人工神经网络在量化膜污染界面作用力中的应用  

Application of artificial neural networks in quantifying interface interaction energy associated with membrane fouling in a membrane bioreactor

作　　者：陈镒锋 申利国[1] 林红军[1] CHEN Yifeng;SHEN Liguo;LIN Hongjun(College of Geography and Environmental Sciences, Zhejiang Normal University, Jinhua 321004, China)

机构地区：[1]浙江师范大学地理与环境科学学院,金华321004

出　　处：《膜科学与技术》2020年第3期47-54,共8页Membrane Science and Technology

基　　金：国家自然科学基金资助项目(51978628)。

摘　　要：膜生物反应器(MBR)中膜污染的黏附过程由污染颗粒和粗糙膜表面之间的界面作用力决定.高级extended-Derjaguin-Landau-Verwey-Overbeek(XDLVO)方法无法快速实现界面作用力的量化.本研究提出了径向基函数(RBF)人工神经网络(ANN)、反向传播(BP)ANN和广义回归神经网络(GRNN)3种模型用于量化这些界面作用力.RBF ANN,BP ANN和GRNN的预测结果均具有较高的回归系数和准确性,表明它们能很好地捕捉界面作用力与各种因素之间复杂的非线性映射关系.与高级XDLVO方法相比,RBF ANN,BP ANN和GRNN的量化效率均得到了显著提高.同时,BP ANN的预测性能优于RBF ANN和GRNN模型.案例研究进一步证明了BP ANN用于界面作用力定量分析的可行性.本研究为量化与膜污染相关的界面作用力提供了新方法.Interfacial energy between sludge foulants and rough membrane surface critically determines adhesive fouling in membrane bioreactors(MBRs).As a current available method,the advanced extended Derjaguin-Landau-Verwey-Overbeek(XDLVO)approach cannot efficiently quantify the interfacial energy.In this study,novel methods including radial basis function(RBF)artificial neural network(ANN),back propagation(BP)ANN and generalized regression neural network(GRNN)were proposed to quantify the interfacial energy associated with the membrane fouling in an MBR.The prediction results of RBF ANN,BP ANN and GRNN have high regression coefficients and accuracies,suggesting their high capacity to capture the complicated non-linear mapping relations between interfacial energy and various factors.As compared with the advanced XDLVO approach,both RBF ANN,BP ANN and GRNN showed remarkably improved quantification efficiency.Meanwhile,BP ANN showed better prediction performance than RBF ANN and GRNN model.Case study further demonstrated the robustness and feasibility of BP ANN for interfacial energy quantification.This study provided a new approach to quantify interfacial energy associated with membrane fouling.

关 键 词：人工神经网络 膜生物反应器 膜污染 界面作用力 

分 类 号：TQ028.5[化学工程] X703.1[环境科学与工程—环境工程]