奇偶交叉样本检核在测深数据模型构建中的应用
|
摘要
利用最小二乘支持向量机(LSSVM)算法构造稳健的海底趋势面需要对多波束测深数据进行筛选,得到反映测深数据整体变化趋势的训练样本。根据多波束测深数据采集的特点提出单个扇区测深数据奇偶交叉样本检核的方法,并利用选取的训练样本对测深数据模型进行构建,通过测试样本的交叉检核及调整核参数得到精确的模型。为了检验奇偶交叉样本检核法的有效性,选取实测的多波束测深数据进行验证,计算结果表明奇偶交叉样本检核法可以合理的选取测深训练样本,构建的测深数据模型更为有效。
It is necessary to select the multi-beam sounding data by using LS-SVM algorithm to construct robust seabed trend surface,and select training samples of the overall change trend of the reaction sounding data. In this paper,combining the characteristics of multi-beam sounding data acquisition,the method of checking odd and even cross sample of single sector sounding data is proposed,and selected training samples are used to construct the model,and the accurate data model of sounding data is obtained through cross checking of the test samples and adjustment of the kernel parameters. In order to verify the validity of odd and even cross sample checking method,the measured data of multi-beam sounding are selected. The results show that odd and even cross sample checking method can reasonably select the training samples of the sounding data,and the sounding data model is more effective.
It is necessary to select the multi-beam sounding data by using LS-SVM algorithm to construct robust seabed trend surface,and select training samples of the overall change trend of the reaction sounding data. In this paper,combining the characteristics of multi-beam sounding data acquisition,the method of checking odd and even cross sample of single sector sounding data is proposed,and selected training samples are used to construct the model,and the accurate data model of sounding data is obtained through cross checking of the test samples and adjustment of the kernel parameters. In order to verify the validity of odd and even cross sample checking method,the measured data of multi-beam sounding are selected. The results show that odd and even cross sample checking method can reasonably select the training samples of the sounding data,and the sounding data model is more effective.
引文
[1]邓乃扬,田英杰.数据挖掘中的新方法-支持向量机[M].北京:科学出版社,2006.
[2]阎辉,张学工,李衍达.支持向量机与最小二乘法的关系研究[J].清华大学学报:自然科学版,2001,41(9):77-80.
[3]陈将宏,周德强,李落清.加权最小二乘支持向量机与最小二乘问题的关系[J].湖北大学学报:自然科学版,2004,26(1):16-18.
[4]张清华,黄贤源,隋立芬.基于SVM粗差剔除的有色噪声建模[J].海洋测绘,2010,30(3):52-55.
[5]黄贤源,翟国君,黄谟涛,等.最小二乘支持向量机构造海底趋势面研究[J].海洋测绘,2010,30(3):9-12.
[6]Wang Weixing,Liu Juan.A classification method of SVM based on AFSA[J].Journal of Chongqing University of Posts and Telecommunications(Natural Science Edition),2011,23(1):91-95.
[7]李超群,郭生练,张俊.Hv-block交叉验证法在率定水文模型参数中的比较研究[J].水文,2006,26(5):25-28.
[8]汤宝平,刘文艺,蒋永华.基于交叉验证法优化参数的Morlet小波消噪方法[J].重庆大学学报,2010,33(1):1-6.
[9]陈强,蒋卫国,陈曦,等.基于支持向量回归模型的水稻田甲烷排放通量预测研究[J].环境科学,2013,34(8):2975-2982.
[10]徐富强,刘相国.基于优化的RBF神经网络的变量筛选方法[J].计算机系统应用,2012,21(3):206-208.
[11]Chapelle O,Vapnik V N,Bousquet O,et al.Choosing multiple parameters for support vector machines[J].M achine Learning,2002,46(1-3):131-159.
[12]褚小立,袁洪福,王艳斌,等.遗传算法用于偏最小二乘方法建模中的变量筛选[J].分析化学研究简报,2001,29(4):437-442.
[2]阎辉,张学工,李衍达.支持向量机与最小二乘法的关系研究[J].清华大学学报:自然科学版,2001,41(9):77-80.
[3]陈将宏,周德强,李落清.加权最小二乘支持向量机与最小二乘问题的关系[J].湖北大学学报:自然科学版,2004,26(1):16-18.
[4]张清华,黄贤源,隋立芬.基于SVM粗差剔除的有色噪声建模[J].海洋测绘,2010,30(3):52-55.
[5]黄贤源,翟国君,黄谟涛,等.最小二乘支持向量机构造海底趋势面研究[J].海洋测绘,2010,30(3):9-12.
[6]Wang Weixing,Liu Juan.A classification method of SVM based on AFSA[J].Journal of Chongqing University of Posts and Telecommunications(Natural Science Edition),2011,23(1):91-95.
[7]李超群,郭生练,张俊.Hv-block交叉验证法在率定水文模型参数中的比较研究[J].水文,2006,26(5):25-28.
[8]汤宝平,刘文艺,蒋永华.基于交叉验证法优化参数的Morlet小波消噪方法[J].重庆大学学报,2010,33(1):1-6.
[9]陈强,蒋卫国,陈曦,等.基于支持向量回归模型的水稻田甲烷排放通量预测研究[J].环境科学,2013,34(8):2975-2982.
[10]徐富强,刘相国.基于优化的RBF神经网络的变量筛选方法[J].计算机系统应用,2012,21(3):206-208.
[11]Chapelle O,Vapnik V N,Bousquet O,et al.Choosing multiple parameters for support vector machines[J].M achine Learning,2002,46(1-3):131-159.
[12]褚小立,袁洪福,王艳斌,等.遗传算法用于偏最小二乘方法建模中的变量筛选[J].分析化学研究简报,2001,29(4):437-442.