优化支持向量机在降水预报中的应用

Research on Rainfall Forecasting Based on Optimistic Support Vector Machine

利用粒子群优化算法对支持向量机参数进行优化,主成分分析法对众多的降水因子进行特征提取,将取得的主成分作为优化支持向量机的特征向量建立优化支持向量机降水预测模型,并对广西降水进行预测.仿真结果表明,粒子群优化支持向量机的预测结果与降水实际值有很好的一致性,通过与其他模型进行比较,表明了模型拟合和泛化能力方面的优越性.

To increase accuracy and stability of the rainfall forecast model, the particle swarm optimization（PSO） and support vector machine（SVM） is proposed. On the basis of analyzing the influential factors of rainfall forecast, the principal component analysis（PCA） is employed to extract main features from rainfall influential factors and the principal component is used as the input of support vector machine, and the rainfall forecast model is built. In the model, the parameter is optimized by the particle swarm optimization algorithm. Finally,the model is applied to forecast the rainfall at the Liuzhou station on Guangxi and the result shows that the proposed model has higher prediction accuracy and stability, compared with the RBF network and several other models.