基于预报因子聚类分级的日径流预报深度信念模型及应用
|
摘要
径流预报是水库运行调度的重要决策依据,提高入库径流预报精度,对水库优化调度和水资源高效利用具有重要意义.本文提出一种基于大样本数据分级策略的深度信念网络模型(Deep Belief Networks,DBN),以龙羊峡入库径流预报为例,采用Fuzzy C-means(FCM)聚类方法,将总样本训练集分为不同训练样本子集;不同样本子集下,对不同预报因子(只考虑降雨、考虑不同时期的降雨及同时考虑降雨及前期径流),分别建立DBN模型和人工神经网络模型(Artificial Neural Network,ANN),分析样本分级和考虑不同因子情况下不同模型的预报结果.结果表明:与不考虑预报因子分级的预报模型相比,基于时间序列聚类的预报模型显著提高了径流预报的精度;通过FCM聚类,将样本分为3类,考虑降雨及前期径流作为预报因子进行分级时,比只考虑降雨、考虑不同时期的降雨时建立的预报模型的预测精度更高.用该模型进行龙羊峡水库入流日径流预报,提高了预报精度,可为龙羊峡水库调度提供决策支持.
Streamflow forecasting is important for optimal reservoir operation and management,and improving the efficiency of water utilization. In this paper,we propose a predictors classification strategy with large amounts of samples,that is,Deep Belief Networks( DBN) based on Fuzzy Cmeans( FCM) clustering method. Daily streamflow forecasting of Longyangxia reservoir was taken as the case study. According to FCM clustering method,the total dataset is divided into different sub-datasets. Considering different forecasting factors( only rainfall,rainfall in different periods,both rainfall and previous streamflow),DBN and ANN models were built,and the results were compared to analyze the impact of samples classification and different forecasting factors on the accuracy of the forecasting models. Results showed that the forecasting models based on predictor classification significantly improve the accuracy of streamflow forecasting compared to the traditional forecasting models without samples classification. The total samples were classified into three categories by FCM clustering,and the performance of the models considering the rainfall and previous streamflow as the predictors is better than those of the models considering only the rainfall or rainfall in different periods. Streamflow forecasting in Longyangxia reservoir has been improved the accuracy and provide the decision support for the scheduling of Longyangxia reservoir.
Streamflow forecasting is important for optimal reservoir operation and management,and improving the efficiency of water utilization. In this paper,we propose a predictors classification strategy with large amounts of samples,that is,Deep Belief Networks( DBN) based on Fuzzy Cmeans( FCM) clustering method. Daily streamflow forecasting of Longyangxia reservoir was taken as the case study. According to FCM clustering method,the total dataset is divided into different sub-datasets. Considering different forecasting factors( only rainfall,rainfall in different periods,both rainfall and previous streamflow),DBN and ANN models were built,and the results were compared to analyze the impact of samples classification and different forecasting factors on the accuracy of the forecasting models. Results showed that the forecasting models based on predictor classification significantly improve the accuracy of streamflow forecasting compared to the traditional forecasting models without samples classification. The total samples were classified into three categories by FCM clustering,and the performance of the models considering the rainfall and previous streamflow as the predictors is better than those of the models considering only the rainfall or rainfall in different periods. Streamflow forecasting in Longyangxia reservoir has been improved the accuracy and provide the decision support for the scheduling of Longyangxia reservoir.
