摘要
【目的】掌握沉淀池底泥的时空分布规律,对优化排泥策略、提升水厂运行效率与科学管理水平具有重要意义。【方法】文章以舟山市某水厂的平流沉淀池为研究对象,创新地提出了预测沉淀池泥位时空分布的混合模型。该模型是将水动力模型和数据驱动模型融合:首先,建立沉淀池的水动力模型模拟分析不同工况下沉淀池底泥的时空分布特性;其次,在水动力模型基础上引入数据驱动模型,最后,构建混合模型预测沉淀池泥位。【结果】混合模型具有较好的准确性和可靠性:水动力模型沉淀池底泥厚度的模拟值和实测值平均误差为2.38%,与沉淀池实际底泥分布趋势一致,大部分污泥在进水前段发生沉降,在沉淀池的末端出水处污泥沉降比较小;XGBoost在多种数据驱动模型中预测效果最好,R^(2)为0.995,平均绝对误差(MAE)为0.0097,均方根方误差(RMSE)为0.0274,沉淀池底泥厚度的预测值与模拟值之间的平均误差为1.83%。【结论】文章通过案例分析最终总结出,采用水动力模型耦合XGBoost数据模型的混合模型,进行沉淀池底泥预测分析,具有显著优势。该混合模型不仅能够准确反映沉淀池底泥的时空分布特性,还有助于优化沉淀池的排泥方式,进而提升水厂的管理和运行效能,具有重要的应用价值。
[Objective]Mastering the temporal and spatial distribution laws of sediment in sedimentation tanks is of great significance for optimizing sludge discharge strategies and improving the operating efficiency and scientific management level of water treatment plants.[Methods]This paper took the horizontal sedimentation tank of a water treatment plant in Zhoushan City as the research object,and innovatively proposed a hybrid model to predict the temporal and spatial distribution of the mud level in the sedimentation tank,this model combined a hydrodynamic model and a data-driven model:Firstly,a hydrodynamic model of the sedimentation tank was constructed to simulate and analyze the temporal and spatial distribution characteristics of the sedimentation tank sediment under different working conditions;Then,a data-driven model was introduced based on the hydrodynamic model to build a hybrid model to predict the mud level in the sedimentation tank.[Results]The hybrid model had good accuracy and reliability:the average error between the simulated value and the measured value of sediment thickness in the sedimentation tank in the hydrodynamic model was 2.38%,which was consistent with the actual sediment distribution trend in the sedimentation tank.Most of the sludge settles in the front section of the inflow,and was relatively small at the outlet of the sedimentation tank;XGBoost had the best prediction effect among various data-driven models,with R^(2) of 0.995,mean absolute error(MAE)of 0.0097,and root mean absolute error(RMSE)of 0.0274,The average error between the predicted value and the simulated value of sediment thickness in the sedimentation tank was 1.83%.[Conclusion]Through case analysis,this paper concludes that using a hybrid model of hydrodynamic model coupled with XGBoost data model for sediment prediction and analysis in sedimentation tanks has significant advantages.The hybrid model not only accurately reflects the temporal and spatial distribution characteristics of the sediment in the sedimentation tank,but also helps optimize the sludge discharge method of the sedimentation tank,thereby improving the management and operation efficiency of the water treatment plants,and has important application value.
作者
陈汪洋
张孝洪
柳景青
傅舟跃
黄盼盼
张卫平
CHEN Wangyang;ZHANG Xiaohong;LIU Jingqing;FU Zhouyue;HUANG Panpan;ZHANG Weiping(China Water Investment Group Co.,Ltd.,Beijing 100053,China;Zhoushan Water Co.,Ltd.,Zhoushan 316000,China;College of Civil Engineering and Architecture,Zhejiang University,Hangzhou 310000,China;Binhai Industrial Technology Research Institute,Zhejiang University,Tianjin 300458,China;Tianjin Zhiyun Water Technology Co.,Ltd.,Tianjin 300301,China;College of Software,Northwestern Polytechnical University,Xi'an 710068,China)
出处
《净水技术》
2025年第8期69-77,共9页
Water Purification Technology
关键词
沉淀池
水动力模型
泥位预测
混合模型
时空分布
sedimentation tank
hydrodynamic model
sludge level prediction
hybrid model
spatial and temporal distribution
