Avaliação da correlação entre propriedades colorimétricas e propriedades físico-mecânicas de madeiras tropicais

Abstract

This work aimed to evaluate the use of colorimetric properties (L*, a* and b*) of the CIELAB system and apparent density as estimators of physical and mechanical properties of ten tropical wood species using quadratic multivariable regression models. The models were evaluated following the ANOVA (Analysis of Variance) precepts, as well as the normality of the data. Therefore, analyzes were made of how physical and mechanical properties and colorimetric parameters are related. First, the correlation between colorimetric parameters and physical and mechanical properties was evaluated using the Pearson test. The goodness of fit was measured using the coefficient of determination (R²). In order to understand how physical and mechanical properties and colorimetric parameters are related, linear and quadratic models with an independent variable (L* or a* or b*) were generated. After understanding how the colorimetric parameters are related and verifying the independence between parameters and properties, linear and quadratic models were generated in order to see if the use of quadratic models improves the accuracy of the estimates. After the evaluation of the quadratic and linear models, the quadratic multivariable models were validated with data from nine species and the properties of the tenth species were estimated and compared with the experimental data. After validation, quadratic multivariable models were generated with data from the ten species evaluated and the error of the estimates was evaluated using the models obtained as data estimators in the literature. Additionally, the behavior of apparent density and its relationship with physical and mechanical properties was evaluated throughout the entire evaluation. The result of Pearson's correlation showed that the colorimetric parameters, as well as the apparent density have significance in the correlation, indicating a significant relationship between factors and properties, but with a weak correlation. Quadratic models, compared to linear models, increased the significance of the estimate, except for the L* parameter. The results of the validation models showed that fc0, fH0 and fH90 presented good quality in the fit (R²>70%) using only the parameters. Adding density, EM also arrived in good quality in the fit, indicating the possibility of its use. For the models with data from the 10 species, the estimator models of the properties fc0, fH0 and fH90, Ec0 and EM achieved good quality in the adjustment using only the colorimetric parameters, improving the precision with the addition of the apparent density as an independent variable. The comparison with the data estimated by the models and the experimental values show that it is possible to estimate physical and mechanical properties using colorimetric parameters and apparent density.