基于支持向量机风速订正方法的研究  

Modification of Wind Speed Based on SVM

作　　者：孔令彬[1] 赵艳茹[1] 王聚杰[1] 仝纪龙[1] 党张利 张文煜[1] 

机构地区：[1]兰州大学大气科学学院/甘肃省干旱气候变化与减灾重点实验室,兰州730000

出　　处：《西南大学学报（自然科学版）》2014年第5期194-200,共7页Journal of Southwest University(Natural Science Edition)

基　　金：国家自然科学基金资助项目(41225018);国家重大科学研究计划"973"项目(2012CB956200)

摘　　要：风速的人工观测数据和自动观测数据之间存在差异,利用支持向量机建立了一个订正模型,将风速的人工观测数据向自动观测数据订正,并利用榆中站2007年10月1日至12月1日两个观测系统测得的风速数据进行了训练和检验.结果表明:经模型订正后,每个时刻风速的两个观测数据之间的相对误差都有了明显的减小,两个数据序列之间的平均绝对误差减小了20.23%,均方根误差减小了26.47%,平均绝对百分误差减小了8.71%.Differences always exist between the manual observation data and the automatic observation data of wind speed.In this study,a model based on support vector machine（SVM）was built to modify the manual observation data to the automatic observation data.The wind speed data recorded with these two observing systems at Yuzhong Station in Lanzhou,Gansu from October 1to December 1,2007was used to practice and test in the neural network.The result indicated that the differences between wind speed’s manual observation data and its automatic observation data at each moment were significantly reduced after the modification,and the differences between these two sequences reduced by 20.23%for MAE,26.47% for RMSE and 8.71%for MAPE.

关 键 词：风速 双轨观测 差异 支持向量机 订正 

分 类 号：P413.2[天文地球—大气科学及气象学]