Shipping industry is the backbone of global trade. However, the large quantities of greenhouse gas emissions from shipping, such as carbon dioxide (CO₂), cannot be ignored. In order to comply with the international environmental regulations as well as to increase commercial profits, shipping companies have stronger motivations to improve ship energy efficiency. In this study, a two-stage ship fuel consumption prediction and reduction model is proposed for a dry bulk ship. At the first stage, a fuel consumption prediction model based on random forest regressor is proposed and validated. The prediction model takes into account ship sailing speed, total cargo weight, and sea and weather conditions and then predicts hourly fuel consumption of the main engine. The mean absolute percentage error of the random forest regressor is 7.91%. At the second stage, a speed optimization model is developed based on the prediction model proposed at the first stage while guaranteeing the estimated arrival time to the destination port. Numerical experiment on two consecutive-8-day voyages shows that the proposed model can reduce ship fuel consumption by 2–7%. The reduction in ship fuel consumption will also lead to lower CO₂ emissions.

航运业是全球贸易的支柱。但是，运输产生的大量温室气体排放，例如二氧化碳（CO ₂），不能忽略。为了遵守国际环境法规并增加商业利润，船公司有更强的动机来提高船的能源效率。在这项研究中，提出了一种用于干散货船的两阶段船舶油耗预测和减少模型。在第一阶段，提出并验证了基于随机森林回归的油耗预测模型。该预测模型考虑了船舶航行速度，总货物重量以及海洋和天气状况，然后预测了主机每小时的油耗。随机森林回归变量的平均绝对百分比误差为7.91％。在第二阶段 在第一阶段提出的预测模型的基础上，开发了速度优化模型，同时保证了估计到达目标端口的时间。在连续两个为期8天的航行中进行的数值实验表明，该模型可以将船舶燃油消耗降低2–7％。船舶燃油消耗的减少也将导致二氧化碳的减少₂排放。