The Three Gorges new channel and the Gezhouba shipping capacity expansion project are fundamental ways of solving the problem of ship congestion at the Three Gorges-Gezhouba hydrojunction. This study develops a prediction model based on neural network to simulate and calculate the freight throughput of the Gezhouba ship lock during the construction from the aspects of implementing cargo transfer over the dam and optimizing ship lock facilities and reservoir scheduling. Then，it explores a feasible navigation guarantee scheme for alleviating the pressure of ships passing through the lock during the construction. The example calculation results show that implementing transfer over the dam reduces the need of ships to pass through the lock by about 5.2% and that optimizing the berthing facilities upstream the Gezhouba No. 1 lock and the discharge of the Three Gorges Reservoir area in dry and flood seasons increases the freight throughput capacity of the Three Gorges-Gezhouba hydrojunction by about 5%.