自来水厂混凝剂智能投加系统的构建与应用  

Construction and Application of Intelligent Coagulant Dosing System for Waterworks

作　　者：吴志金 赖源平 许振楚 贺涛 李伟廉 魏日强 王益发 WU Zhi‑jin;LAI Yuan‑ping;XU Zhen‑chu;HE Tao;LI Wei‑lian;WEI Ri‑qiang;WANG Yi‑fa(Guangzhou Water Supply Co.Ltd.,Guangzhou 510000,China;Guangdong Longquan Technology Co.Ltd.,Guangzhou 510000,China)

机构地区：[1]广州市自来水有限公司,广东广州510000 [2]广东龙泉科技有限公司,广东广州510000

出　　处：《中国给水排水》2024年第18期90-95,共6页China Water & Wastewater

摘　　要：在常规水处理工艺中,混凝沉淀是一个多参数、非线性且大时滞的过程。某大型水厂在混凝阶段采用机器视觉技术和相关AI算法,构建混凝剂智能投加系统。该系统采用微距成像的方式,利用水下摄像头捕捉混凝阶段所形成“矾花”的微距特征,并应用图像识别技术实现对矾花特征的量化表达,同时引入与混凝沉淀相关的多维生产参数,采用分段集成学习的方式,建立智能投加算法模型,并开发多参数混凝剂投加的自动分析和评估决策系统。该水厂的实际应用表明,实现混凝剂智能投加的精准控制,可起到降低混凝剂耗量、节省生产成本的作用。In conventional water treatment processes,coagulation and sedimentation is multi‑parameter,nonlinear and large time‑delay process.An intelligent coagulant dosing system is constructed by using machine vision technology and related AI algorithm during the coagulation stage in a large waterworks.The system adopts macro imaging and utilizes underwater cameras to capture the macro features of alum floc formed during the coagulation stage,and applies image recognition technology to quantify the alum floc features.At the same time,combined with segmented ensemble learning,multi‑dimensional production parameters related to coagulation and sedimentation are introduced to establish an intelligent dosing algorithm model.And then a multi‑parameter coagulant dosing automatic analysis and evaluation decision‑making system is developed.The application of the waterworks shows that precise control of intelligent coagulants dosing can reduce coagulant consumption and save production cost.

关 键 词：混凝沉淀 矾花 图像识别 智能投加 精准控制 

分 类 号：TU991[建筑科学—市政工程]