In order to improve the enthusiasm in English-medium college students, so as to improve students’ reading performance in English, the GLE teaching mode is integrated into the curriculum. Firstly, a planning model is constructed, and the secondary planning is realised through the transition probability matrix; then, the training vocabulary related data are clustered through the membership function; the weights of neurons in each layer are modified by adjusting the membership function again so that the error signal is flagged. Finally, through the knowledge ontology calculation, a reinforcement learning model is obtained after obtaining the vocabulary set for training; the experiment shows that after adopting the teaching mode of this paper and the traditional teaching mode, the average score of the experimental class is improved by 22 points compared with the control class. And the significance level is 0.000<0.05; so, at the 95% confidence level, there is a significant difference in the final grades of the two classes, which shows that the GLE teaching model proposed in this paper has better teaching effect than the traditional model.