Calibração multivariada e imagens digitais no controle de qualidade de farinha de trigo

Abstract

Wheat flour is one of the most consumed products in the world. In the wheat milling industries, different rheological and physical chemical analyzes are carried out to determine the flour specifications. These are routine analyzes that are time-consuming and are extremely important to the process. In this work, an alternative methodology was developed for these analyzes using digital images and multivariate calibration based on partial least squares (PLS). The proposed method was effective in determining the parameters of flour quality, with accuracy indicators evaluated through errors in the calibration and external validation steps, as well as the consideration of the adjust. The results obtained showed the possibility of using digital images combined with multivariate calibration by PLS to determine moisture, color, ash, fall time, gluten, strength, mass resistance to deformation, mass extensibility, tenacity, extensibility ratio, and elasticity index in wheat flour samples from the wheat milling industry. The models were linear, according to the residual plot that showed a random distribution of errors. The proposed method is an alternative to conventional ones for determining the described analyzes, highlighting the advantages of presenting lower cost, quickness in the determinations, and not generating residues.