Novel ship fuel consumption modelling approaches for speed and trim optimisation: Using engine data as auxiliary

被引：2

作者：

Wang, Kangli ^{[1
]}

Zhang, Defu ^{[1
]}

Shen, Zhenyu ^{[1
]}

Zhu, Wei ^{[2
]}

Ye, Hongcai ^{[3
]}

Li, Dong ^{[1
]}

机构：

[1] Tianjin Univ Technol, Maritime Coll, Tianjin, Peoples R China

[2] Northern Nav Serv Ctr, Yingkou AtoN Div, Liaoning, Peoples R China

[3] COSCO SHIPPING Bulk Co Ltd, Guangzhou, Peoples R China

来源：

OCEAN ENGINEERING | 2023年 / 286卷

关键词：

Fuel consumption model; Deep neural network; Branch structure; Operational optimisation; ENERGY EFFICIENCY;

D O I：

10.1016/j.oceaneng.2023.115520

中图分类号：

U6 [水路运输]; P75 [海洋工程];

学科分类号：

0814 ; 081505 ; 0824 ; 082401 ;

摘要：

An accurate fuel consumption model is essential for optimising ship operations. This study examined the impact of main engine data on the precision of ship fuel consumption models. When compared to models constructed without engine data, models utilising engine data can decrease prediction errors by 18.49%-31.25%. However, since speed and trim optimisation tasks necessitate control variables which are not available before the voyage, they cannot be employed as inputs to the fuel consumption model developed for the optimisation task. The first approach involves a two-stage process, while the second entails incorporating an auxiliary branch into a deep neural network. In the experimental findings of the second strategy, the mean absolute error was 0.001139, signifying a 20.9% reduction in fuel consumption model error compared to not utilising the main engine data. These strategies present novel methods for establishing precise fuel consumption models in ship operation optimisation research.

引用

页数：12