ARTIFICIAL NEURAL NETWORK SIMULATION OF ADVANCED BIOLOGICAL WASTEWATER TREATMENT PLANT PERFORMANCE

Abstract

Artificial neural network (ANN) simulation of chemical oxygen demand (COD), total nitrogen (TN), and total phosphorus (TP) removal efficiencies of an advanced biological wastewater treatment process is presented in this study. Seven input parameters (predictors) were used: influent COD, TN, and TP concentrations, internal recycle (IR) and return activated sludge (RAS) ratios, wastewater temperature, and total hydraulic retention time (HRT) of process reactors. Results showed that open-source ANN tools can easily be employed for quick and reliable simulation results. ANN with the logistic, the sinc, and the Elliot functions can be confidently employed for predicting COD, TN, and TP removal efficiencies. Mean square errors were 5.54*10(-7), 2.06*10(-4), and 2.26*10(-3), respectively, for COD, TN, and TP removal efficiencies. Besides, wastewater temperature was found to be the major factor that determines the performance of a wastewater treatment system while RAS ratio, HRT, and influent wastewater characteristics are also effective on the performance.