Prediction modelling and validation of mould level fluctuation in steel continuous casting mould based on Minimal Adjustment Control method

Precise prediction and control of mould level fluctuations (MLF) are essential for minimising casting defects, which significantly affect the micro structure and mechanical properties of steel. Stable MLF ensures uniform solidification, reduces internal defects, and contributes to efficient and high-quality continuous casting processes. This study presents a novel MLF model, based on Minimal Adjustment Control (MAC), to enhance prediction accuracy and control precision. By integrating actual production data with simulation models, the proposed model incorporates the complex dynamics of liquid steel flow and adapts to practical production conditions, providing a more robust prediction compared to traditional models. Non-linear fitting methods estimate model parameters, and cross-validation ensures reliability. The model's superior performance is demonstrated by comparing simulated and real production data across different casting speeds and mould widths. Results show that the proposed MAC model significantly outperforms the K. Kariya model, with a fitting optimization index of 0.8503 compared to 0.4298. The MAC model also achieves superior performance in correlation coefficient, maximum square error, and mean square error, effectively capturing the relationship between operational parameters and MLF. These findings have significant implications for improving the quality and efficiency of steel production.