摘要
河流水位预测一直以来都是水文预报中研究的热点。河流水位变化不定,具有时间上和空间上的变化性、多维性、动态性和不确定性等,给水位预测带来了挑战。本文综合考虑河流水位时空信息,建立基于时空序列的RBF神经网络预测模型来预测河流水位。实验中预测了金沙江下游向家坝水文站的水位信息,并将实验结果与其他多种水位预测方法比较,实验结果显示基于时空序列的RBF神经网络模型在河流水位预测中具有较高精度,证明了方法的可行性。
River water level prediction is not only an important part of hydrological forecasting,but also a hot topic. It is a challenge to river water level prediction,for its level fluctuation,time and space variability,multidimensional,dy-namic and uncertainty. Considering the temporal and spatial information of river water level,this paper proposes a method based on spatio-temporal series model and RBF neural network,then predicts river water level of Xiangjiaba Station with the method. Moreover,the obtained results are compared to other forecast method. The experimental results show that the forecast method based on spatio-temporal series model and RBF neural network has the excellent performance of higher prediction precision.
出处
《城市勘测》
2016年第5期34-39,共6页
Urban Geotechnical Investigation & Surveying
基金
国家高技术研究发展计划(863计划)(2013AA010308)
关键词
河流水位预测
水文预报
时空序列
RBF神经网络
river water level prediction
hydrological forecasting
spatio-temporal series model
RBF neural net-work
