Energy efficiency evaluation of new energy vehicles is a prerequisite for monitoring vehicle performance and making timely warnings. Using the operational data of new energy vehicles taken from the real-time big data monitoring platform, the performance characteristics of new energy vehicles was summarized and an energy efficiency evaluation algorithm was constructed. Based on the self-organized mapping neural network, the shortcoming of K-means clustering which was affected by the initial point, was corrected and combined with the principal component analysis to give the results of energy efficiency rating of new energy vehicles. The effectiveness and practicability of the method are verified through practical cases.