[Objective] To mine the release sequence relationship in information and solve the problem of predicting information popularity on campus information platforms. [Methods] This paper proposes a new dual-layer GRU model for predicting articles popularity. The model incorporates mechanisms such as time interval awareness, author reputation awareness, and sequence window awareness. By leveraging a dual-layer GRU network, the model learns the popularity features of articles and predicts their popularity. Based on this, the calculated popularity values are used to rank and identify the popular articles within a recent time period. [Results] Using a dataset from a campus information platform as an example, the dual-layer GRU model outperforms ARIMA, Stacked LSTM, and BiLSTM-Attention models with the lowest MSE (0.000038), RMSE (0.0059), MAE (0.0043), and highest accuracy (93.04%).  [Limitations] The prediction effect of the model is limited when dealing with information without clear temporal characteristics.  [Conclusion] The introduction of the dual-layer GRU model improves the prediction effect of information popularity models.