基于支持向量机的船舶横摇运动辨识建模研究

项目名称： 基于支持向量机的船舶横摇运动辨识建模研究

项目编号： No.51509193

项目类型： 青年科学基金项目

立项/批准年度： 2016

项目学科： 其他

项目作者： 徐锋

作者单位： 武汉第二船舶设计研究所

项目金额： 16万元

中文摘要： 船舶横摇运动建模一直是船舶与海洋工程领域的研究难点和热点之一，本项目拟使用支持向量机（Support Vector Machines, SVM）算法，结合随机减量技术，在分析随机波浪中船舶横摇数据的基础上，建立精确的船舶横摇运动数学模型。该技术仅通过船舶航行过程中在线实测得到的横摇角时历，即可辨识得到船舶横摇运动数学模型中包含的横摇阻尼、恢复力矩、固有频率和横摇激励等参数，进而可以对船舶的横摇稳性进行在线计算，为评价船舶在航行中的安全性提供依据。本项目拟通过仿真试验、模型试验和实船试验数据对提出的辨识技术的有效性进行广泛验证，并开发快速、高效的SVM算法，深入研究门槛值和样本长度对随机减量曲线的影响，致力于开发可应用于实船的横摇运动在线辨识建模程序。本项目的成功实施，将为船舶横摇运动建模提供新的有效手段。

中文关键词： 船舶横摇运动；系统辨识；支持向量机；随机减量技术

英文摘要： Dynamic modeling of ship rolling motion is one of the most difficult and focused problem in the research area of shipbuilding and ocean engineering. This project aims to establish an accurate mathematical model of ship rolling motion based on analyzing the rolling data in random waves by using support vector machines in combination with random decrement technique. With the proposed identification approach, the parameters including the rolling damping, restoration, natural frequency, rolling excitation and so forth can be obtained via the analysis of the real-time rolling data for a sailing ship. Thereafter the transverse stability can be determined, and the results can be used as the safety evaluation for ships. The proposed identification approach will be extensively validated by using experimental data including simulation tests, model tests, and full-scale trials. Besides, to design an on-line identification modeling program of ship rolling motion for real ships, fast and efficient SVM algorithm will be developed, and the influence of initial rolling angle and sample length on the random decrement curves will be also deeply studied. The successful implement of this project will provide an efficient method for the modeling of ship rolling motion.

英文关键词： ship rolling motion;system identification;support vector machines;random decrement technique