摘要
Abstract
In order to support the pilot construction task of digital twin Taipu gate and to build the"four forecast"closed-loop chain of digital twin water conservancy project,the multi-dimensional dynamic stage discharge prediction model of digital twin Taipu gate based on machine learning has been developed and launched,which has resolved relevant operational challenges in project scheduling and safe operation.The model is based on the optimization and improvement of the original BP neural network prediction model,coupling the water quality and quantity model of Taihu Basin,and reversing the upstream and downstream water level forecast results,so as to predict the discharge capacity of the project in the next 24 hours,and provide reference for decision-making for flood control and water resources dispatching in the basin.By optimizing the weight of the influence factor and adjusting the training sample,the precision control rate of the gate increased by more than 10%,and the precision rate of scheduling execution reached 100%,a breakthrough was made in the"four forecasts"of reverse flow,which greatly guaranteed the safe operation of the project.In 2022,the application of typhoon defense and the work of pressing salt and supply to downstream areas has achieved remarkable results.关键词
数字孪生/机器学习/水位流量预测/防洪/水资源调度/高质量发展Key words
digital twin/machine learning/stage discharge prediction/flood control/water resources dispatching分类
水利科学
