融合大气环流异常因子的径流预报研究

发布时间：2018-03-05 18:44

本文选题：径流预报　切入点：遗传算法　出处：《水力发电学报》2017年08期 　论文类型：期刊论文

【摘要】：径流预报对区域水资源开发与管理具有重要的作用,当前的研究主要聚焦在先进的算法而忽视了丰富预报因子对提高径流预报精度的贡献。本研究以泾河径流为例,将遗传算法(GA)和回归支持向量机模型耦合,建立了改进的支持向量机回归模型(GA-SVR)。预报变量在常规预报因子(降雨与蒸发)的基础上增加了对径流影响较强的大气环流异常因子。结果表明,预测变量未含大气环流异常因子的情况下,GA-SVR模型的预测精度和泛化能力皆优于神经网络模型(ANN);考虑大气环流异常因子后,GA-SVR模型预测精度进一步提高。由此说明,SVR模型耦合GA后可提高月径流的预报精度,考虑大气环流异常因子后其预测精度可进一步提高。
[Abstract]:Runoff forecasting plays an important role in the development and management of regional water resources. The current research focuses on advanced algorithms and neglects the contribution of rich forecasting factors to the improvement of runoff forecasting accuracy. The genetic algorithm (GA) is coupled with the regression support vector machine (RSVM) model. An improved support vector machine regression model (GA-SVR) was established. Based on the conventional prediction factors (rainfall and evaporation), the anomalous factors of atmospheric circulation with strong influence on runoff were added to the prediction variables. The prediction accuracy and generalization ability of GA-SVR model without atmospheric circulation anomaly factor is better than that of neural network model, and the prediction accuracy of GA-SVR model is further improved after considering the atmospheric circulation anomaly factor. The forecasting accuracy of monthly runoff can be improved by coupling GA with the model. Considering the anomalous factor of atmospheric circulation, the prediction accuracy can be further improved.
【作者单位】：西安理工大学水利水电学院西北旱区生态水利工程国家重点实验室培育基地;
【基金】：陕西省水利科技计划项目(2017slkj-19);陕西省水利科技计划项目(2016slkj-8) 国家自然科学基金(91325201) 水利部公益项目(201501058)
【分类号】：P338.2

【相似文献】