The current online education platform has gradually replaced the traditional teaching mode and has become an efficient teaching method. Flipping classroom is a new teaching mode under the background of the rapid development of information technology. It is also an important way of multimedia network teaching. However, compared with the traditional teaching mode, teachers in the online teaching platform cannot judge the students' psychological activities through the students' state of mind, and they can grasp the students' learning status through the teaching process. Based on this, based on the cloud computing platform, this study improves the data transmission effect and improves the algorithm according to the learning process of the online education platform. Moreover, this study combines support vector machine to construct a student state recognition system suitable for online education platform and conducts algorithm performance analysis through experiments. In addition, this study uses MKmeans algorithm, Kmeans algorithm and improved Kmeans algorithm, that is, K-mediods and Xmeans algorithm to compare the pre-processed final data sets. The research results show that the proposed algorithm is suitable for network teaching platform and has certain practical effects.