基于ANN的废纸造纸废水处理过程的动态建模  

ANN-Based Dynamic Modeling of Wastewater Treatment Process in Paper Making by Waste Paper

作　　者：李迪[1] 唐辉[1] 万金泉[2] 马邕文[2] 

机构地区：[1]华南理工大学机械工程学院,广东广州510640 [2]华南理工大学资源科学与造纸工程学院,广东广州510640

出　　处：《华南理工大学学报（自然科学版）》2005年第12期42-45,共4页Journal of South China University of Technology(Natural Science Edition)

基　　金：广东省科技厅重大专项(2003A3040406);广州市科技计划项目(2004Z3-D0271)

摘　　要：研究了人工神经网络技术在废纸造纸废水处理过程动态建模中应用的可行性,采用误差反向传播网络(BP网)建立了表征原水COD、加药量、进水流量、历史出水COD与预计出水COD之间复杂关系的动态模型,并对不同训练方法进行了比较,发现带有动态调整的方法具有较好的效果,其模型的计算输出值与过程的实际输出值具有较好的一致性.利用所建立模型对造纸厂现场排放废水进行实验,结果表明该模型可用于废纸造纸废水处理的动态描述,并为进一步的废水处理系统在线智能控制奠定了基础.This paper aims at exploring the possibility of building a dynamic model by artificial neural network for the wastewater treatment process in the paper making by waste paper. During the investigation, the error BP （ Back Proragation） neural network is used to establish a dynamic model and describe the complex relationship among the COD of the influent wastewater, the dosage of the added medicament, the flowrate of the influent, the historical COD of the effluent and the predictive COD of the effluent. Different training methods for the network are then compared and discussed, demonstrating that the training method with dynamic adjustment results in good consistency between the calculated ouputs and the practical ouputs. The proposed model is finally applied to the wastewater treatment of a paper mill, with the results verifying the effectiveness of the BP model for the dynamic description of wastewater treatment. This model lays a foundation for the further investigation into the on-line intelligent control of the wastewater treatment system.

关 键 词：造纸 废水处理 人工神经网络 动态建模 

分 类 号：X703.1[环境科学与工程—环境工程]