With the wide application of cloud computing and other modern information technologies in the field of physical education, college students’ physical education teaching and training have also stepped into the intelligent era. Based on cloud computing technology, this paper designs sports prescription recommendation algorithm and constructs college students’ sports teaching and training platform. Firstly, the cloud computing step structure and student big data service framework are designed, and then the K-Means clustering algorithm improved based on simulated annealing algorithm is utilized to divide the sports status of college students into groups. Then the similarity of the class where the target is located is calculated to get the prediction score of the target for the recommended program and recommend the exercise prescription. Analyzing the algorithm recommendation accuracy on the dataset and carrying out experimental comparisons on the actual effect of the platform, it is found that the simulated annealing optimization K-Means++ clustering algorithms designed in this paper are all far more accurate than the baseline model, with accuracies of 0.829, 0.907, and 0.916, respectively. And the baseline model accuracy was below 0.8. Except for the 50-meter run and the 800-meter/1000-meter run, the performance of the experimental group in all sports qualities was better than that of the control group. This paper makes a useful exploration for enhancing the training efficiency of professional athletes in colleges and universities and improving the physical health of college students.