Neural network modeling and optimization of scheduling backwash for membrane bioreactor

Backwash scheduling for a membrane bioreactor was experimentally examined and theoretically modeled via neural networks. Flux was determined for different backwash and service time runs. Vacuum and backwashing streams pressure for different timing regimes were used to observe and monitor the fouling and a cake layer accumulated on the membrane surface causing a decline in the flux for the submerged membrane bioreactor. Experimental results were employed to develop an artificial neural network model (ANN) to predict the membrane flux as a function of the backwash and service times. Such modeling entails using a fairly large number of experimental data to reconcile model predictions with actual flux measurements in order to validate the ANN model. The ANN model was shown to be accurate in predicting the flux of the membrane and can be utilized to find optimum backwash scheduling strategy for such reactors.