This study is aimed at promoting the growth and development of college students from two aspects: independent student learning and school management. Firstly, deep neural networks (DNNs) are used to build a personalized learning resource recommendation model for college students based on an online education platform. The model is tested and passed the experiment. A new comprehensive quality evaluation index system for college students is established in school management based on the traditional complete quality evaluation method. The thorough fuzzy evaluation (FCE) method and the backpropagation neural network (BPNN) model are combined to construct a comprehensive quality evaluation model for college students. The students’ comprehensive quality evaluation scores obtained using this model are compared with the actual scores through experiments. The results show that the number of learners is 50 in the constructed personalized learning resource recommendation model for college students. When the number of learning resources is 180, F 1 ‐ score = 3 is more significant than other programs. The FCE method and the comprehensive quality evaluation model constructed by BPNN scored six students. The maximum error value between the evaluation result and the actual value is only 0.039. This shows that the effect of the constructed personalized learning resource recommendation model and college students’ comprehensive quality evaluation model is excellent and practical. This provides a feasible method for improving students’ independent learning ability and school management efficiency.