摘要
建立了常规给水处理工艺处理效率神经网络预测模型和混凝剂投量预测模型。应用结果表明,建立的预测模型具有较高的预测精度:对浊度和UV254处理效率的相关系数(R2)分别为0.86和0.80,对混凝剂投加量的预测精度相关系数为0.72。模型的预测精度可基本满足常规工艺的在线控制和实时调控,使水处理系统在原水水质变化情况下,实现系统优化运行控制。分析模型误差的原因,并对比偏最小二乘回归模型说明神经网络模型的精度,指出该模型在系统优化运行中的可行、及时、准确性。
This paper attempted to accurately and quickly predict the efficiency of the conventional water treatment system and the coagulant dosage with the Artificial Neural Network （ANN） modeling technique. An ANN model developed in the paper was applied to one water plant and the result showed that the ANN model could be used to predict the treatment efficiency of the system with the correlation coefficient （R^2） of 0.86 and 0. 80 for turbidity and ultraviolet absorbance （UV254） respectively, and with the correlation coefficient （R2） for predicting the dosing rate at 0. 72. All the efforts are prepared to realize the online and real time control when water quality was greatly changed. The reason and conditions of the error was analyzed, and as a result, the ANN model could be used to direct the water treatment system feasibly, promply and precisely when compared with the PLS mode.
出处
《西安建筑科技大学学报（自然科学版）》
CSCD
北大核心
2005年第4期488-491,共4页
Journal of Xi'an University of Architecture & Technology(Natural Science Edition)
基金
国家高技术研究发展计划(863)项目(2002AA601140)
北京市水质科学与水环境恢复工程重点实验室开放课题
陕西省自然科学基金项目(2002E214)
关键词
预测模型
神经网络
控制模型
常规工艺
偏最小二乘回归
Forecasting Modeling
Artificial Neural Network
Control Modeling
Conventional Water Treatment System
Partial Least Square Regression