引文
[1]覃光华,丁晶,李眉眉,等.敏感型人工神经网络及其在水文预报中的应用[J].水科学进展,2003,14(2):163-166Qin Guanghua,Ding Jing,Li Meimei,et al. Application of an ANNs with sensitive ability to hydrologic forecast[J].Advances in Water Science,2003,14(2):163-166
[2]韩义超,徐炜,张弛,等.径流贝叶斯概率预报在水库发电优化调度中的应用[J].水利水电科技进展,2014,34(4):39-45Han Yichao,Xu Wei,Zhang Chi,et al. Building optimization model for reservoir hydropower generation using Bayesian forecasting system and stochastic dynamic programming[J]. Advances in Science and Technology of Water Resources,2014,34(4):39-45
[3]缪益平,邓俊.基于BP人工神经网络的枯水径流预报方案研究[J].水文,2008,28(3):33-36Miao Yiping,Deng Jun. Research on low-flow forecasting scheme based on BP artificial neural network[J]. Journal of China Hydrology,2008,28(3):33-36
[4]赵铜铁刚,杨大文,蔡喜明,等.基于随机森林模型的长江上游枯水期径流预报研究[J].水力发电学报,2012,31(3):18-24Zhao Tongtiegang,Yang Dawen,Cai Ximing,et al. Predict seasonal low flows in the upper Yangtze River using random forests model[J]. Journal of hydroelectric engineering,2012,31(3):18-24
[5]何振奇,乔光建.基于多项式回归模型的枯季径流预报与分析[J].南水北调与水利科技,2010,8(5):85-88He Zhenqi,Qiao Guangjian. Polynomial regression model based on the low flow forecasting and analysis[J]. South-toNorth Water Transfers and Water Science&Technology,2010,8(5):85-88
[6]王文圣,熊华康,丁晶.日流量预测的小波网络模型初探[J].水科学进展,2004,15(3):382-386Wang Wensheng,Xiong Huakang,Ding Jing. Study on wavelet network model and its application to the prediction of daily discharge[J]. Advances in Water Science,2004,15(3):382-386
[7] Moradkhani H,Hsu K,Gupta H V,et al. Improved streamflow forecasting using self-organizing radial basis function artificial neural networks[J]. Journal of Hydrology,2004,295(1):246-262
[8]冯小冲,王银堂,刘勇,等.基于物理统计方法的丹江口水库月入库径流预报[J].河海大学学报:自然科学版,2011,39(3):242-247Feng Xiaochong,Wang Yintang,Liu Yong,et al. Monthly runoff forecast for Danjiangkou Reservoir based on physical statistical methods[J]. Journal of Hohai University(Natural Sciences),2011,39(3):242-247
[9] Sivapragasam C,Liong S Y. Flow categorization model for improving forecasting[J]. Hydrology Research,2005,36(1):37-48
[10]罗向龙,焦琴琴,牛力瑶,等.基于深度学习的短时交通流预测[J].计算机应用研究,2017,34(1):91-93Luo Xianglong,Jiao Qinqin,Niu Liyao,et al. Short-term traffic flow prediction based on deep learning[J]. Application Research of Computers,2017,34(1):91-93
[11] Chang X,Wang Q,Liu Y,et al. Sparse regularization in fuzzy c-means for high-dimensional data clustering[J]. IEEE Transactions on Cybernetics,2017,47(9):2616-2627
[12]徐学良,易鹏飞,王海滨,等.基于FCM聚类算法的呼吸音分类识别研究[J].现代电子技术,2015,38(12):16-19Xu Xueliang,Yi Pengfei,Wang Haibin,et al. Study on breath sound recognition based on FCM clustering algorithm[J]. Modern Electronic Technology,2015,38(12):16-19
[13] Bezdek J C,Ehrlich R,Full W. FCM:The fuzzy c-means clustering algorithm[J]. Computers&Geosciences,1984,10(2-3):191-203
[14]汪丽娜,陈晓宏,李粤安,等.基于人工鱼群算法和模糊C-均值聚类的洪水分类方法[J].水利学报,2009,40(6):743-748Wang Lina,Chen Xiaohong,Li Yue'an,et al. Method for flood classification based on fuzzy C-mean clustering and artificial fish swarm algorithm[J]. Journal of Hydraulic Engineering,2009,40(6):743-748
[15] Mahela O P,Shaik A G. Power quality recognition in distribution system with solar energy penetration using S-transform and Fuzzy C-means clustering[J]. Renewable Energy,2017,106:37-51
