Artificial neural network modelling in biological removal of organic carbon and nitrogen for the treatment of slaughterhouse wastewater in a batch reactor.

Wastewater containing high concentration of oxygen-demanding carbonaceous organics and nitrogenous materials (chemical oxygen demand (COD) and total Kjeldahl nitrogen (TKN)) as nutrients emanated from small- to large-scale slaughterhouse units cause depletion of dissolved oxygen in water bodies and attributes to the threat of eutrophication. Biological treatment of wastewater is a useful tool through ages for the treatment of wastewater owing to its cost-effectiveness, reliability along with its innocuous output features. This paper deals with the treatment of slaughter house wastewater by conducting a laboratory scale batch reactor with different input characterized samples, and the experimental results were explored for the formulation of feed-forward back-propagation artificial neural network (ANN) to predict the combined removal of COD and TKN. The ANN modelling was carried out using neural network tool box of MATLAB (version 7.0), with the Levenberg-Marquardt training algorithm. Various trials were examined for the training of the ANN model using the number of neurons in the hidden layer varying from 2 to 30. The mean square error function and regression analysis were also applied for performance analysis of the ANN model. All the input data were logged-in after carrying out detailed experiment in the laboratory with a view to examine the performance of the batch reactor for the treatment of slaughterhouse wastewater. The experimental results were used for testing and validating the ANN model.