基于集合卡尔曼滤波的短期非线性波浪场预测  

Short-term Nonlinear Wave Field Prediction Based on Ensemble Kalman Filter

作　　者：张林风 刘曾 Zhang Linfeng;Liu Zeng(School of Naval Architecture and Ocean Engineering,Huazhong University of Science and Technology,Wuhan 430070,China;Hubei Engineering Research Center of Ship Data Technology and Support Software,Wuhan,430074,China)

机构地区：[1]华中科技大学船舶与海洋工程学院,武汉430074 [2]船舶数据技术与支撑软件湖北省工程研究中心,武汉430074

出　　处：《水动力学研究与进展（A辑）》2024年第3期463-471,共9页Chinese Journal of Hydrodynamics

基　　金：国家自然科学基金(12072126)。

摘　　要：采用非线性波浪模型进行波浪场预测时,初始数据误差会导致模型预报精度逐渐下降。该文采用集合卡尔曼滤波(Ensemble Kalman Filter,EnKF)方法,结合伪谱傅里叶-勒让德(Pseudo-spectral Fourier Legendre,PFL)波浪模型在二维非线性波浪场上开展了数值模拟,有效提高了模型在初始数据受扰动情况下的预测精度。首先,基于波浪理论及典型海浪谱生成波浪场初始数据;随后,引入符合一定空间分布相关性的随机扰动,完成波浪场初始化;其后,对比结合EnKF与PFL模型与单独使用PFL模型的预测误差迭代情况,验证该文提出方法的有效性;最后,通过控制变量法探讨各同化参数的选取对同化效果的影响。该研究成果表明:利用集合卡尔曼滤波方法可有效提高波浪短期预测效果。When using nonlinear wave models for wave field prediction,the initial data errors may lead to a gradual decrease in prediction accuracy.In this paper,the Ensemble Kalman Filter(EnKF)method is combined with the Pseudo-spectral Fourier Legendre(PFL)wave model to conduct numerical simulations for two dimensional nonlinear wave fields,which effectively improves the predictive performance of the model under initial data disturbance.Firstly,nonlinear wave field data is generated based on wave theory and typical wave spectrum.Then,random perturbations which conform to a certain spatial distribution correlation are added to these data for initialization.Next,comparing the prediction error iteration of the method combining EnKF and PFL models with the method using PFL models alone to verify the effectiveness of the proposed method.Finally,the influence of various assimilation parameters on the assimilation effect is explored through a variable-controlling approach.The results show that the EnKF can effectively improve the short-term prediction of nonlinear waves.

关 键 词：非线性波浪预测 集合卡尔曼滤波 数据同化 同化参数选取 

分 类 号：O352[理学—流体力学]