Prediction of anoxic sulfide biooxidation under various HRTs using artificial neural networks.

Abstract

Objective: During present investigation the data of a laboratory-scale anoxic sulfide oxidizing (ASO) reactor were used in a neural network system to predict its performance.

Methods: Five uncorrelated components of the influent wastewater were used as the artificial neural network model input to predict the output of the effluent using back-propagation and general regression algorithms. The best prediction performance is achieved when the data are preprocessed using principal components analysis (PCA) before they are fed to a back propagated neural network.

Results: Within the range of experimental conditions tested, it was concluded that the ANN model gave predictable results for nitrite removal from wastewater through ASO process. The model did not predict the formation of sulfate to an acceptable manner.

Conclusion: Apart from experimentation, ANN model can help to simulate the results of such experiments in finding the best optimal choice for ASObased denitrification. Together with wastewater collection and the use of improved treatment systems and new technologies, better control of wastewater treatment plant (WTP) can lead to more effective maneuvers by its operators and, as a consequence, better effluent quality.