基于主成分分析法和改进随机森林算法的锚泊安全性评价  

Anchoring safety evaluation based on principal component analysis and improved random forest algorithm

作　　者：张宇航 史国友[1] ZHANG Yuhang;SHI Guoyou(Navigation College,Dalian Maritime University,Dalian 116026,Liaoning,China)

机构地区：[1]大连海事大学航海学院,辽宁大连116026

出　　处：《上海海事大学学报》2025年第1期60-67,103,共9页Journal of Shanghai Maritime University

基　　金：中国海油海洋环境与生态保护公益基金(CF-MEEC/TR/2023-8)。

摘　　要：针对传统锚泊安全评价方法存在的模糊性和局限性问题,提出一种基于真实数据的客观评价方法,以提高评价的准确性和可靠性。基于国内外多个组织的锚泊事故报告以及船舶自动识别系统(automatic identification system,AIS)和船舶交通管理系统(vessel traffic services,VTS)提供的信息,建立锚泊安全评价指标体系,并构建锚泊船的小规模数据集。运用主成分分析法(principal component analysis,PCA)得出综合影响因子,再通过基于样本增量学习的随机森林(incremental sample-based random forest,ISRF)算法得出锚泊安全指数,并开发锚泊安全评估软件,使评价过程更加便捷、高效。实例验证表明,PCA与ISRF算法结合的方法不仅能准确评价锚泊安全性,还能快速得出评价结果,满足船舶在抛锚前对锚泊安全性进行快速、准确评估的要求。Aiming at the problems of fuzziness and limitation of traditional anchoring safety evaluation methods,an objective evaluation method based on real data is proposed to to improve the accuracy and reliability of evaluation.Based on the anchoring accident reports from multiple domestic and international organizations and the information provided by automatic identification system(AIS)and vessel traffic services(VTS),an evaluation index system of anchoring safety is established,along with a small-scale dataset of anchoring ships.The principal component analysis(PCA)is used to get the comprehensive influencing factors,the incremental sample-based random forest(ISRF)algorithm is used to get the anchoring safety index,and an anchoring safety evaluation software is developed to make the evaluation process more convenient and efficient.Empirical validation demonstrates that,PCA and ISRF algorithm combined method can not only accurately evaluate anchoring safety but also quickly obtain evaluation results,meeting the requirements of fast and accurate evaluation of the anchoring safety before ship anchoring.

关 键 词：锚泊安全评价 随机森林 增量学习 主成分分析法(PCA) 

分 类 号：U675.92[交通运输工程—船舶及航道工程]