摘要
【目的】二次供水风险点多、设施分布广,是供水环节风险管控的重点难点。余氯是二次供水水质重要参数,通过预测二次供水出水余氯变化,为二次供水环节水质风险管控提供决策依据。【方法】文章建立了具备实时、高频、连续的自动检测、分析、预警功能的水质在线监测系统,以深圳市已完成二次供水设施提标改造的某小区为例,对生活二次供水水池出水余氯实时监测,分析余氯与时间关系,结合人工神经网络建模训练预测余氯值。【结果】余氯随时间呈明显周期变化,变化拐点滞后于用水高峰期2 h左右,模型输出值与真实监测值整体变化趋势保持一致,结果相近,大部分点相对误差均能控制在5%以内,输出值与真实监测值整体均方根误差为0.0238。【结论】影响水池出水余氯的主要因素为市政进水量及水力停留时间,基于水质在线监测数据考虑时间因素对余氯参数的影响,结合人工神经网络工具实现对余氯数据的高精度预测,研究成果为更大范围的余氯预测研究及后续修正管网余氯模型、优化水厂药剂投放、稳定用水终端水质参数研究提供必要依据。
[Objective]Risk points of secondary water supply are widely distributed by multiple facilities,which is the key and difficult point of risk control of water supply link.Residual chlorine is an important parameter of water quality of secondary water supply,and the change of residual chlorine in effluent of secondary water supply can be predicted to provide decision-making basis for water quality risk control of secondary water supply link.[Methods]An online water quality monitoring system with real-time,highfrequency and continuous automatic detection,analysis and early warning functions was established.Taking a community in Shenzhen City that had completed the upgrading of secondary water supply facilities as an example,the residual chlorine of domestic secondary water supply tank was monitored in real time,the relationship between residual chlorine and time was analyzed,and the residual chlorine value was predicted by artificial neural network modeling training.[Results]Residual chlorine showed an obvious periodic change with time,and the change turning point lagged behind the peak water consumption by about 2 hours.The overall change trend of the model output value was consistent with the real monitoring value,and the result were similar.The relative error of most points could be controlled within 5%,and the overall root-mean-square error of the output value and the real monitoring value was 0.0238.[Conclusion]The main factors affecting residual chlorine in water discharge are municipal water intake and hydraulic residence time.Based on online water quality monitoring data,the influence of time factors on residual chlorine parameters is considered,and the high-precision prediction of residual chlorine data is achieved by combining artificial neural network tools.The research result provide a necessary basis for a larger range of residual chlorine prediction research and subsequent modification of the residual chlorine model of the pipe network,optimization of pharmaceutical delivery in the water treatment plant,and water quality parameters at the stable water terminal.
作者
周云鹏
ZHOU Yunpeng(Shenzhen Water<Group>Co.,Ltd.,Shenzhen 518000,China)
出处
《净水技术》
2025年第8期188-192,共5页
Water Purification Technology
关键词
在线监测
余氯
人工神经网络
模型结构
预测
on-line monitoring
residual chlorine
artificial neural network(ANN)
model structure
forecast
