水厂精准混凝投药实时控制系统的构建  

Real-time control system development of precise coagulation dosing in water treatment plant

作　　者：王艺颖 WANG Yiying(Shenzhen Koron Soft Co.,Ltd.,Shenzhen 518020,China)

机构地区：[1]深圳市科荣软件股份有限公司,深圳518020

出　　处：《给水排水》2023年第5期156-164,共9页Water & Wastewater Engineering

摘　　要：基于水厂运行大数据的采集,分别运用随机森林、XGBoost、BP神经网络3种人工智能算法,对混凝投药过程进行非线性建模和非线性预测,构建出水浊度预测模型并进行评估和对比,再以沉淀池出水浊度为控制目标进行最佳投药量求解。同时,模型通过统计短时间隔流速计算混凝至出水之间的滞后时间,提高系统时滞计算的准确性。系统通过算法调用实现实时投药量计算,并与SCADA系统实现数据同步,最终通过SCADA下发指令实现对投药泵的控制。Based on the acquisition of big-data in water treatment plant(WTP)operation,three artificial intelligence algorithms of random forest,XGBoost and BP neural network were used to carry out nonlinear modeling and nonlinear prediction of the coagulation and dosing process,and the effluent turbidity prediction model was constructed,evaluated and compared,and then the optimal dosage was solved with the effluent turbidity of the sedimentation tank as the control target.At the same time,the model calculates the lag time between coagulation and effluent by counting the short-time interval flow rate,which improves the accuracy of the system time delay calculation.The system realizes real-time dosing amount calculation through algorithm invocation,realizes data synchronization with SCADA system,and finally realizes the control of dosing pumps through SCADA command delivery.

关 键 词：混凝投药 大数据 人工智能 水处理优化 

分 类 号：TU991.22[建筑科学—市政工程]