基于多维信息相似度的大渡河流域月径流量预测研究  

Research on Monthly Runoff Forecasting of Dadu River Basin Based on Multi-dimensional Information Similarity

作　　者：李佳 牟时宇 朱艳军 陶思铭 胡义明[2] LI Jia;MOU Shi-yu;ZHU Yan-jun;TAO Si-ming;HU Yi-ming(Dadu River Hydropower Development Co.,Ltd.,Chengdu 610041,China;College of Hydrology and Water Resources,Hohai University,Nanjing 210098,China)

机构地区：[1]国能大渡河流域水电开发有限公司,四川成都610041 [2]河海大学水文水资源学院,江苏南京210098

出　　处：《水电能源科学》2022年第12期38-41,共4页Water Resources and Power

基　　金：国家自然科学基金项目(41730750);国能大渡河流域水电开发有限公司科技项目(CEZB200505212)。

摘　　要：基于降雨径流信息的相似性特征,构建了以最近邻抽样回归模型为核心的中长期径流预测模型。该模型以前期径流和降雨等要素为基础构建多维信息特征因子集,依据预报月份对应的特征因子集,采用一维离散Wasserstein距离法计算其与历史时期特征因子集间的相似度,并以相似度较高特征因子集对应的历史径流值作为该预报月份的预测结果,进而实现中长期月径流预报。将模型应用于大渡河流域汛期各月(6~9月)的径流预测,结果表明该模型具有较高的预测精度,其中,7、8、9月的月流量预报相对误差绝对值在10%左右。研究结果可为流域水资源精准调配和发电效益的提高提供支撑。Based on the similarity characteristics of rainfall runoff information,a medium and long-term runoff prediction model based on nearest neighbor sampling regression model was constructed.According to the previous runoff,rainfall and other elements of the model,a multi-dimensional information feature factor set was constructed.According to the feature factor set corresponding to the forecast month,the similarity between it and the feature factor set of historical period was calculated by one-dimensional discrete Wasserstein distance method.The historical runoff value corresponding to the feature factor set with high similarity was taken as the prediction result of the forecast month.Then the medium and long-term monthly runoff forecast was realized.The model was applied to the runoff prediction of each month(June-September)in the flood season of Dadu River Basin.The results show that the model has high prediction accuracy.Among them,the absolute value of the relative error of monthly flow prediction in July,August and September is about 10%.The research results can provide support for the accurate allocation of water resources and the improvement of power generation benefits.

关 键 词：月径流预报 Wasserstein距离 最近邻抽样回归模型 多维信息相似度 大渡河 

分 类 号：TV124[水利工程—水文学及水资源]