Surrogate Model‐Based Visualization Program for Groundwater Engineering Computation

ABSTRACT In groundwater engineering practice, dewatering design and groundwater flow analysis are essential for the stability assessment of underground excavations, such as foundation pits and tunnels. Traditional groundwater engineering education often suffers from several limitations, including abstract theoretical concepts, complex modeling procedures, and insufficient experiential learning. To address these challenges, intelligent foundation pit dewatering prediction (IFPDP) simulation software was developed in this study, designed specifically for groundwater engineering education to enhance students’ understanding of seepage mechanisms, dewatering processes, and groundwater numerical simulation. The IFPDP employed MODFLOW files generated in GMS and uses FloPy as a Python interface to enable model execution and repeated simulations. In addition, a surrogate model based on the BP neural network was integrated to enhance computational efficiency and supports parameterized teaching. The system was designed to integrate numerical simulation, visualization, and interactive learning with autonomous modeling and case‐library expansion, thereby enabling students to complete the full workflow from theoretical learning to model construction and intelligent prediction. A teaching experiment and questionnaire survey were conducted involving 120 undergraduate students across different academic levels. The results indicate that the IFPDP significantly improves students’ knowledge acquisition, learning motivation, and modeling practice skills. Students widely acknowledged the software's intuitive visualization, engineering relevance, and interactive experience, while also suggesting further expansion of the case library and enhanced parameter guidance. Overall, the study demonstrates that the IFPDP provides an efficient virtual simulation platform for groundwater engineering education and shows strong potential for future application.