Desenvolvimento de método alternativo paraquantificação de clorofilas em indústria de óleo desoja utilizando imagens digitais

• Ano de defesa: 2019
• Autor(a) principal: Bassetti, Bruna Daniely
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Tecnológica Federal do Paraná; Campo Mourao; Brasil; Programa de Pós Graduação em Inovações Tecnológicas; UTFPR
• Palavras-chave em Português: Quimiometria; Óleos vegetais; Calibração; Chemometrics; Vegetable oils; Calibration; CNPQ::CIENCIAS AGRARIAS; Engenharia/Tecnologia/Gestão
• Link de acesso: http://repositorio.utfpr.edu.br/jspui/handle/1/5159

Chlorophyll is the most abundant pigment in nature whose main role is the absorption of sunlight and its conversion to chemical energy during photosynthesis. It is present in various plant species and among them are oilseeds, which provide us with edible oils. Due to the demand from consumers that vegetable oils have a clear appearance, there is a need to remove these pigments in their processing. Thus, there is the bleaching step, which by adding clarifying earth, remove this pigment. To ensure that refined oils are within the industry specified range, analyzes are performed using a spectrophotometer. The method consists of taking wavelength readings at which the green color is emitted (630, 670 and 710 nm) using dichloromethane as a white test. In order to streamline the process and decision making and reduce reagent consumption, it is proposed to use an alternative method from digital webcam images associated with chemometric methods for chlorophyll determination in soybean oil industries. Each image generates a tensor of numerical data that is broken down into color histograms and modeled by partial least squares regression (PLS). The proposed method was effective in the determination of chlorophylls, with accuracy indicators evaluated through errors in the calibration and external validation stage, as well as the consideration of the adjustment. In addition, the model can detect chlorophyll concentrations of 10.48 ppb and is able to accurately and accurately quantify samples with chlorophyll concentrations greater than 31.77 ppb.