Regressão multivariada para determinação de sacarose na presença de cacau usando diferentes instrumentos de espectroscopia NIR
Arquivos
Data
2019-02-19
Autores
Título da Revista
ISSN da Revista
Título de Volume
Editor
Universidade Federal de Viçosa
Resumo
O objetivo deste trabalho foi realizar um estudo comparativo entre os dois tipos de instrumentos da técnica NIR (o NIR de bancada, com maior resolução (1000nm até 2500nm), e o portátil, com menor resolução (900nm até 1700nm)) e averiguar se o NIR portátil é um substituto ao NIR de bancada. A fim de elucidar a viabilidade, ou não, da utilização do NIR portátil, foram realizados experimentos de mistura entre cacau e sacarose (Experimento 1) e cacau, sacarose e frutose (Experimento 2). Para ambos os experimentos, observou-se que a diferença dos erros relativos entre os instrumentos (NIR portátil e de bancada) não foram tão expressivas, sendo, em média, 4% de diferença entre as amostras do NIR portátil e do bancada para o Experimento 1, e 6% para o Experimento 2. É importante dizer que quando se utiliza a mesma faixa espectral coincidente em ambos os instrumentos (1000nm até 1700nm), pode-se concluir que o NIR portátil é recomendado para estudos de mistura de cacau e sacarose, e para os que envolvem frutose na mistura. Considerando os resultados para as amostras dos produtos industrializados, observou-se que os modelos dos Experimentos 1 e 2, utilizando o NIR portátil, foram capazes de predizer, de forma significativa, os percentuais de sacarose correspondente nas embalagens dos produtos de interesse. Desta forma, o instrumento portátil apresentou ser uma boa alternativa para realizar as análises para predição de sacarose, considerando o custo-benefício, podendo-se reduzir custos com aquisição de instrumento e proporcionar rapidez e maior mobilidade para análises.
The objective of this work was to perform a comparative study between the two types of instruments of the NIR technique (the bench NIR, with higher resolution (1000nm to 2500nm), and the portable, with lower resolution (900nm to 1700nm)) and Portable NIR is a replacement for benchtop NIR. In order to elucidate the feasibility or otherwise of the use of portable NIR, experiments were performed between cocoa and sucrose (Experiment 1) and cocoa, sucrose and fructose (Experiment 2). For both experiments, it was observed that the difference of the relative errors between the instruments (portable and bench NIR) were not as expressive, being, on average, 4% difference between the portable and bench NIR samples for the Experiment 1, and 6% for Experiment 2. It is important to say that when using the same coincident spectral range in both instruments (1000nm to 1700nm), it can be concluded that portable NIR is recommended for studies of cocoa and sucrose , and for those involving fructose in the blend. Considering the results for the samples of the industrialized products, it was observed that the models of Experiments 1 and 2, using portable NIR, were able to predict, in a significant way, the corresponding percentages of sucrose in the packages of the products of interest. In this way, the portable instrument presented a good alternative to carry out the analysis for sucrose prediction, considering the cost-benefit, being able to reduce costs with instrument acquisition and to provide speed and greater mobility for analysis.
The objective of this work was to perform a comparative study between the two types of instruments of the NIR technique (the bench NIR, with higher resolution (1000nm to 2500nm), and the portable, with lower resolution (900nm to 1700nm)) and Portable NIR is a replacement for benchtop NIR. In order to elucidate the feasibility or otherwise of the use of portable NIR, experiments were performed between cocoa and sucrose (Experiment 1) and cocoa, sucrose and fructose (Experiment 2). For both experiments, it was observed that the difference of the relative errors between the instruments (portable and bench NIR) were not as expressive, being, on average, 4% difference between the portable and bench NIR samples for the Experiment 1, and 6% for Experiment 2. It is important to say that when using the same coincident spectral range in both instruments (1000nm to 1700nm), it can be concluded that portable NIR is recommended for studies of cocoa and sucrose , and for those involving fructose in the blend. Considering the results for the samples of the industrialized products, it was observed that the models of Experiments 1 and 2, using portable NIR, were able to predict, in a significant way, the corresponding percentages of sucrose in the packages of the products of interest. In this way, the portable instrument presented a good alternative to carry out the analysis for sucrose prediction, considering the cost-benefit, being able to reduce costs with instrument acquisition and to provide speed and greater mobility for analysis.
Descrição
Palavras-chave
Análise multivariada, Predição, Espectroscopia de infravermelho
Citação
SILVA, Raphael Henrique Teixeira. Regressão multivariada para determinação de sacarose na presença de cacau usando diferentes instrumentos de espectroscopia NIR. 2019. 31 f. Dissertação (Mestrado em Estatística Aplicada e Biometria) - Universidade Federal de Viçosa, Viçosa. 2019.