Abstract: To develop a grouting construction quality prediction model that is both highly accurate and efficient, we established a curtain grouting construction quality model based on an integration of the Rough Set Theory, Salp Swarm Algorithm, and Random Forests. The model is specifically designed for practical application in engineering projects. Comparisons were made with the SVM and BP neural network models, revealing that the proposed model achieved superior performance. Specifically, the proposed model required a mere 219.313 s for computation, and exhibited a Pearson correlation coefficient of 0.936 between predicted and measured values. Furthermore, the average absolute error, mean square error, and average absolute percentage error were measured at 0.140, 0.037, and 0.059, respectively. These findings highlight the potential of the proposed model to serve as a valuable reference for grouting construction quality control.

Key words: curtain grouting, rough set theory, salp swarm algorithm, random forest, construction quality, regression prediction