Modelagem e predição de propriedades mecânicas e físicas de fibras lignocelulósicas por FTIR via regressão multivariada
Nenhuma Miniatura disponível
Data
2025-08-27
Autores
Lauriano, Laiza Andrade
Título da Revista
ISSN da Revista
Título de Volume
Editor
Universidade Federal do Espírito Santo
Resumo
In the face of contemporary challenges, the pursuit of sustainable materials that minimize environmental impact is essential. In this context, this study evaluated the use of lignocel lulosic fibers in composites as an alternative to synthetic materials. Lignocellulosic fibers are notable for their advantageous characteristics, such as biodegradability, low cost, and competitive mechanical performance in specific applications, such as the reinforcement of polymer composites used in automobiles and civil construction. However, determining the mechanical and physical properties of lignocellulosic fibers remains a challenge, primarily due to their microscopic dimensions, heterogeneity, and the complexity of experimental testing. Despite these difficulties, detailed knowledge of these properties is fundamental for the efficient selection and application of fibers in polymer composites. Given this scenario, this dissertation proposes the development of Principal Component Regression (PCR) and Partial Least Squares (PLS) mathematical models to predict physical and mechanical properties from spectral data obtained by Fourier-Transform Infrared (FTIR) Spectroscopy, thereby optimizing the characterization process. The models showed satisfactory performance in explaining the variability of data for diameter, density, and tensile strength, with PLS showing the best results for diameter (R2 = 0,766 e Q2 = 0,811) and density (R2 = 0,874 e Q2 = 0,877), and PCR for tensile strength (R2 = 0,721 e Q2 = 0,722). The results highlight the potential of chemometrics as an auxiliary tool for the selection and application of lignocellulosic fibers in sustainable composites.
Descrição
Palavras-chave
Modelagem quimiométrica , Regressão por componentes principais , Mínimos quadrados parciais