聚类算法在船舶能效数据挖掘中的应用  

Application of Clustering Algorithm for Data Mining in Ship Energy Efficiency

作　　者：高梓博 杜太利[1,2] 张勇[1] 黄连忠[1,2] GAO Zibo;DU Taili;ZHANG Yong;HUANG Lianzhong(Marine Engineering College,Dalian Maritime University,Dalian 116026,China;Unmanned Ship Collaborative Innovation Research Institute, Dalian Maritime University,Dalian 116026,China)

机构地区：[1]大连海事大学轮机工程学院,大连116026 [2]大连海事大学无人船协同创新研究院,大连116026

出　　处：《武汉理工大学学报（交通科学与工程版）》2019年第2期286-290,共5页Journal of Wuhan University of Technology（Transportation Science & Engineering）

基　　金：工信部高技术船舶科研计划项目(工信部联装(2014)508号);中央高校基本科研业务费专项资金(3132018306)资助

摘　　要：以船舶节能减排为数据挖掘目标,以目标船一个完整航次的转速、功率、油耗量、GPS数据为基础,通过统计分布图和高斯混合模型聚类,得到了压载航次中主机的三种运行工况.采用二分K均值聚类法对整个航次各工况的数据分别进行聚类,比较得出最佳工况点,并利用其聚类点进行多项式拟合,最终得到了功率-单位海里油耗关系曲线.通过图像得出可以得出,单位海里油耗量随着主机功率的增大有增大趋势,但当功率小到一定程度,单位海里油耗量会反而增大,抛物线的最低点为该航次最佳工况下的理论最低单位海里油耗量.Based on the speed,power,fuel consumption and GPS data of the target ship in a complete voyage,taking energy saving and emission reduction of the ship as the data mining target,three operating conditions of the main engine in the ballast voyage were obtained through statistical distribution map and Gaussian mixture model clustering.The bisecting K-means clustering method was used to cluster the data of each operating point of the entire voyage respectively,and the best operating points were obtained by comparison.Polynomial fitting was performed by using the clustering point and the power-fuel consumption per nautical mile was finally obtained.It can be concluded from the image that the oil consumption per unit sea mile has an increasing trend with the increase of the main engine power,but when the power is small to a certain extent,the oil consumption per unit sea mile will increase instead,and the lowest point of the parabola is the theoretical lowest oil consumption per unit sea mile under the optimal operation conditions of the voyage.

关 键 词：船舶能效数据 聚类 高斯混合模型 二分K均值 

分 类 号：U676.3[交通运输工程—船舶及航道工程]