Due to the worldwide energy crisis and environmental issues, the so-called green ship is recently urgent to be developed for energy-saving and emission-reduction. When a photovoltaic (PV) system is installed on a ship, solar generation can provide auxiliary energy as backup of engine. However, different from the land-based PV system with fixed position, the marine PV system always suffers from complex and frequently-changed environmental conditions, e.g. the partial and dynamic shading, so a maximum power point tracking (MPPT) approach with higher accuracy and faster response is required. In this paper, a novel configuration of large-scale PV array on green ocean-going ship is studied, and its MPPT problem is described as a large-scale optimization model. Then, a meta-heuristic optimization is employed to solve this model offline, and the model predictive control is employed to achieve the online MPPT control in real-time. Result of simulation experiments shows that the proposed configuration and MPPT approach can achieve effective performance, and also can improve the output power under complex environmental conditions significantly.