基于改进灰狼算法的船舶数学模型参数辨识  

Parameter identification of a ship mathematical model based on the modified grey wolf algorithm

作　　者：孟耀 张秀凤[1] 陈雨农 MENG Yao;ZHANG Xiufeng;CHEN Yunong(Navigation College,Dalian Maritime University,Dalian 116026,China)

机构地区：[1]大连海事大学航海学院,辽宁大连116026

出　　处：《哈尔滨工程大学学报》2023年第8期1304-1312,共9页Journal of Harbin Engineering University

基　　金：国家自然科学基金项目(51779029).

摘　　要：为了解决船舶操纵性指数计算复杂以及辨识参数泛化性较低问题,本文利用改进灰狼算法对辨识参数进一步优化。基于大型油轮运动数据以及一阶、二阶线性响应型数学模型,利用支持向量回归得到辨识参数的参考值。当参考值泛化性较低或者不准确时,利用改进灰狼算法实现辨识参数的范围内寻优,并将所得的辨识结果与基于遗忘因子的递推最小二乘的辨识结果对比。研究表明:利用改进灰狼算法优化后得到的辨识参数结果精度较高并且具有一定的泛化性。改进灰狼算法具有较强的搜索能力,同时可以对其他算法得到的不准确的参数进一步优化,使得参数辨识值更为准确。The modified grey wolf optimizer(GWO)is used to optimize identification parameters further to acquire good prediction results for solving the problems of the complex calculation of the ship maneuverability indices and the low generalization of identification parameters.The reference values of identification parameters were obtained by using support vector regression based on the motion data of large oil tanker and first and second-order linear ship response models.When the generalization of the reference value is low or inaccurate,the improved GWO algorithm is used to optimize identification parameters within a certain range.Moreover,the identification results provided by the proposed algorithm are compared with the identification results acquired through recursive least squares based on the forgetting factor.The simulation prediction results show that the identification parameter results obtained by the modified GWO have high accuracy and generalization.Results indicate that the modified GWO algorithm has strong searchability and can further optimize the inaccurate parameters given by other algorithms,increasing the accuracy of parameter identification.

关 键 词：船舶响应型数学模型 参数辨识 船舶操纵性指数 支持向量回归 改进灰狼算法 基于遗忘因子的递推最小二乘 辨识参数优化 泛化性验证 

分 类 号：U661.3[交通运输工程—船舶及航道工程]