摘要
为了提高径流预测的精度,提出了小波分析与遗传算法优化BP神经网络相结合的预测模型。采用呼兰河流域兰西水文站1956-2019的实测年径流序列进行预测和测试,选取均方根误差(RMSE)、拟合系数(R^(2))和平均相对误差(MAPE)对预测结果进行对比评价,与其GA-BP模型和BP模型进行对比,有着更高的精度和更低的误差。为呼兰河流域径流的预测提供了一条新的方法。
In order to improve the accuracy of runoff prediction,a prediction model combining wavelet analysis and genetic algorithm optimized BP neural network was proposed.The measured annual runoff series from Lanxi Hydrological Station in the Hulan River Basin from 1956 to 2019 were used for prediction and testing.Root mean square error(RMSE),fitting coefficient(R^(2)),and mean relative error(MAPE)were selected to compare and evaluate the prediction results.Compared with its GA-BP model and BP model,it has higher accuracy and lower error.This provides a new method for predicting runoff in the Hulan River Basin.
作者
李杰
孙颖娜
曹越
LI Jie;SUN Yingna;CAO Yue(School of Hydraulic and Electric-power,Heilongjiang University,Harbin 150080,China;Heilongjiang Water Resources and Hydropower Survey and Design Institute,Harbin 150080,China)
出处
《云南水力发电》
2025年第1期9-13,共5页
Yunnan Water Power
基金
国家社会科学基金项目(20BS042)。
