水文预测的BP网络模型及其拟合度分析

BP MODEL FOR HYDROLOGIC SERIES PREDICTION AND GOODNESS-OF-FIT ANALYSIS

结合实例分析计算 ,论证了 BP网络模型对水文预测的有效性 .为了克服 BP网络收敛速度慢的缺点 ,文中采用了恰当的输入输出值规范化、输出函数值限幅和最佳步长搜索等方法 ,实践表明这些方法在一定条件下可以有效地加快学习误差的收敛 .为了验证模型预测值的精度 ,对预测系列的拟合度作了一定分析 .

The effectiveness of BP model for hydrologic series prediction is verified by computation of practical examples. Appropriate normalization of inputoutput values, limitation of activation function output ranges and optimization of learning rates are employed simultaneously to improve efficiency of BP network. The computing results of examples show effects of these methods on speeding convergence of learning errors. Analysis for goodnessoffit of predicting series is made to test precision of model.