Functional words, as an important component of Tibetan literature, has caused great difficulties in the process of document recognition. Based on the traditional Tibetan grammar and grammar rules, this paper mainly studies and puts forward three kinds of recognition algorithms for a large number of Tibetan free function words in Tibetan historical documents, and establishes a rule base of 284 rules for Tibetan free function words disambiguation. In the process of digitalization, the ambiguity of unfree function words in Tibetan sentences can be quickly identified and eliminated, and the accuracy of automatic identification of Tibetan documents can be improved.