Continuous casting is the most important route for the production of steel today. Due to the physical, mechanical, and chemical components involved in the production, continuous casting is a very complex process, pushing conventional methods of monitoring and control to their limits. In recent years, this complexity and the increasing global competition created a demand for new methods to monitor and control the continuous casting process. Due to the success and associated rise of machine learning techniques in recent years, machine learning nowadays plays an essential role in monitoring and controlling complex processes. This publication presents a scientific survey of machine learning techniques for the analysis of the continuous casting process. We provide an introduction to both the involved fields: an overview of machine learning, and an overview of the continuous casting process. Therefore, we first analyze the existing work concerning machine learning in continuous casting of steel and then synthesize the common concepts into categories, supporting the identification of common use cases and approaches. This analysis is concluded with the elaboration of challenges, potential solutions, and a future outlook of further research directions.

连续铸造是当今钢铁生产中最重要的途径。由于生产中涉及的物理，机械和化学成分，连续铸造是一个非常复杂的过程，将常规的监视和控制方法推向了极限。近年来，这种复杂性和日益加剧的全球竞争对监控和控制连续铸造过程的新方法提出了要求。由于近年来机器学习技术的成功发展以及随之而来的兴起，如今的机器学习在监视和控制复杂过程中起着至关重要的作用。该出版物介绍了机器学习技术的科学概况，以分析连续铸造过程。我们提供了有关这两个领域的介绍：机器学习概述，以及连续铸造过程的概述。因此，我们首先分析与钢水连铸中的机器学习有关的现有工作，然后将通用概念归纳为几类，以支持通用用例和方法的识别。通过对挑战的阐述，潜在的解决方案以及对进一步研究方向的未来展望来结束本分析。