Projecte llegit

Títol: Introduccion y desarrollo de la vision hiperespectral SWIR en la determinacion de grasas en rosquillas fritas

---

Tutor/Cotutor: PUJOLA CUNILL, MONTSERRAT

Departament: DEAB

Títol: Introduccion y desarrollo de la vision hiperespectral SWIR en la determinacion de grasas en rosquillas fritas

Data inici oferta: 30-07-2021      Data finalització oferta: 30-03-2022

---

Estudis d'assignació del projecte:

GR ENG ALIMENT 2016

Lloc de realització:


En empresa (cal signar un conveni de cooperació)

        Tutor Extern: Juan Antonio Mena Gil
        Institució/Empresa: Bimbo Iberia SA

Paraules clau:
Validación, Soxhlet, SWIR, quimiometría, predicción

Descripció del contingut i pla d'activitats:
En los procesos industriales de fabricacion de rosquillas fritas se ha detectado una elevada desviación del contenido de grasa. En este trabajo se realizara un estudio del sistema de inspeccion por imagen hiperespectral para obtener informacion quimica y fisica de cada muestra obtenida en el proceso de fritura .

Overview (resum en anglès): BIMBO DONUTS IBERIA, S.A has been producing bakery-related products since 1963.
In recent years, a deviation in the fat percentage of fried donuts (Donuts) has been detected. The main goal of this project is to obtain a predictive model that allows the determination of the percentage of fat. This will be determined by a detection system that uses hyperspectral vision imaging technology in the Donuts production line.
There are numerous applications of the hyperspectral image interpretation system in the food industry, but they only apply to simple and / or intermediate raw materials or food products. This makes us think about the possibility of replacing the traditional ¿Soxhlet¿ fat determination system with new control systems based on the reading of the SWIR spectrum for more complex and / or semi-finished food products. Therefore, the aim of the work will be to obtain a predictive model using SWIR hyperspectral vision technology. In order to achieve this goal, a three-step methodology has been developed. A first part where a predictive model is developed at pilot level. The second part focuses on developing and integrating online, the system that allows you to estimate the percentage of fat and finally, a third stage that will validate the predictive model created. The results indicate that it is possible to estimate the percentage of fat in a fast and efficient way allowing a control in production line saving 25% of the costs of fats, one of the main raw materials of the production.
The development of the project will regulate the parameters and conditions of the production line that alter the percentage of fat depending on the results, achieving a more homogeneous and economically profitable product. The results obtained have made it possible to obtain a predictive model that allows the estimation of the percentage of fat of a fried donut with a significant degree of confidence and to check the effect of different parameters such as location, temperature, fermentation and kneading in the final result.

---