In the catalytic reforming process, aromatic yield is a standard for evaluating the production quality of the process, and studies are underway to improve productivity by optimizing the cost and energy. In particular, the activity and lifetime prediction of reforming catalysts can contribute to overall process efficiency improvement, such as product quality, productivity, and predictive maintenance. However, it is difficult to predict real-time catalyst activity and lifetime according to changes in process operation with the existing method that requires experimental data. In this study, a multiple linear regression (MLR) model and GRU model with the real process operating data are proposed for long-term plant operation and optimization in the counter-current continuous catalytic reforming. The MLR-GRU model predicts catalyst performance degradation and lifetime according to operating conditions by defining a new variable, reforming catalyst activity. The proposed model can predict the future reformate yield with an error of less than 1%. As a result of predicting the catalyst lifetime according to various operating temperatures, feed flow patterns, and feed quality, the feed flow rate had the greatest influence on the catalyst lifetime profile. In terms of the amount of produced reformate oil, the case with maximum feed rate is the worst (25.6%); on the other hand, the case with minimum feed rate is the best (+11.4%). Thus, it is important to establish an appropriate production plan of the produced reformate oil. The model proposed in this study can predict the reformate yield and lifetime, reflecting the degradation of catalyst performance according to the operating profile in real-time, which is expected to improve productivity by production scheduling, optimization, and predictive maintenance.