全部 标题 作者
关键词 摘要

OALib Journal期刊
ISSN: 2333-9721
费用:99美元

查看量下载量

相关文章

更多...

死端微滤牛血清白蛋白溶液膜通量的预测

Keywords: 死端微滤,通量,BP神经网络,RBF神经网络

Full-Text   Cite this paper   Add to My Lib

Abstract:

为实现对不同操作条件(操作压力、料液质量浓度和温度)下的牛血清白蛋白溶液死端微滤膜通量的预测,以训练步数、绝对相对误差和相关系数作为预测的衡量指标,并对所建立的3层BP神经网络和RBF神经网络基本模型的内部参数进行了优化.优化的BP神经网络模型的拓朴结构为3-9-1,学习率为0.05,学习/训练函数为traingdx,隐层到输出层的传递函数为logsig,该网络对牛血清白蛋白(BSA)溶液膜通量预测的平均绝对相对误差为2.37%,相关系数为0.9960;优化的RBF神经网络的网络设计函数为newrbe,散布常数为400,该网络对BSA溶液膜通量预测的平均绝对相对误差为4.83%,相关系数为0.9870.结果表明,BP神经网络优于RBF神经网络.

References

[1]  MOHAMMADI T,MOGHADAMK M,MADAENI S S.Hydrodynamic factors affecting flux and fouling during reverse osmosisof seawater[J].Desalination,2002(151):239-245.
[2]  DORNIER M,DECLOUX M,TRYSTRAMG,et al.Dynamic modelingof crossflowmicrofiltration using neural networks[J].J Membr Sci,1995,98(1/2):263-273.
[3]  PIRON E,LATRILLE E,RENE F.Application of artificial neural networks for crossflow microfiltration modeling:“black-box”and semi-physical approaches[J].Comput Chem Eng,1997,21(2):1021-1030.
[4]  HAMACHI M,CABASSUD M,DAVIN A.Dynamic modelling of crossflow microfiltration of bentonite suspension usingrecurrent neural networks[J].Chem Eng Process,1999,38(3):203-210.
[5]  GOLOKA B S,CHITTARANJAN R.Predicting flux decline in crossflow membranes using artificial neural networks andgenetic algorithms[J].J Membr Sci,2006,283(1/2):147-157.
[6]  MHURCHU′J N,FOLEY G.Dead-end filtration of yeast suspensions:correlationg specific resistance and flux data usingartificial neural networks[J].J Membr Sci,2006,281(1/2):325-333.
[7]  周开利,康耀红.神经网络模型及其MATLAB仿真程序设计[M].北京:清华大学出版社,2004:89-90.
[8]  董长虹.Matlab神经网络与应用[M].北京:国防工业出版社,2005:71-90.
[9]  WOODS D,PIERCE C,ZEIHER E H.Biofouling of reverse osmosis systems:three case studies[J].Ultrapure Water,1991,8(7):50-64.
[10]  MATTHEUS F A G,SHYAM S S,SALHA S A.Effect of feed temperature on permeate flux and mass transfer coefficient inspiral-wound reverse osmosis systems[J].Desalination,2002(144):367-372.
[11]  陈明杰,倪晋仁,薛安.典型前馈神经网络在潮流模拟中的应用与比较[J].泥沙研究,2003(5):41-48.CHEN Ming-jie,NI Jin-ren,XUE An.Comparative study on typical feed-forward ANNs for tidal simulation[J].Journal ofSediment Research,2003(5):41-48.(in Chinese)
[12]  飞思科技产品研发中心.MATLAB6.5辅助神经网络分析与设计[M].北京:电子工业出版社,2003:10-12,64-69.
[13]  FU R Q,XU T W,PAN Z X.Modelling of the adsorption of bovine serum albumin on porous polyethylene membrane byback-propagation artificial neural network[J].J Membr Sci,2005,251(1/2):137-144.
[14]  CHELLAM S.Artificial neural network model for transient crossflow microfiltration of polydispersed suspensions[J].JMembr Sci,2005,258(1/2):35-42.
[15]  RAMASWAMY S,GREENBERG A R,PETERSON M L.Non-invasive measurement of membrane morphology via UFDR:pore-size characterization[J].J Membr Sci,2004,239(1):143-154.
[16]  AYDINERA C,DEMIRAI,YILDIZB E.Modeling of flux decline in crossflowmicrofiltration using neural networks:the caseof phosphate removal[J].J Membr Sci,2005,248(1/2):53-62.

Full-Text

Contact Us

service@oalib.com

QQ:3279437679

WhatsApp +8615387084133