Neural network identification of wastewater treatment plants

Abstract

Wastewater treatment plants (WWTPs) are highly complex systems. Therefore, it is difficult to predict the key parameters of water quality. Researches show that feed-forward neural networks have strong ability to approximate nonlinear functions. In order to predict the parameters of water quality, this paper proposes a modeling method by using artificial neural networks to predict the effluent quantity, including the concentration of chemical oxygen demand, biological oxygen demand and total suspended solid. The appropriate architecture of ANN models is determined through several steps of training and testing of the model. The performance of the artificial neural network model was assessed through the correlation coefficient (R) and mean square error (MSE). The results demonstrate that the proposed modeling method is effective and useful.