Neural network modelling and prediction of an Anaerobic Filter Membrane Bioreactor

Anaerobic membrane bioreactors have become an environmentally friendly solution for wastewater treatment. The lack of sufficiently accurate mathematical procedures to model their behaviour and the fouling process membranes, poses a challenge when trying optimise energy consumption maintenance costs. An membranes is critical make most this technology. This perfect scenario in which introduce neural networks (NN) as alternative modelling. However, duration experiments difficulties measuring some relevant variables, it hard collect high quality datasets train NN. Our goal obtain good prediction status enable adjustment operation conditions ahead time. To do so we must our networks. combination static dynamic enables us leverage best capabilities each one. requires preprocessing that separates trends from oscillations. outputs obtained need be put together build up predicted evolution fouling. Accurate predictions are then extended 25 75 filtration cycles. maintain even extend accuracy after sudden changes operating conditions, retraining NN every cycles proposed. AI based real time open new scope decision making, optimisation field anaerobic reactors.