目的 工业数字孪生技术在钢铁行业的应用展现了其显著的潜力，成为工厂数字转型的核心技术之一，尤其在数据建模方面。本文梳理工业数字孪生中数据建模技术在钢铁行业中的应用。方法 通过梳理相关文献，重点分析四种数据建模方法，即基于知识的方法、基于机理的方法、基于传统机器学习的方法和基于深度学习的方法，详细介绍这些方法的优势、局限性和具体应用案例。探索了这些方法在数字物理实体融合的工业数字孪生系统构建中的潜力，以及未来模型的可扩展性设计，特别是针对大模型技术的应用。结论 通过这一综述，梳理钢铁行业现有工业数字孪生数据建模技术，为钢铁行业的数字化转型提供有价值的见解，并为未来的研究和实践提供方向。

With great application potential in the iron and steel industry, the industrial digital twin technology in the iron and steel industry becomes a core technology for the digital transformation of the plant, especially in terms of data modeling. The work aims to review the application of data modeling technology in industrial digital twin in the iron and steel industry. The paper focused on analyzing four data modeling methods based on literature research:knowledge-based method, mechanism-based method, traditional machine learning method and deep learning method, and introduced the advantages, limitations and specific application cases of these methods in detail. The fusion potential of these methods in the construction of industrial digital twins was discussed. In addition, it also took into consideration the scalable design of future models, especially for the application of large model technologies. This review provides an in-depth understanding of existing industrial digital twin data modeling technologies in the iron and steel industry, valuable insights into the digital transformation, and directions for future research and practice.