A Hybrid Modeling Method for the Reaction Zone of a Mineral Electric Furnace Based on Mechanism and Dynamic Compensation of Data

To achieve precise control of the submerged arc furnace smelting process, it is first necessary to accurately predict the dynamic characteristics and development trends of the reaction zone furnace conditions. This paper establishes a mechanism model for the operating resistance of the molten pool during the submerged arc smelting process in an electric arc furnace, based on an in-depth study of the heat generation mechanisms of arc resistance and charge layer resistance, their interactive relationships, and the continuous changes in the resistivity of the molten charge layer caused by heat accumulation in the reaction zone. Additionally, a data-model-based dynamic compensation mechanism is proposed. This mechanism utilizes a bidirectional long-short term memory network to capture the temporal characteristics of data and employs a multi-head attention mechanism to achieve dynamic feature focusing. This approach addresses the limitations of the mechanism model under complex operating conditions, thereby enhancing the accuracy and robustness of tension prediction. Experimental results demonstrate that the hybrid model effectively improves the prediction accuracy of reaction zone furnace condition trends, laying the foundation for further refinement of the smelting process control.