Data and knowledge driven approach for burden surface optimization in blast furnace

This paper presents a hybrid optimization strategy for determining the setting values of burden surface through measured data and domain knowledge integration manner. The proposed hybrid optimization strategy, including broad learning-based soft sensing models for production indicators, novel twin information fusion based pre-setting model, knowledge-mining based feedback compensation model, data-based production status evaluation and knowledge-based adjustment model, can adjust the setting values of burden surface in response to the changes in production status and safe operation can be reached as well. Finally, comprehensive experiments are conducted to verify the effectiveness and feasibility of the proposed method.