基于高阶动态模态分解的波浪中船舶操纵运动时序预测  

Time Series Prediction of Ship Maneuvering Motion in Waves Based on Higher Order Dynamic Mode Decomposition

作　　者：陈昌哲 邹早建[1,2] 邹璐[1,2] 刘金洲 CHEN Changzhe;ZOU Zaojiann;ZOU Lu;LIU Jinzhou(School of Naval Architecture,Ocean and Civil Engineering,Shanghai Jiao Tong University,Shanghai 200240,China;State Key Laboratory of Ocean Engineering,Shanghai Jiao Tong University,Shanghai 200240,China)

机构地区：[1]上海交通大学船舶海洋与建筑工程学院,上海200240 [2]上海交通大学海洋工程国家重点实验室,上海200240

出　　处：《中国造船》2023年第6期216-224,共9页Shipbuilding of China

基　　金：国家自然科学基金资助项目(51979164,51979165);中央高校基本科研业务费(AF0100142/005/002)。

摘　　要：船舶运动时序预测是海上自主水面船舶的关键技术之一,在现阶段存在实时性差和易受海况影响等问题。论文应用一种新颖的降阶模型,即高阶动态模态分解,对船舶在波浪中的操纵运动进行时序预测。将KCS船作为研究对象,建模所需的训练数据来自国际比较研究SIMMAN 2020专题研讨会提供的规则波中船模回转试验数据。首先采用高阶动态模态分解和标准动态模态分解分别对船舶运动进行模态分解,然后将这两种方法分解得到的主导模态用于重构船舶运动,以实现对船舶运动的时序预测。研究结果表明,高阶动态模态分解对船舶运动的预测精度高于标准动态模态分解,其计算效率和精度能满足船舶实时运动预测的要求。Time series prediction of ship motion is one of the key technologies of maritime autonomous surface ships,and there still exist some problems,such as poor real-time performance and ship motions easily affected by sea conditions.To solve these problems,a novel reduced-order model,i.e.,higher-order dynamic mode decomposition(HODMD),was applied to time series prediction of ship maneuvering motion in waves.With the KCS ship model taken as the study object,the data of turning circle maneuver in regular waves provided by the international comparative study SIMMAN2020 Workshop were utilized as the training data required for modeling.Ship motion modes in the test data were decomposed by HODMD and standard DMD.Then,the obtained dominant modes were used to reconstruct ship motion and perform time series prediction.The results show that the prediction accuracy of HODMD is higher than that of the standard DMD,and the computational efficiency and prediction accuracy of HODMD meet the requirements of real-time prediction of ship motions.

关 键 词：波浪中船舶操纵运动 高阶动态模态分解 时序预测 降阶模型 

分 类 号：U661.33[交通运输工程—船舶及航道工程]