As a key means to improve the quality of education and promote the development of students, the analysis and intervention of students’ behavior has attracted increasing attention from educators and researchers. Student behavior data covers many aspects of students’ campus life, such as classroom performance, after-school activities and social interaction, etc. These data exist in various forms such as text, images and videos. The purpose of this article is to explore the application of deep learning (DL) in the analysis and intervention of students’ behavior, and to build a predictive analysis model of students’ behavior based on convolutional neural network (CNN). In order to verify the application effect of the model, this article uses real student behavior data for experimental verification. The experimental results show that the student behavior prediction and analysis model based on DL can accurately predict students’ learning status and behavior trends, and provide strong decision support for educators. At the same time, compared with the traditional student behavior analysis method, the model has higher accuracy and efficiency, and can better meet the needs of actual scenes.