Aiming at the poor performance of existing algorithms in solving large-scale ship path planning problems and the lack of consideration of marine environmental factors such as eddies, a ship path planning method based on punishment pheromone ant colony optimization was proposed. Firstly, three evaluation functions were designed for the planned path: length, risk and heading. Secondly, ACO(ant colony optimization) algorithm inspired by reinforcement learning was designed to search the optimal path, which adds punishment pheromone to the traditional guidance pheromone, which can prevent ants from conducting ineffective searches. Finally, the simulation experiments of the improved algorithm under static environments demonstrate that the proposed algorithm is superior to traditional ACO, jump point search algorithm, and bi-directional search improved ACO in terms of path length, risk value and turn accumulation angle. Compared to the best metrics among these three algorithms, proposed algorithm still achieves a significant improvement in path length reduction of 6.1%, risk value reduction of 5.6%, heading accumulation angle reduction of 78.6%, and iteration number reduction of 53.3%. Especially when the mesoscale eddies and water flow are introduced, the proposed algorithm can still plan a more suitable path for ship navigation, which has positive application significance.