In the network environment, in order to effectively evaluate the comprehensive influence of papers in different ages, this paper proposes firstly citation speed Cs, the monthly citation peak Ctop, the monthly average usage Ut and the first month discussion T1 according to the different performances of different indicators over time, and use the standardization and distribution weights to combine the evaluation indicators to build a dynamic and comprehensive evaluation system for the influence of the paper. Taking the PLOS usage index, Scopus citation index and Twitter discussion index on the PLOS website as examples, the top 10 highly cited papers with the highest citation frequency from 2011 to 2017 were selected as samples for empirical research. It is found that the multi-dimensional dynamic evaluation system effectively solves the cumulative effect of time factors. While evaluating the papers of different ages fairly and objectively, it can also effectively identify high-value new papers with short published time.