Neural network simulation of the chemical oxygen demand reduction in a biological activated-carbon filter

This paper is primarily aimed at encouraging further use of neural networks by the water- and wastewater treatment industry. The study demonstrates the principle of using a network method of simulating the performance of a biological activated-carbon filter based on a biological water-quality assessment and measurements of pH and dissolved oxygen during the bio-regeneration mode with untreated river water. Protozoa, worms, rotifers, bacteria, fungi and algae were used as biological parameters. The neural network model could reasonably estimate the chemical oxygen demand reduction in an exhausted filter. The neural network model gave much better results than a second-order polynomial regression model; however, a much larger database is required than is currently available.