furnace inner wall temperature based on mechanism constraints and dynamic updates is established.
It improves prediction efficiency and stability, can automatically identify parameter drift and update the model, maintains high-precision furnace wall temperature prediction, and supports blast furnace thermal state analysis and anomaly early warning.
Smart Images
Figure CN122414005A_ABST