Artificial neural network modeling of discrete settling process


Demir S. , Avşar Y. , Manav Demir N.

1st International Conference on Environment, Technology, and Management, Niğde, Turkey, 27 - 29 June 2019, vol.1, no.1, pp.147-156

  • Publication Type: Conference Paper / Full Text
  • Volume: 1
  • City: Niğde
  • Country: Turkey
  • Page Numbers: pp.147-156

Abstract

Performance of artificial neural network (ANN) in predicting efficiency of discrete settling process was evaluated in this study. For this purpose, a layered settling approach was used in a primary sedimentation tank and suspended solids (SS) removal efficiencies were calculated for various operating conditions by solving the mass balance equations. The removal efficiencies from the layered settling calculations were employed for training and validating the ANN. The results showed that an ANN with the logistic function satisfactorily predicts SS removal efficiency in discrete settling process at a given set of influent flowrate, split ratio, particle density, particle diameter, and water temperature. ANN predicted the SS removal efficiencies with mean square errors as low as 0.0064%.