Predicting membrane fouling in a high solid AnMBR treating OFMSW leachate through a genetic algorithm and the optimization of a BP neural network model

Anaerobic membrane bioreactors are a promising technology in the treatment of high-strength wastewater; however, unpredictable membrane fouling largely limits their scale-up application. This study, therefore, adopted a backpropagation neural network model to predict the membrane filtration performance in a sub-merged system, which treats leachate from the organic fraction of municipal solid waste. Duration time, water yield flow, influent COD, pH, bulk sludge concentration, and the ratio of & UDelta;TMP to filtration time were selected as input variables to simulate membrane permeability. The membrane pressure slightly increased by 1.1 kPa within 62 days of operation. The results showed that the AnMBR membrane filtration performance was acceptable when treating OFMSW leachate under a flux of 6 L/(m(2).h). The model results indicated that the sludge concentration largely determined the membrane fouling with a contribution of 33.8%. Given the local minimization problem in the BP neural network process, a genetic algorithm was introduced to optimize the simulation process, and the relative error of the results was reduced from 5.57% to 3.57%. Conclusively, the artificial neural network could be a useful tool for the prediction of an AnMBR that is so far under development.

Automated summary

This study used a backpropagation neural network model to predict the membrane filtration performance in a submerged system and applied a genetic algorithm to optimize the simulation process. The results showed that artificial neural networks can be a useful tool for predicting AnMBR.