Chapter 22 - Modeling Undefined Complexities of Wastewater Treatment Processes With Artificial Neural Network

Abstract

A number of physical, chemical, and biological processes are used for wastewater treatment, and several variables that define these unit operations and processes are controllable. However, wastewater quality and interactions among process variables affect the treatment efficiency. Modeling and simulation of these processes by conventional mathematical modeling techniques are difficult due to complexity of these processes. Artificial neural networks (ANNs) have shown the ability to model complicated and nonlinear processes, including the complex behavior in treatment operation and processes. This chapter presents the use of ANN in modeling wastewater treatment processes. Neural networks such as feed forward backpropagation which are proved to be effective in modeling treatment processes are presented. The models are trained for various types of networks, topography, training algorithms, and transfer functions to obtain a general predictive model. This chapter also summarizes the use of ANN in wastewater treatment showing the better performed models for predictability. Applications of ANN models for prediction and optimization of different processes for removing pollutants, including toxic metals and dyes, and for modeling anaerobic processes, are also presented. Finally, the chapter provides a perspective on the directions for future research in utilizing ANN with other modeling techniques to improve predictability with cost reduction.