摘要
由于当前的船舶分类较为单一,不同类型船舶的尺度和航行油耗等特征参数存在很大差异,采用相同的油耗标准衡量不同类型船舶的油耗会产生很大偏差。为有效解决该问题,以某公司的干散货船、集装箱船和油船为研究对象,提出一种“两步式”船型分类方法。采用K-Means算法对该公司内部船舶的9个属性进行分类,并基于肘部法则确定分类数量;根据得到的簇的数量,采用K-Means模型对船舶进行分类,并打上分类标签。针对该公司外部船舶属性数据缺失严重、数据质量较差的情况,基于上述分类标签,采用XG-Boost算法对该公司内部的船舶进行二次训练,使船舶分类模型具有处理数据缺失问题和提供分类概率的能力。实际应用结果表明,该“两步式”船型分类方法能对公司内外船舶能耗表现一致的船舶进行合理分类,并建立公司内外船舶的映射关系。
Ships characterized as same type can be significantly different in technical parameters.For reasonably evaluate ship's fuel efficiency,they should be divided in a better way.This paper studies the way of grouping ships belonging to a shipping company for fuel efficiency evaluation.The ships involved covering bulk carriers,container ships,and tankers.A"two-step"ship grouping method is proposed.The unsupervised machine learning algorithm weighted K-Means is employed to divide the company's ships into groups according to nine ship attributes of the vessels,and the number of clusters is determined using the elbow method.Based on the obtained cluster number,the K-Means model is applied again to group the ships and assign category labels.The ship data are further processed using the supervised machine learning algorithm weighted XG-Boost according to the above categories to gain the capability of handling incomplete/bad quality data and providing categorizing probability.The enhanced model is used to process the ships belonging to other companies,the accurate attributes of which may be available.The effectiveness of the proposed method is verified through actual fuel consumption estimation.
作者
王绍函
韩懿
王翔宇
任飞扬
WANG Shaohan;HAN Yi;WANG Xiangyu;REN Feiyang(Shanghai Ship and Shipping Research Institute Co.,Ltd.,Shanghai 200135,China;COSCO SHIPPING Technology Co.,Ltd.,Shanghai 200135,China;School of Naval Architecture,Ocean and Civil Engineering,Shanghai Jiao Tong University,Shanghai 200241,China)
出处
《上海船舶运输科学研究所学报》
2023年第3期28-34,53,共8页
Journal of Shanghai Ship and Shipping Research Institute
