Validação prática de modelos de infravermelho próximo para tomate: sólidos solúveis e acidez

Abstract

Tomatoes are the most commonly used and consumed vegetables, being acceptable for fresh consumption as well as for industrialization. Despite the wide acceptance of tomatoes, the consumer presents quality as a requirement for the acquisition of the fruit, which is measured through parameters such as soluble solids (SS) and titratable acidity (TA). A chemical analysis technique, non-destructive and non-invasive and with fast response, is an absorption spectroscopy in the near infrared region that has been widely used in several industries, from agricultural to petrochemicals. Demanding large-scale production, the quality of tomatoes demanded by consumers and as it is a non-destructive and noninvasive technique of near infrared spectroscopy (NIR), it is still necessary to test models under commercial conditions, ensuring a near infrared model for tomatoes making sorting easy. An external and practical validation of the near infrared models for fresh tomatoes was carried out, comparing them with the destructive methods and checking the accuracy of the models in qualifying the fruit in terms of soluble solids and acidity. This paper presents project models obtained from 2018 to 2019, built from the Model Builder and The Unscrambler programs and selected from parameters such as calibration coefficient, cross validation coefficient, average error of the calibration set, and cross validation and calculation of the residual calibration deviation. The results obtained through the prediction of the model for the soluble solid attributes and titratable acidity, of the table tomato with greater production variability. The model for SS of table tomato has potential for commercial use, either in determining the harvest point, or quantifying the vegetable’s quality. For AT or portable NIR, it did not produce an applicable model due to the wavelength limitation.