The invention discloses a fuel-saving method for an intelligent route based on navigation characteristic data learning, which includes a fuel consumption model verification system, an intelligent route selection system, input information, and output information, and is characterized in that the fuel consumption model verification system includes water flow information calibration verification model, ship type verification, draft verification model, navigation speed verification model, wave information verification model, meteorological information verification model, predicted fuel consumption of past routes, and actual Fuel consumption; the beneficial effect is that the most fuel-efficient route can be intelligently selected according to the time-varying external factors in the long-distance voyage, which can not only meet the actual engineering needs of marine ships to save fuel, fill in the technical gap, but also generate large social and economic benefits.