排灌机械工程学报2025，Vol.43Issue(10)：1040-1048,9.DOI:10.3969/j.issn.1674-8530.24.0104
基于Stacking集成学习的水电机组负荷分配
Load distribution for hydroelectric units based on Stacking ensemble learning
摘要
Abstract
In the context of large-scale hydroelectric units and complex operating conditions,the tradi-tional load distribution of hydroelectric units faces problems such as long optimization time,easy falling into local optimum,and unstable results.A Stacking ensemble learning model and constraint correction-based load distribution method for hydroelectric units were proposed in this study.Firstly,historical data were input into the Stacking ensemble learning model for training,and the K-fold cross-validation method was used to alleviate the overfitting caused by repeated learning,resulting in an ini-tial plan for unit load distribution.Secondly,the initial plan was modified with constraints such as load balancing,output limitations,and unit combinations to continuously approximate the historical decision and form the final decision plan.Taking a power station as an example,indicators such as water con-sumption and output fluctuation rate were used to evaluate the distribution results,and then compared with traditional dynamic programming methods.Through the integration,the time required for online prediction of load allocation after the model training is only 2.04 s.The decision time is greatly shor-tened,and the prediction accuracy and robustness are significantly improved.It can provide certain references for unit load distribution.关键词
水电机组负荷分配/厂内经济运行/Stacking集成学习/约束修正/机器学习Key words
load distribution of hydroelectric units/economical operation within the factory/Stacking ensemble learning/constraint correction/machine learning分类
水利科学引用本文复制引用
郑晓楠,于洋,潘虹,郑源,杭晨阳,杨杰,马晓瑶,陈致远..基于Stacking集成学习的水电机组负荷分配[J].排灌机械工程学报,2025,43(10):1040-1048,9.基金项目
国家自然科学基金重点项目(52339006) （52339006）
江苏省创新支撑计划国际科技合作项目(BZ2023047) （BZ2023047）
