A novel fault detection methodology tailored for marine photovoltaic (PV) grid-connected systems is presented in this study, with a focus on mitigating challenges stemming from module-level faults and environmental disturbances, specifically irradiance variations and ship-induced mechanical oscillations. A tolerance-based power variation index is proposed, which considers both sensor inaccuracies—typically within ±1.5%—and transient power deviations caused by rolling and pitching, modeled within a ±0.05 kW fluctuation range. By integrating this tolerance into a dynamic threshold framework, the method improves the accuracy and reliability of fault detection in practical shipboard scenarios. A dual-loop PI control strategy is implemented to ensure stable tracking of the d-q axis current components, minimizing overshoot and ensuring fast dynamic response under fault conditions. Simulation results using a detailed Simulink model confirm that the proposed method can accurately differentiate between power drops due to irradiance variation and those caused by physical PV module faults. The approach provides a robust, real-time solution for fault-tolerant marine PV energy systems.