There can be frequency out-of-limit caused by the pre-synchronization control of the microgrid during ship berthing. Thus this paper proposes a pre-synchronization control strategy for a ship microgrid based on intelligent self-learning control. First, the phase difference between the ship microgrid and the main grid is calculated by using the phase angle of a virtual synchronous generator. Then, based on an improved dynamic linearization method, the discrete phase difference nonlinear system is transformed into a compact linearized data model. The intelligent self-learning virtual synchronous generator pre-synchronization control algorithm is designed according to the data model. Finally, the adaptive adjustment of the power frequency droop coefficient of the generator is designed to avoid the frequency out-of-limit during the pre-synchronization process. The proposed control strategy realizes smooth grid-connected pre-synchronization control and effectively avoids the frequency out-of-limit by comparison simulation. This ensures the stable operation of the asynchronous motor load. The proposed control strategy demonstrates its superiority and accuracy when applied to the ship microgrid.