摘要
采用相关分析的方法确定了灌区退水量的主要影响因素,将神经网络的在线学习功能与数据库技术相结合,建立了灌区退水量动态模型,实现了模型的在线学习,以动态的模型反映灌区退水系统的动态变化,保证了退水量模型使用的长期有效性。与实测资料对比表明,模型能够较好的模拟灌区退水系统的变化,利用灌区渠首的引水量、降水量和地下水埋深资料能够较准确的对灌区的退水量进行预测。
The major factors influencing the irrigation return flow discharge are determined by means of correlation analysis, and a dynamic model simulating the variation of the flow is established by neural network method combining with database. The model possesses the function of online study so that the online update and long-term availability can be achieved and the variation of the irrigation return flow system can be dynamically simulated. The comparison of the application of the model with observation data indicates that the variation of irrigation return flow can be accurately forecasted according to the canal water diversion, precipitation and groundwater table.
出处
《水利学报》
EI
CSCD
北大核心
2006年第6期717-721,共5页
Journal of Hydraulic Engineering
基金
国家自然科学基金项目(50179030)
陕西省水资源与环境重点实验室重点科研项目(03JS041)
关键词
灌区退水量
神经网络
LM优化算法
动态模型
irrigation return-flow
neural network
Levenberg-Marquardt algorithm
dynamic model