摘要
采用支持向量机对海浪要素中的有效波高进行预测,采用风场和波浪场作为学习要素,对比不同特征向量对有效波高预测结果的准确度。取台湾岛东部海区作为实验区域,使用NCEP再分析的数值模式数据作为学习样本。选用支持向量分类机,建立了4组不同特征向量的模型进行海浪有效波高的预测,并对4种模型的结果进行比较和分析。实验表明,当输入的特征向量过多或过少时,会对模型的预测结果和计算效率产生不同的影响。当使用风场和波浪场共同作为特征向量进行学习时,在该区域预测结果与模式预报结果相比更接近,相关系数将近99%,均方根误差约0.2 m。
Support vector machine(SVM) is used to predict the significant wave height, in which wind field and wave field are adopted as learning parameters, and the influence of different eigenvectors on the prediction is analyzed. The domain of the SVM is located to the southeast of the Taiwan Island and the NCEP reanalysis data are used as learning samples. By using support vector classification, we built 4 models with different feature vectors and predicted the significant wave height. Results show that feature vectors can impact the accuracy and computation speed. When wind field and wave field are adopted as eigenvectors for learning, the correlation coefficient is nearly 99% and the root mean square error is about 0.2 m in comparison with numerical model simulations.
引文
[1] GROUP T W.The WAM model—a third generation ocean wave prediction model[J].Journal of Physical Oceanography,1988,18(12):1775-1810.
[2] YUAN Y L,PAN Z D,HUA F,et al.LAGFD-WAM ocean wave numerical model Ⅰ:basic physical model[J].Haiyang Xuebao,1992,14(5):1-7.袁业立,潘增弟,华锋,等.LAGFD-WAM海浪数值模式——Ⅰ:基本物理模型[J].海洋学报,1992,14(5):1-7.
[3] YUAN Y L,HUA F,PAN Z D,et al.LAGFD-WAM ocean wave model Ⅱ:territorial characteristic inlaid method and its application[J].Haiyang Xuebao,1992,14(6):12-24.袁业立,华锋,潘增弟,等.LAGFD-WAM海浪数值模式——Ⅱ.区域性特征线嵌入格式及其应用[J].海洋学报,1992,14(6):12-24.
[4] YIN B S,WANG T.A third generation shallow water wave numerical Model-YE-WAM[J].Chinese Journal of Oceanology & Limnology,1996,14(2):106-112.
[5] YANG Y Z,QIAO F L,ZHAO W,et al.MUSNUM ocean wave numerical model in spherical coordinates and its application[J].Haiyang Xuebao,2005,27(2):1-7.杨永增,乔方利,赵伟,等.球坐标系下MASNUM海浪数值模式的建立及其应用[J].海洋学报,2005,27(2):1-7.
[6] BOOIJ N,RIS R C,HOLTHUIJSEN L H.A third-generation wave model for coastal regions:1.model description and validation[J].Journal of Geophysical Research:Oceans,1999,104(C4):7649-7666.
[7] TOLMAN H L,CHALIKOV D.Source terms in a third-generation wind wave model[J].Journal of physical oceanography,1996,26(26):2497-2518.
[8] JAMES S C,ZHANG Y,O'DONNCHA F.A machine learning framework to forecast wave conditions[J].Coastal Engineering,2018,137:1-10.
[9] VAPNIK V.The nature of statistical learning theory[C]//Conference on Artificial Intelligence.Springer-Verlag,1995:988-999.
[10] CORTES C,VAPNIK V.Support-vector networks[J].Machine Learning,1995,20(3):273-297.
[11] KUBAT M,MATWIN S.Addressing the curse of imbalanced training sets:one-sided selection[C]//Proceedings of the 14th International Conference on Machine Learning.Nashville,USA,1997:179-186.
[12] BARANDELA R,VALDOVINOS R M,SáNCHEZ J S,et al.The imbalanced training sample problem:under or over sampling[C]//Proc of the Joint IAPR International Workshops on Structural,Syntactic and Statistical Pattern Recognition.Lisbon,Portugal,2004:806-814.
[13] CHAWLA N V,BOWYER K W,HALL L O,et al.SMOTE:synthetic minority over-sampling technique[J].Journal of Artificial Intelligence Research,2011,16(1):321-357.