基于知识与数据驱动的铁矿原料自适应评价研究OA

In the processing flow of raw materials before ironmaking in blast furnaces,the evaluation of iron ore is crucial for achieving cost reduction,efficiency improvement and efficient resources utilization.A comprehensive evaluation model system for iron ore which integrated knowledge and data-driven approaches was proposed in this paper to resolve the issues of singular indicator dimensions and difficulties in feature information fusion in iron ore evaluation.Firstly,a multi-dimensional evaluation indicator system encompassing physical,metallurgical,and economic properties was constructed.The entropy weight method and the analytic hierarchy process(AHP)were adopted to achieve the weight representation of objective feature information and subjective prior knowledge,respectively.Secondly,a fusion evaluation model of iron ore raw materials was established based on deep neural network,and the posterior evaluation information was used as supervision to realize the calibration and adaptive learning of the evaluation model.Finally,a browser-server(B/S)system architecture for the evaluation model was implemented using technologies such as Vue.js with bootstrap,Java,and MySQL.By integrating domain-specific knowledge of ore blending and model-represented data,a customized knowledge base was instantiated through the DeepSeek large language model,thereby creating a human-machine model interface capable of natural language interaction.Case studies and system applications indicate that the adaptive evaluation model can more accurately capture the features of iron ore raw materials,and the constructed integrated system possesses stronger information self-learning capabilities and interpretability.The established iron ore evaluation framework,along with its associated model and system,provides robust support for intelligent management of raw materials in the ironmaking process and facilitates digital twin implementation in steelmaking enterprises.