机器学习在各个领域以及交叉学科中有着广泛的应用。本文运用两种机器学习方法支持向量机回归SVR和BP人工神经网络,对长江流域河溶水文站2009~2014年逐日径流与日降水资料进行学习、训练以及预测。采用 效率系数和相对偏差 作为比较指标,与传统的分布式水文模型（THREW）进行比较。结果表明：THREW模型模拟效果好, 效率系数为0.609,具有清晰的物理过程和水文机理,但是模拟结果的相对偏误差 较大,数据要求较多,操作复杂。两种机器学习方法在相对偏误差 指标表现较好,都较好的模实现了对径流的预测,泛化能力较好,即具有将学习成果应用于新知识的能力。但是对数据依赖较大,数据样本容量越大,预测的结果会更理想。BP神经网络预测径流结果在相对偏误差 指标表现很好,但是在径流极大值的预测失真。本次模拟结果显示BP神经网络优于THREW模型优于SVR模拟结果。
Abstract
Machine learning has a wide range of applications in various fields and interdisciplinary subjects． In this paper,two machine learning methods,support vector regression SVＲ and BP neural networks,are used to learn,train and predict daily runoff based on the daily precipitation and runoff data of 2009 － 2014 in Yangtze Ｒiver Basin． To compare with the traditional distributed hydrological model( THＲEW) ,The Ns efficiency coefficient and percent bais PB are adopted as indicators． The results show that the THＲEW model,with clear physical process and hydrological mechanism,has good simulation effect and the Ns coefficient efficiency is 0．503． However,the PB is big,
and the data and the operation is complex． The two machine learning methods work better in relative error index,and all have good runoff prediction and generalization ability． That is the ability to apply learning outcomes to new knowledge． But they are more dependent on data,
and the larger the data sample size is,the prediction results will be fitter． The prediction of runoff results by BP neural networks performs well in index,but it is distorted in the prediction of maximum runoff． The simulation results showe that the BP neural network is superior to the
THＲEW model while the THＲEW model is better than the SVＲ simulation results．
关键词
机器学习 /
径流预测 /
THREW模型 /
SVR回归 /
BP人工神经网络 /
效率系数
{{custom_keyword}} /
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[ ] 黄宜华. 大数据机器学习系统研究进展[J]. 大数据, 2015, 1(1):28-47. (HUANG Yihua. Research progress on large data machine learning system [J]. big data, 2015, 1 (1): 28-47. (in Chinese))
[2] 雷晓云, 张丽霞, 梁新平. 基于MATLAB工具箱的BP神经网络年径流量预测模型研究——以塔城地区乌拉斯台河为例[J]. 水文, 2008, 28(1):43-46.
[3] 张利平, 王德智, 夏军,等. 相空间神经网络模型及其在水文预测中的应用[J]. 水电能源科学, 2004, 22(1):5-8.
[4] 机器学习理论、方法及应用,王雪松,2009,科学出版社。(The theory, method and application of machine learning, Wang Xuesong, 2009, Science Press. (in Chinese))
[5] 杨淑莹, 张桦. 模式识别与智能计算:MATLAB技术实现[M]. 电子工业出版社, 2015. (YANG Shuying, ZANG Hua. Pattern recognition and Intelligent Computing: MATLAB technology implementation [M]. Electronics Industry Press, 2015. (in Chinese))
[6] 机器学习理论及应用,李凡长,2009,中国科学技术大学出版社。(The machine learning theory and application, Fanchang Li, 2009, University of Science & Technology China press. (in Chinese))
[7] 刘佩瑶, 郝振纯, 王国庆,等. 新安江模型和改进BP神经网络模型在闽江水文预报中的应用[J]. 水资源与水工程学报, 2017, 28(1):40-44. (LIU Peiyao, HAO Zhenchun, WANG Guoqing, et al. Application of Xin'An river model and improved BP neural network model in hydrological prediction of Minjiang [J]. Journal of water resources and water engineering, 2017, 28 (1): 40-44. (in Chinese))
[8] 贾仰文, 王浩, 倪广恒, 等. 分布式流域水文模型原理与实践[M]. 中国水利水电出版社, 2005. (JIA Yangwen, WANG Hao, NI Guangheng, et al. Principle and practice of Distributed Watershed Hydrological Model [M]. China Water Conservancy and Hydropower Press, 2005. (in Chinese))
[9] 徐中民, 蓝永超. 人工神经网络方法在径流预报中的应用[J]. 冰川冻土, 2000, 22(4):372-375. (XU Zhongmin, LAN Yongchao. Application of artificial neural network in runoff prediction[J]. Journal of Glaciology and Geocryology, 2000, 22 (4): 372-375. (in Chinese))
[10] 蓝永超, 康尔泗, 徐中民,等. B-P神经网络在径流长期预测中的应用[J]. 中国沙漠, 2001, 21(1):97-100. (LAN Yongchao,KANG Ersi,XU Zhongmin,et al. BP neural network in the run-off long-term prediction[J]. Journal of Desert Research, 2001, 21(1):97-100.(in Chinese)）
[1 ] H. RAMAN, N. SUNILKUMAR. Multivariate modelling of water resources time series using artificial neural networks[J]. International Association of Scientific Hydrology Bulletin, 1995, 40(2):145-163.
