无站条件下BP神经网络年径流量预测模型研究

针对无监测站的河流断面的径流量预测难的问题,选取了有针对性的影响因子,采用人工神经网络的BP改进算法—自适应调节学习速率算法,以Matlab神经网络工具箱为工具,以辽河干流铁岭站为假想对象,建立了无站条件下年径流量预测模型,模型训练48次误差达到要求,以2004～2006年3年实测资料作为检验样本进行仿真,验证模型的精度。仿真结果表明:3年预测结果全部满足要求,说明该模型可用于河流任一断面年径流量的预测。

The BP Network Prediction Model for Annual Runoff for Lacking of Hydrological Station

In connection with the problem of runoff prediction of river section for lacking of station,with the targeted impact factor the annual runoff prediction model was established taking Tieling station of Liao River mainstream as the imaginary example which used improved BP algorithm of Artificial Neural Network-adaptive learning rate algorithm and Matlab Toolbox as the tool.The error met the requirement after forty-eight times training.Take the measured data from 2004 to 2006 as test sample to simulate and verify the model accuracy.The simulation result showed that the prediction of the three years met the requirement and the model could be used in annual runoff prediction on any section of river.