Kriging水动力学代理模型在水库群优化调度中的应用  

Application of Kriging-based Hydrodynamic Surrogate Modelfor Optimal Scheduling of Cascade Reservoirs

作　　者：徐杨 吕昊 刘帅[3,4] 方威 覃晖[3,4] XU Yang;L Hao;LIU Shuai;FANG Wei;QIN Hui(Three Gorges Cascade Dispatch and Communication Center,Yichang 443000,China;Power China Huadong Engineering Corporation Limited,Hangzhou 311122,China;School of Civil and Hydraulic Engineering,Huazhong University of Science and Technology,Wuhan 430074,China;Hubei Key Laboratory of Digital Valley Science and Technology,Huazhong University of Science and Technology,Wuhan 430074,China)

机构地区：[1]三峡水利枢纽梯级调度通信中心,湖北宜昌443000 [2]中国电建集团华东勘测设计研究院有限公司,杭州311122 [3]华中科技大学土木与水利工程学院,武汉430074 [4]华中科技大学数字流域科学与技术湖北省重点实验室,武汉430074

出　　处：《长江科学院院报》2024年第2期7-13,20,共8页Journal of Changjiang River Scientific Research Institute

基　　金：国家重点研发计划项目(2021YFC3200303)。

摘　　要：为解决传统水动力学模型结合调度模型求解耗时长的难题,使用代理模型模拟水动力学模型以提升求解效率。从金沙江下游—三峡梯级水库调度实际出发,结合水动力学和代理模型相关理论,建立了基于Kriging代理模型的水库群多目标调度模型并采用多目标进化算法对模型进行求解,最后采用投影追踪法对得到的多目标帕累托前沿进行决策。研究结果表明,代理模型的平均模拟误差≤1.5%,能够较为精确地替代原有长耗时水动力学模型并结合调度模型给出具有竞争关系的帕累托前沿及合理的折衷方案,对促进梯级水库综合效益发挥提供理论依据。To address the time-consuming nature of traditional hydrodynamic model combined with scheduling model,we propose a surrogate model as an approximation to enhance efficiency.The research focuses on the practical scheduling of cascade reservoirs from lower Jinsha River to the Three Gorges.By combining hydrodynamics theory with the Kriging surrogate model,we established a multi-objective dispatching model for a multi-reservoir system,and subsequently solved the model using a multi-objective evolutionary algorithm.Ultimately,we identified the multi-objective Pareto front of the model by adopting the projection pursuit method in decision-making.Research findings demonstrate that the surrogate model exhibits an average simulation error of less than 1.5%,effectively replacing the time-consuming hydrodynamic model while generating a competitive Pareto front and providing reasonable compromise solutions.These outcomes lay a theoretical foundation for advancing the comprehensive benefits of cascade reservoirs.

关 键 词：多目标优化调度 KRIGING模型 水动力学模型 投影追踪法 梯级水库 

分 类 号：TV697.11[水利工程—水利水电工程]