[ 2] Gui G Q. Face Recognition Using Support Vector Machines[J]. Computer Science, 2003.
[ 3] Hearst M A, Dumais S T, Osuna E, et al. Support vector machines[J]. IEEE Intelligent Systems & Their Applications, 2002, 13(4):18-28.
[ 4] 王景雷, 吴景社, 孙景生,等. 支持向量机在地下水位预报中的应用研究[J]. 水利学报, 2003(5):122-128. (WANG Jinglei, WU Jingshe, SUN Jingsheng, et al. Research on the application of support vector machine in the prediction of groundwater level[J]. Journal of Hydraulic Engineering, 2003 (5): 122-128. (in Chinese))
[ 5] 田富强, 胡和平, 雷志栋. 基于代表性单元流域的水文模拟理论与方法[J]. 2005. (TIAN Fuqiang, HU peace, LEI Zhidong. Theory and method of hydrological simulation based on representative unit basin[J]. 2005. (in Chinese))
[ 6] Mou L, Tian F, Hu H, et al. Extension of the Representative Elementary Watershed approach for cold regions: constitutive relationships and an application[J]. Hydrology & Earth System Sciences Discussions, 2007, 4(5):565-585.
[ 7] Tian F, Li H, Sivapalan M. Model diagnostic analysis of seasonal switching of runoff generation mechanisms in the Blue River basin, Oklahoma[J]. Journal of Hydrology, 2012, s 418–419(4):136-149.
[ 8] Liu D F, Tian F Q, Hu H P, et al. Sediment simulation at Upper Sangamon River basin using the THREW model[C]// Hydrological Modelling and Integrated Water Resources Management in Ungauged Mountainous Watersheds. Proceedings of A Symposium Held at Chengdu, China, November. 2009.
[ 9] TIAN Fuqiang, HU Heping, LEI Zhindong, et al. Extension of the representative elementary watershed approach fof cold regions via explicit treatment of energy related processes[J]. Hydrology and Earth System Sciences, 2006, 10(5):619-644.
[20] 崔东文. 多隐层BP神经网络模型在径流预测中的应用[J]. 水文, 2013, 33(1):68-73. (CUI Dongwen. Application of Hidden Multilayer BP Neural Network Model in Runoff Prediction [J]. Journal of China hydrology, 2013, 33(1):68-73. (in Chinese))
[2 ] 杨洪. 改进 BP 神经网络集成模型在径流预测中的应用[J]. 水资源与水工程学报, 2014(3):213-219. (YANG Hong. Application of improved BP neural network integrated model in runoff prediction [J]. Journal of water resources and water engineering, 2014 (3): 213-219. (in Chinese))
[22] 聂敏, 刘志辉, 刘洋,等. 基于PCA和BP神经网络的径流预测[J]. 中国沙漠, 2016, 36(4):1144-1152. (NIE Min, LIU Zhihui, LIU Yang, et al., Runoff Forecast Based on Principal Component Analysis and BP Neural Network[J]. Journal of Desert Research,2016, 36(4):1144-1152. (in Chinese))
[23] 季云. BP算法及其应用实例[J]. 中国新技术新产品, 2010(9):29-29. (JI Yun. BP algorithm and its application example [J]. new technology and new products in China, 2010 (9): 29-29. (in Chinese))
[24] 明琦, 刘俊萍. 兰州站径流支持向量机预测[J]. 水力发电学报, 2010, 29(4):32-38. (MING Qi, LIU Junping. Forecast of the run-off support vector machine of LanZhou Railway Station, [J]. Journal of Hydroelectric Engineering, 2010, 29 (4): 32-38. (in Chinese))
[25] 郭俊, 周建中, 张勇传,等. 基于改进支持向量机回归的日径流预测模型[J]. 水力发电, 2010, 36(3):12-15. (GUO Jun, ZHOU Jianzhong, Zhang Yong, et al. The daily runoff prediction model based on Improved SVM regression [J]. hydroelectric power, 2010, 36 (3): 12-15. (in Chinses))
[26] 卢敏, 张展羽. 径流预测的支持向量机应用研究[J]. 中国农村水利水电, 2006(2):47-49. (Lu Min, Zhang Zhang Yu. Prediction of the runoff at Lanzhou station with a support vector machine model [J]. China Rural hydropower, 2006 (2): 47-49. (in Chinese))
[27] Doherty J. PEST: A Unique Computer Program for Model-independent Parameter Optimisation[J]. 1994.
PDF(672 KB)
鄂公网安备 42010602001274号 鄂ICP备14015464号-1