Controle de qualidade de águas potáveis utilizando análise multivariada de imagens

• Autor(a) principal: Damasceno, Deangelis
• Data de Publicação: 2015
• Tipo de documento: Tese
• Idioma: por
• Título da fonte: Repositório Institucional da UFG
• Texto Completo: http://repositorio.bc.ufg.br/tede/handle/tede/5558
• topic: Água potável; Controle de qualidade; Análise de imagens; Calibração multivariada; PLS; Potability of water; Physical chemical parameters; PLS; CIENCIAS EXATAS E DA TERRA::QUIMICA

New digital image based-analytical methodologies are proposed to measure pH, alkalinity, total hardness, calcium hardness, magnesium hardness, chloride, fluorine, and total iron of potable water samples. Multivariate image analysis and partial leastsquares regression were applied to BMP digital images acquired from a CCD-scanner. RGB, HIS, XYZ and YCbCr color spaces and 300 dpi images of 200 μL water samples were employed. This micro volume per sample yielded micro volumes of analytical waste per sample (400.0 μL). PLS root mean square error of prediction (RMSEP) for pH analyses was 0.16. RMSEP values for alkalinity, total hardness, calcium hardness, magnesium hardness, chloride, fluorine, and total iron were 0.03, 1.20, 2.01, 0.07, 0.04, and 0.06 mg L-1, respectively. Analytical figures of merit were computed for all PLS proposed methods. Mean relative errors ranging from 0.20% to 1.33 were found. The proposed methods were validated against standard analytical procedures for water quality control. There were no statistical differences between mean PLS value and the one found using the respective standard procedure (ttest, = 0.05). Precision was found statistically equivalent for pH, alkalinity, chloride, fluoride, and total iron when compared to the related reference method (Ftest, = 0.05). Therefore, the new PLS analytical methods proposed for water control quality can be employed as an alternative to standard methods.