摘要
针对猫街水文站采用物理模型进行洪水预报时过程复杂、适用性和快捷性较差等问题,笔者选取BP神经网络模型作为预报模型进行洪水预报研究,选取2013—2018年猫街水文站共计6年主汛期逐日水文观测资料作为训练样本,2019—2021年共3年主汛期资料作为测试样本。研究结果表明,在现有数据条件下,除部分特殊年份外,采用BP神经网络模型进行洪水预报的精度较高,整体预报精度较好,对实际预报作业有一定的指导意义。同时,BP神经网络预报模型具有误差修正功能,随着模型学习训练期的延长、预报次数增加,预报精度还会相应提高,未来可运用到实际预报作业当中。
In view of the complex process, poor applicability and rapidity of the physical model used by the Maojie Hydrologic Station for flood forecasting, the BP neural network model is selected as the forecasting model for flood forecasting research. The daily hydrological observation data of the Maojie Hydrological Station in the main flood season of six years from 2013 to 2018 is selected as the training sample, and the data of the main flood season of three years from 2019 to 2021 is selected as the test sample. The research results show that under the existing data conditions, except for some special years, the BP neural network model has higher accuracy for flood forecasting, and the overall forecasting accuracy is better, which has certain guiding significance for the actual forecasting operation. At the same time, the BP neural network prediction model has the function of error correction. With the extension of the model learning and training period and the increase of the prediction times,the prediction accuracy will also be improved accordingly, which can be applied to the actual prediction operation in the future.
作者
柳志强
LIU Zhiqiang(Hydropower Plant of Tianshengqiao-I Hydropower Development Co,Ltd,Xingyi 562400,China)
出处
《红水河》
2023年第1期32-36,共5页
Hongshui River
关键词
洪水预报
BP神经网络模型
猫街水文站
flood forecasting
BP neural network model
Maojie Hydrologic Station
作者简介
柳志强(1993),男,黑龙江伊春人,助理工程师,学士,主要从事水库调度管理工作。E-mail:phoenixhhu@qq.com。
