人民黄河
人民黃河
인민황하
Yellow River
2014年
2期
30-32,36
,共4页
K最近邻算法%Pareto前沿解集法%BP网络%新安江模型%XPBK神经网络模型
K最近鄰算法%Pareto前沿解集法%BP網絡%新安江模型%XPBK神經網絡模型
K최근린산법%Pareto전연해집법%BP망락%신안강모형%XPBK신경망락모형
K-nearest neighbor algorithm%Pareto solution set%BP neural network%Xinanjiang model%XPBK neural network model
为了较好地模拟流域汇流过程,采用K最近邻算法、Pareto前沿解集法等方法对传统的BP模型进行了改进,然后结合新安江模型的产流模块构建了XPBK模型,以呈村、东湾、阜平3个流域作为研究对象,对各流域进行了模拟。结果表明:①在呈村、东湾和阜平3个流域中,XPBK模型的模拟效果总体上优于新安江模型的;②呈村流域新安江模型和XPBK模型的确定性系数均值均为0.97,东湾流域、阜平流域中XPBK模型的确定性系数均值均大于新安江模型的;③XPBK模型运用于流域模拟是可行的,尤其对湿润地区的模拟精度较高。
為瞭較好地模擬流域彙流過程,採用K最近鄰算法、Pareto前沿解集法等方法對傳統的BP模型進行瞭改進,然後結閤新安江模型的產流模塊構建瞭XPBK模型,以呈村、東灣、阜平3箇流域作為研究對象,對各流域進行瞭模擬。結果錶明:①在呈村、東灣和阜平3箇流域中,XPBK模型的模擬效果總體上優于新安江模型的;②呈村流域新安江模型和XPBK模型的確定性繫數均值均為0.97,東灣流域、阜平流域中XPBK模型的確定性繫數均值均大于新安江模型的;③XPBK模型運用于流域模擬是可行的,尤其對濕潤地區的模擬精度較高。
위료교호지모의류역회류과정,채용K최근린산법、Pareto전연해집법등방법대전통적BP모형진행료개진,연후결합신안강모형적산류모괴구건료XPBK모형,이정촌、동만、부평3개류역작위연구대상,대각류역진행료모의。결과표명:①재정촌、동만화부평3개류역중,XPBK모형적모의효과총체상우우신안강모형적;②정촌류역신안강모형화XPBK모형적학정성계수균치균위0.97,동만류역、부평류역중XPBK모형적학정성계수균치균대우신안강모형적;③XPBK모형운용우류역모의시가행적,우기대습윤지구적모의정도교고。
In order to accurately simulate the hydrological processes in a watershed,this paper introduced a new hydrologic model-XAJ Pareto Back Propagation K-nearest neighbour model (XPBK). The XPBK coupled the XAJ model with traditional BP artificial neural network model. Optimiza-tion of the XPBK model parameters was achieved by using the Pareto front set of the K-nearest neighbor algorithm. The XPBK model was applied to hourly streamflow simulations in Chengcun,Dongwan and Fuping watersheds. The results indicate that:a)The performance of XPBK is superior to XAJ in all the watersheds;b)In Chengcun,an average coefficient of determination of both XPBK and XAJ is 0. 97,however,XPBK outperforms XAJ in the Dongwan and Fuping watersheds;c)XPBK model is a promising tool for simulating various hydrologic processes and it achieves higher simulation accuracy in humid regions.