基于HHT变换的PSO-LSSVM耦合模型估算气象资料短缺地区ET0  

Estimation Method for ET_0 with PSO-LSSVM Based on HHT Transform in Data Shortage Area

作　　者：曾向辉[1] 赵礼曦 张宇[3] 

机构地区：[1]水利部国际合作与科技司,北京100053 [2]Georgia Institute of Technology, 10010th St.NWAPT 505, ATL, GA 30332 [3]中国农业大学水利与土木工程学院,北京100083

出　　处：《灌溉排水学报》2015年第2期1-6,共6页Journal of Irrigation and Drainage

基　　金：公益性行业科研专项经费项目(201401078)

摘　　要：为准确估算气象资料短缺地区参考作物腾发量,构建了一种基于HHT变换的PSO-LSSVM耦合模型,并利用新疆和田气象站2000—2009年单日数据做训练、双日数据做验证。结果表明,该模型估算ET0方法明显优于常规的PSO-LSSVM和GRNN,预测精度较二者分别提高了15.7%~85.6%和15.8%~93.7%;该方法预测ET0的气象要素重要性为Rs〉Tmax〉Tmin〉RH〉Wn,利用该方法对气象要素组合为Tmax/Tmin/RH/Wn、Tmax/RH/Wn、Tmin/Wn、Wn条件下的ET0预测,MSE分别为0.407、0.185、0.149、0.135,说明该方法可以很好地估算资料缺失地区ET0。In order to accurate estimation of reference crop evapotranspiration in area where could not get required meteorological elements,a coupling model of PSO-LSSVM was established based on HHT transform,and Chinese Hetian Xinjiang meteorological station daily data of 2000—2009were used for training,double-day data for validation.The results showed that：both single factor and multiple factors after HHT transformation for PSO-LSSVM prediction results had significantly higher accuracy rates than that of PSOLSSVM and the GRNN prediction.The prediction accuracy was improved by 15.7%-85.6%,15.8%-93.7%respectively;and the importance of prediction elements was：Rs〉Tmax〉Tmin〉RH〉Wn.When the deletion combination was Tmax/Tmin/RH/Wn,Tmax/RH/Wn,Tmin/Wnand Wn,the MSE was 0.407,0.185,0.149,0.135 respectively,which showed that this method was a good estimation method for ET0 in data shortage area.

关 键 词：参考作物腾发量 HHT变换 PSO-LSSVM 预测模型 

分 类 号：S161.4[农业科学—农业气象学]