基于BP神经网络模型模拟塔里木河下游潜水埋深变化——以英苏断面C5井为例

Simulating Change of Groundwater Level Based on BP Neural Network Model in Lower Reaches of Tarim River——C5 Well in the Yengsu Section is Selected as an Example

本文以塔里木河下游英苏断面350 m处C5井为研究对象,分析了影响下游英苏断面的潜水埋深影响因素,通过三层BP神经网络模型模拟了潜水埋深变化.以Matlab7.0为工作平台,将2000.7-2008.12期间的英苏C5井的步长为3个月数据资料作为一个样本,选取每个样本的输水量、输水持续天数、上季度该井的潜水埋深平均值作为模型输入量,输出量为相应的C5井的本季度的潜水埋深平均值,建立3-11-1的BP神经网络模型,模拟了C5井潜水埋深.结果表明,网络模拟相对误差小于5%,模型具有较高的精度.通过BP模型模拟潜水埋深,为塔河下游生态恢复和水资源决策提供一定的依据.

Studying on the C5 well of the Yengsu section 350 meter away from river in lower reaches of Tarim River, analysizing on the factors impacting on groundwater level in the lower reaches of Tarim River, simulating change of groundwater level with three BP neural network models. With Matlab 7.0 as the working platform, the data of C5 well for every 3 months is selected as a sample during 2000.7-2008.12, the transportion quantity of each sample, the number of days of every sample transporting water, average depth of groundwater level in the last sample is selected as model input, the output is the groundwater level on average in this quarter, 3-11-1 BP neural network model is established to simulate groundwater level of C5 well. The results show that the network simulation is less than 5% relative error, the model has high accuracy. Using the BP neural network model for simulating the depth of groundwater level, will provide a basis for decision-making for the downstream ecological restoration and water resources of Tarim River.