基于粒子群-支持向量机定量降水集合预报方法

首先对ECMWF不同物理量场预报因子群进行自然正交展开,选取能充分反映每个预报因子场主要信息的第一主分量作为模型输入.进一步利用粒子群算法对支持向量回归机的相关参数进行优化,以南宁市8个气象站单站逐日降水作为预报对象,建立粒子群-支持向量回归集合预报模型,进行单站逐日降水的数值预报产品释用预报方法研究.利用模型对2015年5-6月南宁市8站进行了逐日降水预报业务试验,结果表明,模型具有较好的预报效果.并提出了利用隶属函数建立可信度函数对不同的预报模型进行评价.

A New Support Vector Regression Ensemble Method Based on PSO for Quantitative Rainfall Forecasting

In this paper,the natural orthogonal expansion is utilized to select first principal component of ECMWF different physical factor groups.And the particle swarm optimization (PSO) algorithm is used to optimize the parameters of support vector regression (SVR).Further,a particle swarm optimization SVR model is set up to daily rainfall of the singlestation in May-June,2015.As shown from the model results of the forecast experiment,it is suggested that the model does well in forecasting rainfall over the 8 station of Nanning area.Finally,the confidence function is used to evaluate the different forecasting models that is established with the membership function.