Prediction of physico-chemical indices in intact olives and olive paste by means of visible/near infrared spectroscopy

The challenge of producing a high quality extra virgin olive oil is of great interest and the selection of olive fruit with defined properties which ensure positive attributes in the olive oil and the real time process control are foreseeable in the production chain. The investigation of the olives characteristics before the milling process could allow to control the quality of the oil output. A better monitoring of the oil production process also depends on controlling of the paste, the intermediate product between the olives inlet in the process and the oil outlet from the mill, to establish correlations among olives, paste and oil. Nowadays, established methods for quality assessment are generally based on tedious and time consuming techniques that are impractical for processing a big number of samples. Therefore, there is a lack of real time information during the milling process in order to keep monitored the operating parameters. The aim of the work was the application of vis/NIR spectroscopy in order to correlate the spectral data, acquired on intact olives just before or on pastes during the milling process, to the crucial parameters for the optimization of the milling process. A total of 54 samples were collected and each sample was composed of 300g of olive fruits randomly collected from the selected plants. The olives were collected weekly. Spectral acquisitions were performed using an optical portable system operating in the 400-1000 nm wavelength range. Two spectral measurements were taken in reflectance mode on individual fruit along their equator region. Physical measurements (yield point force and total deformation energy) were performed on olives samples, while, after fruits crushing for olive paste production, traditional chemical analysis (moisture, oil and sugars content) and maturity index (MI) were measured and correlated to spectral data. The correlations between spectral data and chemical parameters were carried out also considering the spectra acquired on olive pastes. The olive samples available were used for the calculation of a chemometric regression model for reference parameters by using the partial least squares (PLS) regression analysis. The vis/NIR spectra acquired on single olive were correlated to the texture parameters (one-to-one correlation) using the PLS regression algorithm, while the 30 olive spectra representing each experimental sample were averaged and the resulting mean spectrum was correlated to the chemical indices and MI. Similarly, olive paste samples were also correlated to chemical reference data and to MI to create PLS models. The obtained results were encouraging whether for chemical, texture and MI parameters. The study showed the feasibility of real time prediction of important indices. The sector could be helped by the adoption of rapid analytical systems for olive fruits and olive paste for a quick evaluation of crucial features in a view of a better control of the milling process.