[16] Hinton G E,Osindero S,Teh Y W. A fast learning algorithm for deep belief nets[J]. Neural Computation,2006,18(7):1527-1554
[17] Wang H Z,Wang G B,Li G Q,et al. Deep belief network based deterministic and probabilistic wind speed forecasting approach[J]. Applied Energy,2016,182:80-93
[18] Mohamed A,Dahl G E,Hinton G. Acoustic modeling using deep belief networks[J]. IEEE Transactions on Audio,Speech,and Language Processing,2012,20(1):14-22
[19] Karabulut E M,Ibrikci T. Discriminative deep belief networks for microarray based cancer classification[J].Biomedical Research,2017,28(3):1016-1024
[20]沈延青,靳少波,白钰.黄河上游唐乃亥水文站与龙羊峡水电站水量不平衡原因分析[J].电网与清洁能源,2014,30(1):117-119Sheng Yanqing,Jin Shaobo,Bai Yu,Analysis of the causes of water unbalances at Tangnaihai hydrological station and Longyangxia hydropower station in the upper reaches of the Yellow river[J]. Power System and Clean Energy,2014,30(1):117-119
[21]史玉品,康玲玲,王金花,等.近期黄河上游气候变化对龙羊峡入库水量的影响[J].水利水电科技进展,2005,25(4):5-8Shi Yupin,Kang Lingling,Wang Jinhua,et al,Climate variation in upper Yellow River in 1990s and its influences on inflow of Longyangxia reservoir[J]. China Rural Water and Hydropower,2005,25(4):5-8
[22]巴存胜,赵建英.龙羊峡水库入库径流特征及周期分析[J].青海电力,2003,(4):4-6Ba Cunsheng,Zhao Jianying. Analysis of Longyangxia reservoir's entering radial flow character and period[J]. Qinghai Electric Power,2003,(4):4-6
[23]张昂,李铁键,傅汪,等.基于卫星降水融合产品的黄河源区汛期径流模拟[J].应用基础与工程科学学报,2017,25(1):1-16Zhang Ang,Li Tiejian,Fu Wang,et al. Model simulation of flood season runoff in the headwaters of the Yellow River basin using satellite-ground merged precipitation data[J]. Journal of Basic Science and Engineering,2017,25(1):1-16
[2]韩义超,徐炜,张弛,等.径流贝叶斯概率预报在水库发电优化调度中的应用[J].水利水电科技进展,2014,34(4):39-45Han Yichao,Xu Wei,Zhang Chi,et al. Building optimization model for reservoir hydropower generation using Bayesian forecasting system and stochastic dynamic programming[J]. Advances in Science and Technology of Water Resources,2014,34(4):39-45
[3]缪益平,邓俊.基于BP人工神经网络的枯水径流预报方案研究[J].水文,2008,28(3):33-36Miao Yiping,Deng Jun. Research on low-flow forecasting scheme based on BP artificial neural network[J]. Journal of China Hydrology,2008,28(3):33-36
[4]赵铜铁刚,杨大文,蔡喜明,等.基于随机森林模型的长江上游枯水期径流预报研究[J].水力发电学报,2012,31(3):18-24Zhao Tongtiegang,Yang Dawen,Cai Ximing,et al. Predict seasonal low flows in the upper Yangtze River using random forests model[J]. Journal of hydroelectric engineering,2012,31(3):18-24
[5]何振奇,乔光建.基于多项式回归模型的枯季径流预报与分析[J].南水北调与水利科技,2010,8(5):85-88He Zhenqi,Qiao Guangjian. Polynomial regression model based on the low flow forecasting and analysis[J]. South-toNorth Water Transfers and Water Science&Technology,2010,8(5):85-88
[6]王文圣,熊华康,丁晶.日流量预测的小波网络模型初探[J].水科学进展,2004,15(3):382-386Wang Wensheng,Xiong Huakang,Ding Jing. Study on wavelet network model and its application to the prediction of daily discharge[J]. Advances in Water Science,2004,15(3):382-386
[7] Moradkhani H,Hsu K,Gupta H V,et al. Improved streamflow forecasting using self-organizing radial basis function artificial neural networks[J]. Journal of Hydrology,2004,295(1):246-262
