论文部分内容阅读
将主成分分析和支持向量机回归相结合,以广西5、6月区域平均日降水量作为预报对象,进行区域日降水量预测研究.首先,整理分析大量的T213数值预报产品信息数据进行主成分分析,得到主成分数据序列;其次,根据主成分数据序列建立训练集训练支持向量机,并利用遗传算法优化参数;最后,输入支持向量机所需数据,得到主成分预测结果,建立广西日降水预报模型.实例计算结果表明,支持向量机回归模型比逐步回归模型有更好的预测能力.
Based on the combination of principal component analysis and support vector machine regression, regional daily average precipitation in May and June of Guangxi was used as a forecast object to study the regional daily precipitation forecasting.Firstly, a large number of T213 numerical forecasting product information data were collated and analyzed for principal components Finally, input the data needed by SVM to get the prediction result of principal components, and establish the daily precipitation of Guangxi at the same time.Secondly, the training set training vector machine is established based on the principal component data sequence, and the genetic algorithm is used to optimize the parameters. Forecasting model.Example calculation results show that SVM regression model has better predictive ability than the stepwise regression model.