Abstract: To enhance the intelligent control level of the converter steelmaking process, the research progress in models for endpoint control and slag foaming prediction was reviewed. Through a systematic review of relevant research, the evolution of endpoint control models from static approaches, including mechanistic, incremental, and intelligent models, to dynamic control models based on sub-lance, off-gas, and image data was analyzed.The development trend of slag foaming prediction models from single-modal prediction model based on audio, image and off-gas data to multi-modal prediction model combining these data sources was discussed.The review indicates that while existing models have achieved certain success in both theory and industrial applications, they still face significant challenges, particularly in terms of robustness and interpretability under complex operating conditions, as well as a strong dependence on high-quality data. This study concludes that the deep integration of metallurgical mechanisms with data-driven methods, coupled with the development of efficient multi-modal information fusion techniques, represents a key direction for improving the prediction accuracy and robustness of the models in the future.

Key words: converter steelmaking, end-point control, static control model, dynamic control model, slag foaming prediction model