摘要
以江西省某中型水库为研究对象,探索基于深度学习的水利数字孪生系统在动态数据融合与预测中的应用效果。针对水库流域内数据异构性和调度滞后性问题,设计了一套集动态感知、数据融合、预测和反馈优化的完整技术框架。研究结果表明,系统在短期预测中将水位误差控制在±0.1 m内,在长期预测中将水位误差控制在±0.3m内,同时将数据传输延迟从5s缩短至3s,整体响应时间缩短至10s以内。研究表明,基于深度学习的水利数字孪生技术能够显著提升水库防洪调度的精准性和实时性,对于中小型水库智慧化管理改革具有一定的推广价值。
This paper takes a medium-sized reservoir in Jiangxi Province as the research object,exploring the application effect of deep learning based water conservancy digital twin system in dynamic data fusion and prediction.A complete technical framework integrating dynamic perception,data fusion,prediction,and feedback optimization has been designed to address the issues of data heterogeneity and scheduling lag within the reservoir basin.The research results indicate that the system controls the water level error within±0.1 m in short-term forecasting and±0.3 m in long-term forecasting,while reducing the data transmission delay from 5 s to 3 s and overall response time to within 10 s.Research has shown that deep learning based digital twin technology for water conservancy can significantly improve the accuracy and real-time performance of reservoir flood control scheduling,and has certain promotional value for the intelligent management reform of small and medium-sized reservoirs.
作者
黎春根
程铁洪
王皓
LI Chungen;CHENG Tiehong;WANG Hao(PowerChina Jiangxi Electric Power Engineering Co.,Ltd.,Nanchang,Jiangxi 330096,China)
出处
《自动化应用》
2025年第12期81-84,共4页
Automation Application
关键词
深度学习
水利数字孪生
动态数据
数据融合
预测模型
deep learning
water conservancy digital twin
dynamic data
data fusion
prediction model
