College of Geography and Environmental Sciences, Zhejiang Normal University, Jinhua 321004, China.
College of Geography and Environmental Sciences, Zhejiang Normal University, Jinhua 321004, China.
Bioresour Technol. 2019 Jun;282:262-268. doi: 10.1016/j.biortech.2019.03.044. Epub 2019 Mar 9.
It is of great importance to propose effective methods to quantify interfacial interaction since it directly determines foulant adhesion and membrane fouling process in membrane bioreactors (MBRs). This study developed a radial basis function (RBF) artificial neural network (ANN) to predict the interfacial interactions with randomly rough membrane surface. The interaction data quantified by the advanced extended Derjaguin-Landau-Verwey-Overbeek (XDLVO) approach were used as the training samples for the RBF networks. It was found that, the computing time consumption for the RBF network prediction was only about 1/50 of that for the advanced XDLVO approach under same conditions, indicating the high efficiency of the RBF ANN method. Meanwhile, the calculation accuracy of the method was acceptable to get reliable results. This study demonstrated the breakthrough of the fundamental methodology related with membrane fouling. The proposed RBF ANN method has broad application prospects in membrane fouling and interface behavior research.
提出有效的方法来量化界面相互作用非常重要，因为它直接决定了膜生物反应器（MBRs）中污染物的附着和膜污染过程。本研究开发了一种径向基函数（RBF）人工神经网络（ANN）来预测与随机粗糙膜表面的界面相互作用。通过先进的扩展德加古林-兰德威维尔-奥弗贝克（XDLVO）方法量化的相互作用数据被用作 RBF 网络的训练样本。结果表明，在相同条件下，RBF 网络预测的计算时间消耗仅为先进 XDLVO 方法的 1/50 左右，表明 RBF ANN 方法具有高效性。同时，该方法的计算精度可以接受，以获得可靠的结果。本研究证明了与膜污染相关的基础方法学的突破。所提出的 RBF ANN 方法在膜污染和界面行为研究中有广阔的应用前景。
