Surface defects have a serious detrimental effect on the quality of steel. To address the problems of low efficiency and poor accuracy in the manual inspection process, intelligent detection technology based on machine learning has been gradually applied to the detection of steel surface defects. An improved YOLOv8 steel surface defect detection model called YOLOv8-MGVS is designed to address these challenges. The MLCA mechanism in the C2f module is applied to increase the feature extraction ability in the backbone network. The lightweight GSConv and VovGscsp cross-stage fusion modules are added to the neck network to reduce the loss of semantic information and achieve effective information fusion. The self-attention mechanism is exploited into the detection network to improve the detection ability of small targets. Defect detection experiments were carried out on the NEU-DET dataset. Compared with YOLOv8n from experimental results, the average accuracy, recall rate, and frames per second of the improved model were improved by 5.2%, 10.5%, and 6.4%, respectively, while the number of parameters and computational costs were reduced by 5.8% and 14.8%, respectively. Furthermore, the defect detection generalization experiments on the GC-10 dataset and SDD DET dataset confirmed that the YOLOv8-MGVS model has higher detection accuracy, better lightweight, and speed.

In this study, a CaO–SiO₂–Al₂O₃-based continuous casting mold flux was designed using the FactSage thermodynamics software to determine the composition range of CaF₂. The viscosity characteristics of the mold flux were determined using a rotating viscometer. The results show that the constant temperature viscosity at 1300 °C decreases gradually as CaF₂ content is increased from 3% to 11% in the CaO–SiO₂–Al₂O₃-based slag. Viscosity is reduced from 0.854 to 0.241 Pa·s, viscous the flow activation energy is reduced from 157.74 to 114.34 kJ·mol⁻¹, and the break temperature is reduced from 1280 to 1180 °C. Furthermore, when the CaF₂ content is increased from 3% to 11%, the number of nonbridging fluorine bonds (Al–F structure and Si–F structure) in the melt increases to 287, the number of bridging fluorine bonds (Al–F–Al structure, Si–F–Si structure, and Si–F–Si structure) is only 17, and the network rupture of fluorine ions in the system is larger than the network formation. Consequently, both the degree of polymerization and viscosity are reduced.

Longitudinal crack defects are a frequent occurrence on the surface of thin slabs during the high-speed continuous casting process. Therefore, this study undertakes a detailed analysis of the solidification characteristics of hyper-peritectic steel thin slabs. By establishing a three-dimensional heat transfer numerical model of the thin slab, the formation mechanism of longitudinal cracks caused by uneven growth of the initial shell is determined. Based on the mechanism of longitudinal crack formation, by adjusting the performance parameters of the mold flux, the contradiction between the heat transfer control and lubrication improvement of the mold flux is fully coordinated, further reducing the incidence of longitudinal cracks on the surface of the casting thin slab. The results show that, using the optimized mold flux, the basicity increases from 1.60 to 1.68, the F^- mass fraction increases from 10.67% to 11.22%, the Na₂O mass fraction increases from 4.35% to 5.28%, the Li₂O mass fraction increases from 0.68% to 0.75%, and the carbon mass fraction reduces from 10.86% to 10.47%. The crystallization performance and rheological properties of the mold flux significantly improve, reducing the heat transfer performance while ensuring the lubrication ability of the molten slag. After optimizing the mold flux, a surface detection system was used to statistically analyze the longitudinal cracks on the surface of the casting thin slab. The proportion of longitudinal cracks (crack length/steel coil length, where each coil produced is about 32 m long) on the surface of the thin slab decreases from 0.056% to 0.031%, and the surface quality of the thin slab significantly improves.