Chemometrics, data analysis, molecular spectroscopy (NIR...), multivariate calibration and classification
Agrobiodiversity and territorial systems
Our cooperation interest is on developing multivariate calibration and classification models.
Our collaboration can be devoted to several applications:
1) Related to the application:
- study of new raw materials that are alternative protein sources. The European Union (EU) has a deficit of sources of protein to produce feedstocks, which is partially solved by large imports of soybeans. To reduce the dependency from imports, it is urgent to find other sources of protein, and research is in progress about the use of insects, algae, and slaughter by-products.
- study of any type of potential food: olive oil, species, etc.
2) Related to the data analysis:
- development of nonlinear predictive models based on deep learning methodology (neural networks-ANN) and inverse modelling (ANNi), which are currently showing their applicability in multiple areas, among which is molecular spectroscopy.
- Working with big data set (data coming from more than 1 instrument) requires advances in the implementation of data fusion techniques.
The research group from the University Rovira I Virgili, Tarragona, spain, have three decades of expertise in teaching and research in chemometrics. They have worked both on the quantitative side (multivariate calibration of orders zero, one and two) and on the qualitative aspect (classification and discrimination), including aspects related to model optimization such as variable selection and data fusion. As far as the multivariate modelling (UV, NIR/MIR, Raman, etc.), the group has faced modelling challenges in previous research projects that range from the food industry to the petrochemical industry. The models were implemented and tested during real industrial production during the last stage of that project. From the knowledge transfer point of view, the URV team are yearly providing service to the NIR technology in a refinery in Tarragona for controlling the production of gasoline (1999-2021). We developed the multivariate models for a newly designed, patented (2019), instrument whose purpose is to detect bitter almonds in the reception of the raw product using NIRS. We also collaborated (2009-2011) with the Public Health Laboratory, Tarragona, Spain for the establishment of methods for the determination of Sudan dyes in food spices using multivariate techniques and with the company La Morella_Nuts S.A.U. (Tarragona, Spain) (2011-2015) for the establishment of methods based on NIRS for the detection of food fraud.