Service life extension (SLE) of ship is becoming increasingly vital for ensuring continuous business operations and maximizing profitability, particularly in consideration of the rising costs associated with acquiring new ships. For corroded hulls, traditional SLE strategies are typically based on the surveyor experience, which not only affect the safety and continuity of ship operations but also make it difficult to guarantee an optimal SLE duration and maintenance plan. To address these challenges, this paper introduces a decision-making method based on reinforcement learning, designed to optimize SLE strategies in a more systematic and effective manner. Reinforcement learning, as a type of machine learning method, enables an agent to gradually learn the optimal strategy through continuous interaction with the environment, making adjustments to achieve a set goal. The agent is driven by a reward mechanism, receiving corresponding rewards or penalties after each interaction, guiding it to learn in the direction of maximizing cumulative rewards. The proposed method employs a two-layer reinforcement learning optimization algorithm. In the outer layer, the optimization process uses multiple shorter SLE durations as the action space, iteratively determining the optimal total SLE duration and net present value (NPV) through a series of decisions and inner layer optimizations. The inner layer optimization focuses on a specific SLE duration, developing the most effective maintenance plan and calculating the NPV within that time frame. Validation of this method on a corroded ship demonstrates its effectiveness in providing a well-optimized SLE strategy.