Modeling photosynthetically oxygenated biodegradation processes using artificial neural networks

The complexity of the mechanisms underlying organic matter mineralization and nutrient removal in algal–bacterial photobioreactors during the treatment of residual wastewaters has severely hindered the development of mechanistic models able to accurately describe these processes. Artificial neural networks (ANNs) are capable of inferring the complex relationships existing between input and output process variables without a detailed description of the mechanisms governing the process, and should therefore be more suitable for the modeling of photosynthetically oxygenated systems. Thus, a neural network consisting of a single hidden layer with four neurons accurately predicted the steady-state operation of a continuous stirred tank photobioreactor during salicylate biodegradation by an algal–bacterial consortium. Despite its simplicity and the low number of data sets for ANN training (23), this network topology exhibited a satisfactory fit for both training and testing data with correlation coefficients of 99%. Although the use of ANNs for modeling conventional wastewater treatment systems is not novel, this work constitutes, to the best of our knowledge, the first reported application of ANNs to photosynthetically oxygenated systems and one of the few models for microalgae-based treatment processes. This modeling approach is therefore expected to contribute to improve the understanding of the complex relationships between light, temperature, hydraulic retention time, pollutant concentration and process removal efficiency, which would eventually promote the development of algal–bacterial processes as a cost effective alternative for the treatment of industrial wastewaters.