WANGChenbo,WANGYue,YUHuarong,et al.Machine Learning-based Optimization of Electrocoagulation for Fluoride Removal from Groundwater[J].China Water & Wastewater,2026,42(11):16-25.
基于机器学习模型优化电凝聚法去除地下水中氟化物
- Title:
- Machine Learning-based Optimization of Electrocoagulation for Fluoride Removal from Groundwater
- Keywords:
- groundwater; electrocoagulation; fluoride removal; machine learning; parameter optimization
- 摘要:
- 本研究提出一种基于机器学习的电凝聚工艺参数优化框架，用于高效去除地下水中的氟化物。通过整合文献中收集的204个试验数据点，构建了涵盖反应条件与处理结果的多维数据集，并系统比较了6种典型机器学习算法的预测性能。结果表明，XGBoost模型在剩余氟浓度和能耗的双重预测任务中表现最佳（R2test=0.76~0.89，MSEtest≤0.004 7），优于KNN、SVR、RF、LightGBM 和 MLP。基于SHAP的特征重要性分析揭示，初始氟浓度、电解时间和电流密度是决定除氟效果的关键因子，而电解时间、电流密度与阳极面积则主导能耗水平。进一步将训练良好的XGBoost模型嵌入快速精英多目标遗传算法（NSGA-Ⅱ），实现了“去除率最大化-能耗最小化”的多目标优化，获得了多组帕累托前沿解。验证结果显示，模型预测值与实测值的相对误差在10%~25%之间。该框架有效展示了数据驱动逆向设计的潜力，为电化学水处理工艺的智能化设计提供了新思路，提升了除氟工艺的效率。
- Abstract:
- This study presented a machine learning?based framework for optimizing the electrocoagulation (EC) process parameters to efficiently remove fluoride in groundwater. By integrating 204 experimental data points collected from the literature, a multidimensional dataset encompassing reaction conditions and treatment results was constructed. The performance of six typical machine learning algorithms was systematically compared. The results showed that the XGBoost model performed the best in the dual prediction task of residual fluoride concentration and energy consumption (R2test=0.76-0.89, MSEtest≤0.004 7), outperforming KNN, SVR, RF, LightGBM, and MLP. Feature importance analysis based on SHAP revealed that initial fluoride concentration, electrolysis time, and current density were the key factors determining fluoride removal efficiency, while electrolysis time,current density, and anode area dominated energy consumption levels. Furthermore, the well-trained XGBoost model was integrated into a non-dominated sorting genetic algorithm (NSGA-Ⅱ) for multi-objective optimization of removal rate maximization-energy consumption minimization,yielding multiple Pareto front solutions. Experimental validation showed that the relative error between model predictions and actual measurements ranged from 10% to 25%. This framework effectively demonstrates the potential of data-driven reverse design, providing new insights for the intelligent design of electrochemical water treatment processes, improving fluoride removal efficiency.
相似文献/References:
[1]张安嫒.滤料粒径对地下水中铁、锰、氨氮、浊度去除效果的影响[J].中国给水排水,2020,36(7):47.
[2]袁雅姝,陈正洋,傅金祥,等.地下水源除铁锰滤池调试运行实践及思考[J].中国给水排水,2020,36(15):35.
YUAN Ya-shu,CHEN Zheng-yang,FU Jin-xiang,et al.Practice and Consideration of Commissioning and Operation of Iron and Manganese Removal Filter Treating Groundwater in Liaozhong District of Shenyang City[J].China Water & Wastewater,2020,36(11):35.
[3]李圭白,杨海洋,仲琳,等.锰质滤膜活性对接触氧化除锰及除氨氮效能的影响[J].中国给水排水,2020,36(21):1.
LI Gui-bai,YANG Hai-yang,ZHONG Lin,et al.Effect of Manganese Oxides Film Activity on Removal Efficiencies of Manganese and Ammonia Nitrogen by Contact Oxidation[J].China Water & Wastewater,2020,36(11):1.
[4]由昆,范维利,傅金祥,等.载锰活性氧化铝除氟降铁效能研究[J].中国给水排水,2020,36(23):42.
YOU Kun,FAN Wei-li,FU Jin-xiang,et al.Fluorine and Iron Removal Efficiency of Manganese-loaded Activated Alumina[J].China Water & Wastewater,2020,36(11):42.
[5]赵蓓,李林彬,赵云鹏,等.水源水质对地下水管网铁释放的影响和控制技术研究[J].中国给水排水,2020,36(23):52.
ZHAO Bei,LI Lin-bin,ZHAO Yun-peng,et al.Influence of Source Water Quality on Iron Release in Groundwater Distribution Networks and Its Control Technology[J].China Water & Wastewater,2020,36(11):52.
[6]赵蓓,张海祥,李礼,等.自备井置换区域管网水质风险评估[J].中国给水排水,2021,37(13):32.
ZHAO Bei,ZHANG Hai-xiang,LI Li,et al.Risk Assessment for Water Quality of Distribution Network in Areas Using Self-supply Wells after Water Source Replacement[J].China Water & Wastewater,2021,37(11):32.
[7]王旭峰,杨涛,徐秀丽,等.三维电极生物膜与硫自养耦合工艺脱氮特性[J].中国给水排水,2024,40(17):31.
WANGXu-feng,YANGTao,XUXiu-li,et al.Nitrogen Removal in Three?dimensional Electrode Biofilm Coupled with Sulfur Autotrophic Process：Influence of HRT and Sodium Bicarbonate[J].China Water & Wastewater,2024,40(11):31.
[8]李普壮,黄廷林,程亚.水循环对复合锰氧化物催化氧化氨氮和锰的影响[J].中国给水排水,2025,41(1):52.
LIPu-zhuang,HUANGTing-lin,CHENGYa.Impact of Water Circulation on Catalytic Oxidation of Ammonia Nitrogen and Manganese by Composite Manganese Oxide[J].China Water & Wastewater,2025,41(11):52.
[9]王松波,唐欢,何香雨,等.含氧阴离子对MnO2催化氧化氨氮活化效果的影响[J].中国给水排水,2026,42(11):1.
WANGSongbo,TANGHuan,HEXiangyu,et al.Effect of Oxyanions on Activation of MnO2 for Catalytic Oxidation of Ammonia Nitrogen[J].China Water & Wastewater,2026,42(11):1.