气象科学2011，Vol.31Issue(2)：187-193,7.
最小二乘支持向量机在云量预报中的应用
Application of least squares support vector machine in prediction of cloud cover
摘要
Abstract
Based on WRF model products and station-observed data of January and August from 2003 to 2006, the short-term forecast models of every month by the Least Squares Support Vector Machines (LSSVM) method for total cloud cover and low cloud cover at the Taibei and Xiamen station are established by selecting the appropriate parameter and kernel function, according to different sample length. The models are tested using the data of 2007, and compared by BP-ANN. Results show that the forecasting effect of LSSVM is better than BP-ANN, the forecasting effect of models corresponding to different sample amount is acceptable, and decrease of training data number doesn't reduce the forecast accuracy of LSSVM method. The result also shows that LSSVM is an effective method for total cloud cover and low cloud cover prediction, and it is a hopeful method for other meteorological element forecasting.关键词
支持向量机/核函数/云量/预报Key words
Support vector machine/ Kernel function/ Cloud cover/ Forecast分类
天文与地球科学引用本文复制引用
胡邦辉,刘丹军,王学忠,高传智..最小二乘支持向量机在云量预报中的应用[J].气象科学,2011,31(2):187-193,7.基金项目
国家自然科学基金资助项目(40730953) （40730953）
