In this paper, the evaluation system of college students’ innovation and entrepreneurship education is constructed and the indexes are assigned by combining the hierarchical analysis method. After that, PSO algorithm is introduced in the optimization of weights and thresholds of BP neural network, the neural network model using particle swarm optimization (PSO-BP) is constructed, and the process of PSO algorithm optimization of BP neural network is described. It was found that the combined weight of five indicators, namely, “examination results of innovation and entrepreneurship courses, entrepreneurial experience, participation in centralized entrepreneurship training camps, obtaining financial support from entrepreneurship funds, and participation in innovation or entrepreneurship clubs”, accounted for more than 10%, while the combined weight coefficients of the rest of the indicators were all below 0.1. Compared to the BP model, the PSO-BP model has better network performance and its training samples have higher correlation with the test samples. In addition, the PSO-BP model can be used for predicting data prediction after 9 iterations of training, and the maximum relative error between the actual value and the expected value of the model network test output is very small (<1.4272%), which makes the model ideal. After PSO optimization PSO-BP model has almost no prediction error (<0.34%), which can improve the evaluation efficiency and accuracy.