Developing artificial neural network models of water treatment processes: a guide for utilities

Abstract

Because of the complex nature of drinking water treatment unit processes, utilities have difficulty quantifying the interactions and relationships that exist between process inputs and process outputs. Process models, where they exist, are often site specific and are unable to simultaneously handle continuous variations in more than one or two key process variables. The artificial neural network (ANN) technology is a robust artificial intelligence technology that can handle the complex and dynamic nature of treatment processes. As such, the technology has been gradually gaining acceptance in the drinking water treatment industry as a tool for process modelling and control. While publications on modelling results and applications abound, a detailed account of ANN modelling methodology is lacking. Presented is a detailed methodology for developing successful ANN models of drinking water treatment processes. The utility and applicability of this methodology is demonstrated through a case study where a successful ANN model to predict filtration performance was developed. Key words: artificial neural networks, process modelling, process optimization, water treatment.