Predictive Biodegradation of Multiple Toxic Pollutants in Bioreactors Treating Real Wastewater using ANN and GP

Abstract

Industrial wastewater contains eco-toxic organic pollutants such as phenol, cyanide and xylenol, which are usually the most common toxic compounds from steel industries and their high concentrations negatively affects biological treatments through substrate inhibition. Optimizing the biodegradation of multiple substrates found in wastewater such as phenol, xylenol and cyanide under Aerobic (in presence of oxygen), Anaerobic (in absence of oxygen) and Anoxic (in presence of nitrates) biotreatment systems has been done in this study. To predict the performance and design control systems to meet the environmental regulatory standards, the complex non-linear behavior presented in the biological treatment requires an accurate model. Herein, by using Artificial intelligence (ANN, GA and GP), the mathematical modelling of biological reactors operating under batch and continuous systems treating phenol, xylenol and cyanide were attempted. Mathematical modelling of biological systems is complex due to lack of process knowledge and multiple governing factors thereby significantly increasing the research trends using data based driven models. To correlate the biological non-linear relationship and to achieve the minimum error goal for pollutant removal with an acceptable R2 value of 0.847, 0.892, 0.937 in the anoxic system for phenol, cyanide and xylenol removal, the ANN and GA were designed. These models represent, analyse and reliably make predictions of complex biological systems helping in controlling and enhancing the performance of the wastewater treatment plant. From these results, the better pollutant removal was done by the anoxic system when compared to aerobic and anaerobic systems especially for phenol and cyanide removal.