摘要
选取Daubechies(db6)小波系对陆浑灌区1970—2013年的降水量及作物需水量时间序列进行多时间尺度水平分解,以得到的1970—2003年的小波系数为输入数据、灌溉水量为输出数据,构建多尺度小波变换的小波神经网络进行预测模拟,并利用单隐层BP、双隐层BP神经网络和小波神经网络分别进行了预测比较,结果表明:基于多尺度小波变换的小波神经网络模型的模拟精度和预测的稳定性均高于其他3种模型,4种预测模型的预测值平均相对误差排序为单隐层BP神经网络>双隐层BP神经网络>小波神经网络>多尺度小波变换的小波神经网络。
Firstly,the Daubechies(db6)wavelet system was chosen to conduct multi-time scale horizontal decomposition of the precipitation and crop water requirement time series in the Luhun irrigation area from the year 1970 to 2013.Taking the obtained wavelet coefficients from 1970 to 2003 as the input data and irrigation water amount as the output data,a wavelet neural network based on multi-scale wavelet trans form was used to predict and simulate,and using single hidden layer BP,double hidden layer BP neural network and wavelet neural network respectively predict and compare.The results show that the relative error precision and the prediction stability of the wavelet neural network model with multi-scale wavelet transform are higher than that of the other three models.The average prediction relative error of these four pre diction models ranking:The single hidden layer BP neural network>double hidden layer BP neural network>wavelet neural network>multiscale wavelet transform Wavelet neural network.
作者
郭兵托
孙素艳
张金萍
李佳艺
GUO Bingtuo;SUN Suyan;ZHANG Jinping;LI Jiayi(Yellow River Engineering Consulting Co.,Ltd.,Zhengzhou 450003,China;General Institute of Water Resources and Hydropower Planning and Design,Ministry of Water Resources,Beijing 100120,China;College of Water Conservancy and Environment,Zhengzhou University,Zhengzhou 450001,China;Taiyuan Hydraulic Research Institute of Shanxi Province,Taiyuan 030002,China)
出处
《人民黄河》
CAS
北大核心
2019年第11期154-158,共5页
Yellow River
基金
国家重点研发计划项目(2017YFC0404404)
2018年度河南省高校科技创新人才支持计划项目(18HASTIT014)
天津大学水利工程仿真与安全国家重点实验室项目(HESS-1717)
郑州大学2015年优秀青年教师发展基金资助项目(1521323002)
关键词
多尺度
小波变换
小波神经网络
灌溉水量
multi-time scale
wavelet transform
wavelet neural network
irrigation water amount
作者简介
郭兵托(1978—),男,河南内黄人,高级工程师,主要从事水资源可持续利用和水利水能规划研究工作;通信作者:张金萍(1979—),女,河南郑州人,副教授,研究方向为水文学及水资源E mail:iwhrzhy@sohu.com。
