Article: Artificial neural networks for forecasting wastewater parameters of a common effluent treatment plant Journal: International Journal of Environment and Waste Management (IJEWM) 2019 Vol.24 No.3 pp.313 - 336 Abstract: This paper employs artificial neural networks (ANN) to evolve a framework wherein advance prediction of common effluent treatment plant (CETP) performances can be made using process variables such as BOD, TSS, and pH. To illustrate the efficacy of the framework, ANN models were applied to the case of a CETP having designed treatment-capacity of 900 m<SUP align="right"><SMALL>3</SMALL></SUP>/day. The data was collected over a period of 5-years from the influent and effluent streams for the CETP wherein eight industries discharge their wastewaters. It was observed that multilayer perceptron with online back-propagation algorithm having hyperbolic-tangent function for both hidden and output layers gives excellent results. All ANN models learnt rapidly with training speeds as high as 500 iterations/second. Thus, ANN-based models proved efficient and robust tools giving R values upto 0.98. The evolved models were then used to prepare input importance tables to delineate contributions of load of individual industries into the CETP for evolving a sustainable financial mechanism to charge industries in accordance with their respective loads. Such forecasting may also be beneficially used to curtail the need of continuous monitoring of the CETP, thereby resulting in significant savings besides reducing perpetual dependence on operator-based real time monitoring. Inderscience Publishers - linking academia, business and industry through research

Title: Artificial neural networks for forecasting wastewater parameters of a common effluent treatment plant

Authors: Monika Vyas; Mukul Kulshrestha

Addresses: Environmental Engineering Division, Department of Civil Engineering, MANIT-Bhopal, India ' Environmental Engineering Division, Department of Civil Engineering, MANIT-Bhopal, India

Abstract: This paper employs artificial neural networks (ANN) to evolve a framework wherein advance prediction of common effluent treatment plant (CETP) performances can be made using process variables such as BOD, TSS, and pH. To illustrate the efficacy of the framework, ANN models were applied to the case of a CETP having designed treatment-capacity of 900 m³/day. The data was collected over a period of 5-years from the influent and effluent streams for the CETP wherein eight industries discharge their wastewaters. It was observed that multilayer perceptron with online back-propagation algorithm having hyperbolic-tangent function for both hidden and output layers gives excellent results. All ANN models learnt rapidly with training speeds as high as 500 iterations/second. Thus, ANN-based models proved efficient and robust tools giving R values upto 0.98. The evolved models were then used to prepare input importance tables to delineate contributions of load of individual industries into the CETP for evolving a sustainable financial mechanism to charge industries in accordance with their respective loads. Such forecasting may also be beneficially used to curtail the need of continuous monitoring of the CETP, thereby resulting in significant savings besides reducing perpetual dependence on operator-based real time monitoring.

Keywords: artificial neural networks; ANN; process variables; common effluent treatment plant; CETP; modelling; forecasting.

DOI: 10.1504/IJEWM.2019.103106

International Journal of Environment and Waste Management, 2019 Vol.24 No.3, pp.313 - 336

Received: 02 Jan 2018
Accepted: 29 Dec 2018
Published online: 15 Oct 2019 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article

Title: Artificial neural networks for forecasting wastewater parameters of a common effluent treatment plant

Keep up-to-date