摘要
利用造纸废水处理监控系统取得表征废水水质的各项指标,在此基础上研究了基于BP网络和RBF网络的造纸废水处理建模。仿真结果表明,BP网络较RBF网络对样本数据的仿真误差较小,泛化能力更好;输入量考虑历史出水COD变化趋势的网络,其仿真效果要优于不考虑变化趋势的网络;运用基于BP网络和RBF网络的造纸废水处理模型能够准确地预测出水COD,为实现废水处理的自动控制提供可行途径。
Based on the gaining the water quality index through monitor system for wastewater treatment in papermaking, wastewater treatment models by BP and RBF neural networks are studied. The result of simulation shows that the model established by BP network has a smaller simulated error and a better general ability according to the model established by RBF network, and the network considering the changing trend of historical COD value of effluent in the input layer has a better simulation effect than the one without the changing trend. The wastewater treatment models by BP and RBF neural networks can accurately predict the COD value of effluent, providing a mean to realize automatic control in wastewater treatment.
出处
《造纸科学与技术》
2006年第6期132-136,共5页
Paper Science & Technology
基金
广东省科技厅重大专项基金(项目号2003A3040406)
广州市科技计划项目(项目号2004Z3-D0271)资助
项目名称"二次纤维造纸废水处理智能控制系统"。
关键词
造纸
废水处理
BP神经网络
RBF神经网络
仿真研究
papermaking
wastewater treatment
BP neural network
RBF neural network
simulation study
作者简介
黄明护，男，硕士研究生，主要从事制浆新技术与污染控制的研究。
