Aplicação de aprendizado de máquina na análise de qualidade de água de caldeiras

Abstract

Water is a vital resource for industrial operations, and many industries rely on boilers to produce steam for various purposes such as power generation and heating. The quality of water used in boilers is crucial to ensure their efficiency and operational safety, as well as to prevent corrosion and scale buildup in the equipment. The purpose of this study is to evaluate the quality of water intended for boiler feed, assisting in identifying quality issues such as the presence of impurities or contaminants. This analysis will enable industries to take measures to prevent equipment damage, operational disruptions, and potential regulatory sanctions resulting from non-compliance with regulatory standards. Based on synthetic data, a machine learning model will be developed to classify water quality based on physicochemical parameters such as iron concentration, copper concentration, silica concentration, total alkalinity, total hardness, and conductivity, which are regulated by the governing bodies. The developed model provided an accurate categorization of water quality regarding its suitability for use in industrial boilers, verifying its compliance with the necessary regulations and technical specifications for the efficient and safe operation of boilers