净水技术2025，Vol.44Issue(9)：157-165,9.DOI:10.15890/j.cnki.jsjs.2025.09.020
基于优化XGBoost算法的黄河下游引黄水库富营养化预测
Prediction of Diversion Reservoir Eutrophication in Yellow River Downstream Based on Optimized XGBoost Algorithm
摘要
Abstract
[Objective]In order to study and predict the risk of eutrophication in the water of the Yellow River diversion reservoir in the lower reaches of the Yellow River,and further improve the monitoring,early warning,and emergency response capabilities of water source quality,a model construction research and application of the sparrow search coupled extreme gradient boosting algorithm(SSA-XGBoost)were carried out.[Methods]Taking two typical Yellow River diversion reservoirs in Jinan City as the research objects,using their water quality historical data and meteorological data from 2013 to 2022,targeting their high total nitrogen,high nitrogen phosphorus ratio,high algae and other water pollution characteristics,the XGBoost algorithm with high computational efficiency and excellent predictive performance was adopted,and the sparrow search algorithm was used to optimize their four parameters.The key factors affecting eutrophication,such as physical and chemical properties,nutrients,and solar radiation,including water temperature,pH,dissolved oxygen,permanganate indicator,turbidity,total phosphorus,total nitrogen,ammonia nitrogen,nitrate,nitrogen phosphorus ratio,7 day average sunshine hours,7 day average total solar radiation,and other 12 indices were used as input variables for the model.A prediction and early warning model for water eutrophication suitable for two Yellow River water diversion reservoirs in Jinan City had been established.[Results]The root mean square error(RMSE)of the SSA XGBoost eutrophication prediction model was 4.25 μg/L,the average absolute error(MAE)was 3.19 μg/L,and the goodness of fit(R2)was 0.77.The prediction accuracy was better than that of the BP neural network model and support vector machine model,and the accuracy of chlorophyll a level prediction could reach over 85%.The variable that had the greatest impact on the chlorophyll a prediction result of the two Yellow River diversion reservoirs was pH,followed by nitrate,and then permanganate index.[Conclusion]Overall,the SSA XGBoost eutrophication prediction model has high accuracy,good performance,and strong practicality.The construction and application research of the model will provide reference and basis for algae risk prediction and early warning in two Yellow River water diversion reservoirs.关键词
引黄水库/SSA-XGBoost/富营养化/叶绿素a/预测预警Key words
diversion reservoir of Yellow River/SSA-XGBoost/eutrophication/chlorophyll a/prediction and early warning分类
建筑与水利引用本文复制引用
李祥,孙韶华,刘帅,马中雨,王明泉,宋武昌,陈发明,李桂芳,贾瑞宝..基于优化XGBoost算法的黄河下游引黄水库富营养化预测[J].净水技术,2025,44(9):157-165,9.基金项目
国家重点研发计划(2021YFC3200904) （2021YFC3200904）
山东省重大科技创新工程项目(2020CXGC011406) （2020CXGC011406）
国家水体污染控制与治理科技重大专项(2017ZX07502002) （2017ZX07502002）
