水厂智能加矾控制技术的苏州实践  

Suzhou Practice of Intelligent Coagulation Control Technology in Water Treatment Plants

作　　者：刘阳 蒋福春 张雪 王冬生 LIU Yang;JIANG Fuchun;ZHANG Xue;WANG Dongsheng(School of Automation,Nanjing University of Posts and Telecommunications,Nanjing 210023,China;Suzhou City Construction Investment and Development Co.,Ltd.,Suzhou 215002,China)

机构地区：[1]南京邮电大学自动化学院,南京210023 [2]苏州城市建设投资发展(集团)有限公司,苏州215002

出　　处：《给水排水》2025年第3期124-129,共6页Water & Wastewater Engineering

基　　金：苏州市水利水务科技项目(2023009)。

摘　　要：混凝沉淀工艺是自来水厂净水处理过程的重要环节,加矾量的控制效果直接影响沉淀池出水水质的达标与达优。提出了一种基于深度学习的前馈-反馈复合控制方法及系统,该智能加矾系统已经成功地运用到苏州市3座自来水厂。与先前的人工投加控制相比,该系统使沉淀池的单月平均出水浊度降低了15.2%,矾耗下降了约20.21%,每座自来水厂的班组人员由原先的一班4人减少至一班3人,3座自来水厂通过降低矾耗和人力成本,每年共计节约成本约104万元。The coagulation and sedimentation process is a critical stage in the water purification process at water treatment plants.The control of coagulant dosage directly impacts the quality of the treated water in the sedimentation tank.This paper proposes a deep learning-based feedfor-ward-feedback composite control method and system.This intelligent coagulant dosing system has been successfully applied to the three water treatment plants of Suzhou.Compared with the previ-ous manual dosing control,the system has reduced the average monthly effluent turbidity of the sedimentation tank by 15.2%,decreased coagulant consumption by approximately 20.21%,and reduced the number of shift personnel at each plant from four to three per shift.The three water treatment plants collectively save approximately 1.04 million yuan annually by reducing coagulant consumption and labor costs.

关 键 词：自来水厂 混凝 深度学习 前馈-反馈复合控制 节能降耗 

分 类 号：TU991[建筑科学—市政工程]