考虑季节因子的支持向量机径流预测模型

基于河川径流月时段具有非平稳季节性的特征，构建支持向量机的季节性样本，建立水库入库径流的季节性支持向量机预测模型。以三峡水库1950~2006年的入库径流系列为训练样本，利用季节性支持向量机预测模型对三峡水库2007~2009年的月径流量进行预测。将此预测结果与BP神经网络模型和标准支持向量机模型预测结果进行指标分析，结论显示季节性支持向量机径流预测模型准确度更高，可以用于水库入库径流的预测。

Seasonal support vector machine model of the runoff forecast

Based on the non-stationary and seasonal river monthly runoff series, a seasonal support vector machine forecast model is established by building the seasonal sample of support vector reservoir. This paper takes Three Gorges Reservoirs runoff series from 1950 to 2006 as training sample, and forecasts 2007~2009 monthly runoff by use of the seasonal support vector machine forecast model .Comparing with the standard support vector machine forecast model and the BP neural network forecast model, this model is more accurate and suitable for the reservoir runoff forecast.