Modeling of Permeability of Membrane Bioreactor System Using Artificial Neural Network

Due to the possibility of executing large number of virtual tests over a finite/short time, modeling of complex systems such as membrane bioreactor is considered as a powerful tool, though it requires experimental validation and reliable conversion of the process to a true mathematical model. This paper investigates filtering process modeling via neural network using MATLAB 8.1 (2013), in which the experimental data achieved from a submerged membrane bioreactor (MBR) system equipped with Kubota membranes are used for municipal wastewater treatment with high mixed liquor suspended solids (MLSS) concentration. More precisely, two third of empirical data are employed for training—using trainlm algorithm, test and validation of the network. Then, the designed network is used to estimate the permeability of both the rest one third of data and another typically similar MBR system. The value of determination coefficient (R2) for predicting the permeability of one third of data in the first system is 0.93, and for the second systems is 0.92, attesting the adequacy and effectiveness of the proposed method

کلیدواژه ها:

membrane bioreactor ، resistances in series ، filtration process ، neural network ، permeability