Carbon dioxide capture, utilization, and storage-enhanced oil recovery (CCUS-EOR) is considered as a practical way to reduce carbon dioxide emission. It requires injecting CO2 into oil reservoirs to enhance oil recovery and permanently sequestering CO2 in a geological formation. The coupled optimization of CO2-EOR and storage is crucial to operate the CCUS-EOR project. However, the complex coupling mechanisms of CO2 flooding and geological storage are often oversimplified in the CCUS-EOR process. Moreover, the CO2-EOR is a short-term process compared with the CO2 geological storage, but the current research do not consider the difference in the time scale between CO2-EOR and geological storage when determining the optimization objective function, which is difficult to ensure the reliability of the optimization work. On the basis of the geological model of PUNQ-S3, a numerical simulation model considering the comprehensive coupling mechanisms of CO2-EOR and storage, such as CO2 miscible flooding for enhanced oil recovery, CO2 dissolution in the formation brine, mineralization of CO2 in the geological formation, and the relative permeability hysteresis effects, is established for accurate prediction of the fate of the CO2 in the CCUS-EOR project. An adaptive optimization objective function is innovatively discerned through comprehensive statistical analyses of an extensive simulation sample in this work to ensure the reasonable optimization time scales of the CO2-EOR and storage. The convolutional neural network is used as an efficient and robust proxy prediction model to extract the features of parameter variables. The particle swarm optimization algorithm is realized to achieve the coupled optimization of CO2-EOR and storage. The results show that the main controlling mechanisms change throughout the CCUS-EOR process. Further, there are timescale differences in CO2 storage and cumulative oil production at different CCUS-EOR stages. The optimization results show that the optimal solution set of cumulative oil production and CO2 storage forms an intermittent Pareto front.