Abstract： A data model considering parameter fusion was presented by studying the relation among the geometric, physical and mechanical parameters of different entities during excavation. Based on C++ and the objectoriented program development method, a data structure reflecting the entity class hierarchy was introduced to develop a 3D modeling and risk prewarning software for deep excavation. By using the 3D modeling software, the geometric and mechanical model of deep excavation was unified. The software was applied to the deep excavation of  the metro station at Hanzhong Road in Shanghai, and the results show that the presented data model reflects the relation among different entities well and is effective in detecting potential risks during excavation.