基于在线增量学习支持向量机的径流预测

Runoff Forecasting Based on Online Incremental Learning SVM

针对传统SVM模型及非增量SVM模型在训练过程中会产生冗余向量且效果差的问题,提出在线增量学习SVM预测模型,并利用祁县来远镇盘陀村昌源河上盘陀水文站的月径流历史资料进行的仿真测试。结果显示,在线增量学习SVM模型较传统的SVM模型有较高的预测精度。

The traditional SVM model and non -incremental SVM model in the training process willproduce redundancy vector and the effect is poor, a online incremental learning SVM prediction model was proposed, and to take historical data of Pantuo station in the Changyuan river to simu- late, that lie in Pantuo village learning SVM prediction model of Laiyuan town, Qixian. The resuhs show that Online incremental than the traditional SVM model has high prediction accuracv.