MODELING OF TRIHALOMETHANES (THMS) IN DRINKING WATER BY ARTIFICIAL NEURAL NETWORK

摘要

Chlorine is the main source for drinking water disinfection. In water, chlorine reacts with natural organic matter present in the raw water to form certain organohelides. These include trihalomethanes (THMs) and other halo compounds. These disinfection by-products (DBPs) are suspected to be carcinogenic. The chances of formation of THMs (CHCly CHCl2Br, CHClBr2 and CHBr3) are increased when excess of chlorine is available. Controlling the formation of DBFs is also essential to comply the drinking water qualities guidelines. Analysis of DBPs is time consuming and costly. Prediction of THMs concentrations using appropriate models for particular source water may be useful for monitoring THMs concentrations. For prediction of formation of THMs several modelling approach were used in the past. These include kinetic and regression based models. In the present study ANN based model is developed for the prediction of THMs. pH, TOC, chlorine dose and residual chlorine were considered as responsible for THMs formation and taken as independent variables in the modelling. Back propagation network which shows relation between independent variables has been considered and Stuttgart Neural Network Simulator (SNNS) was used for modelling.