Computational intelligence applied to optimization of water treatment process parameters

• Grant number: 16/02641-1
• Support Opportunities: Regular Research Grants
• Start date: May 01, 2016
• End date: April 30, 2018
• Field of knowledge: Engineering - Electrical Engineering - Industrial Electronics, Electronic Systems and Controls
• Principal Investigator: Andre Felipe Henriques Librantz
• Grantee: Andre Felipe Henriques Librantz
• Host Institution: Universidade Nove de Julho (UNINOVE). Campus Memorial. São Paulo , SP, Brazil
• Associated researchers: Cleber Gustavo Dias ; Fábio Cosme Rodrigues dos Santos

Abstract

Water is an essential resource for maintaining the life. In order to avoid health problems, the water extracted from the natural sources has to be properly treated, so it can be used for consumption. The water treatment process consists of several steps, which include complex physical and chemical phenomena, which presents mostly nonlinear behavior.One of the most important steps is the coagulation process, which aims to electrically destabilize loads of unwanted particles in the water to be retained in the subsequent steps. The coagulant dosage reference is set according to the quality of raw water by jar-tests or test zeta potential measurements.Nevertheless, these methods are time consuming, besides demanding high costs of acquisition and equipment maintenance. An alternative way for setting coagulant dosage values is to use Computational Intelligence, such as metaheuristics techniques to extract knowledge from water treatment process data.In this work, the metaheuristics techniques Genetic Algorithm and Simulated Annealing are proposed for the development of an intelligent classifier parameters of water treatment processes from historical data that can consider all the parameters used to control the coagulant dosage for subsequent entry into a coagulant dosage prediction model, from which it is possible to set the coagulant dosage results regardless of the raw water quality scenarios and subsequent coagulation parameters, as well.This classifier can be used in selecting the data for training intelligent prediction models coagulant dosages in order to improve the accuracy of those models, which can contribute to the reduction of operational costs in water treatment processes. (AU)

Scientific publications

(References retrieved automatically from Web of Science and SciELO through information on FAPESP grants and their corresponding numbers as mentioned in the publications by the authors)

HENRIQUES LIBRANTZ, ANDRE FELIPE; RODRIGUES DOS SANTOS, FABIO COSME; DIAS, CLEBER GUSTAVO. Artificial neural networks to control chlorine dosing in a water treatment plant. ACTA SCIENTIARUM-TECHNOLOGY, v. 40, 2018. (16/02641-1)