STL (stereo lithography) files are widely used model files in 3D printing field. However, when 3D model transform into STL file, not only the data redundancy and topology relation loss will occur, but also restricts further optimization and application of STL mesh models. This paper focus on the problem of the storage process of STL files in ASCII format, a fast topological reconstruction algorithm for STL files based on hash tables is proposed. This algorithm uses the modulo division method to construct a hash function, the chain address method is also used to deals with hash collision, then, point and face tables are combined to store the index value relationship between points and faces. Finally, the topological reconstruction of the STL file is achieved. The experimental results show that the reconstruction algorithm proposed can effectively removes redundant data for models of different levels of complexity. After removing redundancy, the average proportion of vertices is only 16.64%, and the average reduction in STL files storage space is 91.23%. This algorithm also balances the relationship between the space utilization rate of the topological structure and the algorithm time complexity, providing convenience for the fast layered processing of STL files. In addition, the time consumption of topological reconstruction based on four different hash functions are analyzed. The results show that the hash function designed in this paper can reduce the computation time significantly, especially dealing with complex model. In conclusion, the method proposed in this paper has excellent efficiency and stability in dealing with models of different complexity levels, and also provides optimization solutions for the next step of manufacturing printing.