| 摘要: |
| 摘要：为了指导长潭岗水库安全、高效运行，保护下游人民生命财产，本文对长潭岗水库的日平均进库流量进行了建模。本文使用水库上游集雨面积的降水量、蒸发量资料以及水库的日平均进库流量数据建立人工神经网络模型，通过对模型隐含层和输出层的层函数进行不同的组合，得到了4种复杂程度和表达能力逐渐增加的模型。各模型训练结果较好，验证结果均能在0.05的显著性水平上通过检验，其中模型2111（第一隐含层为一元二次线性函数其余层次均为一元一次线性函数）的效果最好，层函数的复杂程度对峰值输出的准确度有一定的正面的贡献。该模型能够满足气象局的专业气象服务需求。 |
| 关键词: 长潭岗水库、进库流量、相关性分析、人工神经网络 |
| DOI： |
| 投稿时间：2022-06-02修订日期：2022-12-13 |
| 基金项目:湖南省气象局短平快课题（湘气函〔2021〕45号）：湘西州暴雨强度与暴雨雨型研究 |
|
| Design of daily average inflow forecast model of Changtangang Reservoir in flood season |
| michuyang,guxue,qingxiangtao,zhengfuwei |
| (Meteorological Bureau of Xiangxi Tujia and Miao Autonomous Prefecture, Hunan Province) |
| Abstract: |
| Abstract: Changtangang Reservoir is a medium-sized water conservancy project with flood control and various comprehensive benefits, accurate inflow flow prediction is of great significance to guide the safe and efficient operation of the reservoir and protect the safety of people"s lives and property downstream. In this paper, the artificial neural network model is established by using the precipitation and evaporation data of the rainwater collection area in the upper reaches of the reservoir and the daily average inflow data of the reservoir. By combining the layer functions of the hidden layer and the output layer of the model, four models with increasing complexity and expression ability are obtained. The training results of each model are good, and the validation results can pass the test at the significance level of 0.05, model 2111 (the first hidden layer is a univariate quadratic linear function, and the other levels are univariate linear functions) has the best effect, the complexity of layer function has a certain positive contribution to the accuracy of peak output. It can be better used in the business services of the Meteorological Bureau. |
| Key words: precipitation, evaporation, inflow, correlation analysis, artificial neural network |
