The machine learning algorithm is one of the key technical means for the research of intelligent construction technology of full-face tunnel boring machine, which needs a large number of comprehensive and detailed data-based geological information. While, the geological information such as the map, table and text description of geotechnical investigation report cannot be directly identified by machine learning algorithm. Hence, a datafication method of multisource geological information is put forward, which operates CAD and excel by programming, uses VBA vector graphic recognition technology, takes the given reference line as the positioning basis, automatically identifies the strata that the tunnel passes through in the CAD geological profile vector map, and integrates with the multi-source geological information in the geotechnical investigation report, so as to obtain the opportunity data-based geological information that can be identified by machine learning algorithm. Compared with the traditional manual method, the method mentionedabove can greatly improve the efficiency and accuracy of the data, and can flexibly adjust the data density to meet different needs. The method has been applied in the research of intelligent auxiliary driving of full-face tunnel boring machine, and good effects have been achieved.