Port on-shore power retrofit is an important method to achieve“carbon peak”and“carbon neutrality”. While formulating an on-shore power retrofit plan, it is necessary to consider the government policy and the willingness of ship company to use on - shore power. However, the existing models usually default that all arriving ships use on-shore power, ignoring the independent choice of ships. Also, they usually ignore the differences of ships, without considering the refined processes such as queuing of ships and berth selection. These are the reasons that the on-shore power retrofit is“hot on the shore and cold on the ship side”. Many ports in the middle and lower Yangtze River have low utilization rates of shore power, and it is difficult for ports to recover the cost of shore power transformation. To this end, a decision-making optimization model for on- shore power retrofit is proposed, based on two-stage dynamic game theory. The model allows ships to independently decide whether to use shore power based on factors such as docking time, oil and electricity costs, port conditions, and waiting costs. Based on the algorithm, the optimal number of modified berths and the optimal shore power service fee are calculated. The numerical results of the case show that this method can achieve energy saving and emission reduction under the premise of ensuring the economic benefits of both port and the ship owners. Moreover, the impact of different government policies on the dual-carbon goal is discussed.