In order to optimize the effect of topic mining of online English teaching data in higher vocational colleges and improve the consistency of topics, a topic mining method of online English teaching data in higher vocational colleges based on LDA model was proposed. Relevant data were collected from online English language teaching in higher education institutions and pre-processed to provide a reliable database. An LDA model was created and a document generation process for the LDA model was designed. On this basis, the document term matrix is constructed, and the LDA model is combined to deeply mine the data theme of online English teaching in higher vocational colleges. The test results reveal that upon the implementation of the proposed method, the theme consistency score consistently surpasses that of the traditional approach. This enhanced methodology is more effective in capturing and representing the underlying themes within the online English teaching data of higher vocational colleges, thereby exhibiting significant performance advantages.