[8]冯小冲,王银堂,刘勇,等.基于物理统计方法的丹江口水库月入库径流预报[J].河海大学学报:自然科学版,2011,39(3):242-247Feng Xiaochong,Wang Yintang,Liu Yong,et al. Monthly runoff forecast for Danjiangkou Reservoir based on physical statistical methods[J]. Journal of Hohai University(Natural Sciences),2011,39(3):242-247
[9] Sivapragasam C,Liong S Y. Flow categorization model for improving forecasting[J]. Hydrology Research,2005,36(1):37-48
[10]罗向龙,焦琴琴,牛力瑶,等.基于深度学习的短时交通流预测[J].计算机应用研究,2017,34(1):91-93Luo Xianglong,Jiao Qinqin,Niu Liyao,et al. Short-term traffic flow prediction based on deep learning[J]. Application Research of Computers,2017,34(1):91-93
[11] Chang X,Wang Q,Liu Y,et al. Sparse regularization in fuzzy c-means for high-dimensional data clustering[J]. IEEE Transactions on Cybernetics,2017,47(9):2616-2627
[12]徐学良,易鹏飞,王海滨,等.基于FCM聚类算法的呼吸音分类识别研究[J].现代电子技术,2015,38(12):16-19Xu Xueliang,Yi Pengfei,Wang Haibin,et al. Study on breath sound recognition based on FCM clustering algorithm[J]. Modern Electronic Technology,2015,38(12):16-19
[13] Bezdek J C,Ehrlich R,Full W. FCM:The fuzzy c-means clustering algorithm[J]. Computers&Geosciences,1984,10(2-3):191-203
[14]汪丽娜,陈晓宏,李粤安,等.基于人工鱼群算法和模糊C-均值聚类的洪水分类方法[J].水利学报,2009,40(6):743-748Wang Lina,Chen Xiaohong,Li Yue'an,et al. Method for flood classification based on fuzzy C-mean clustering and artificial fish swarm algorithm[J]. Journal of Hydraulic Engineering,2009,40(6):743-748
[15] Mahela O P,Shaik A G. Power quality recognition in distribution system with solar energy penetration using S-transform and Fuzzy C-means clustering[J]. Renewable Energy,2017,106:37-51
[16] Hinton G E,Osindero S,Teh Y W. A fast learning algorithm for deep belief nets[J]. Neural Computation,2006,18(7):1527-1554
[17] Wang H Z,Wang G B,Li G Q,et al. Deep belief network based deterministic and probabilistic wind speed forecasting approach[J]. Applied Energy,2016,182:80-93
[18] Mohamed A,Dahl G E,Hinton G. Acoustic modeling using deep belief networks[J]. IEEE Transactions on Audio,Speech,and Language Processing,2012,20(1):14-22
[19] Karabulut E M,Ibrikci T. Discriminative deep belief networks for microarray based cancer classification[J].Biomedical Research,2017,28(3):1016-1024
[20]沈延青,靳少波,白钰.黄河上游唐乃亥水文站与龙羊峡水电站水量不平衡原因分析[J].电网与清洁能源,2014,30(1):117-119Sheng Yanqing,Jin Shaobo,Bai Yu,Analysis of the causes of water unbalances at Tangnaihai hydrological station and Longyangxia hydropower station in the upper reaches of the Yellow river[J]. Power System and Clean Energy,2014,30(1):117-119
[21]史玉品,康玲玲,王金花,等.近期黄河上游气候变化对龙羊峡入库水量的影响[J].水利水电科技进展,2005,25(4):5-8Shi Yupin,Kang Lingling,Wang Jinhua,et al,Climate variation in upper Yellow River in 1990s and its influences on inflow of Longyangxia reservoir[J]. China Rural Water and Hydropower,2005,25(4):5-8
[22]巴存胜,赵建英.龙羊峡水库入库径流特征及周期分析[J].青海电力,2003,(4):4-6Ba Cunsheng,Zhao Jianying. Analysis of Longyangxia reservoir's entering radial flow character and period[J]. Qinghai Electric Power,2003,(4):4-6
[23]张昂,李铁键,傅汪,等.基于卫星降水融合产品的黄河源区汛期径流模拟[J].应用基础与工程科学学报,2017,25(1):1-16Zhang Ang,Li Tiejian,Fu Wang,et al. Model simulation of flood season runoff in the headwaters of the Yellow River basin using satellite-ground merged precipitation data[J]. Journal of Basic Science and Engineering,2017,25(1):1-16