Proc. 18th Int. Conf. Near Infrared Spectrosc., pp. 115–122 (2019)
The control of conformity at reception of bulk olive oil before unloading is an essential step for packing plants to meet quality standards and ensure traceability. New approaches, different to sampling and analysis at-line, are needed to improve efficiency and decision-making processes. Near-infrared spectroscopy (NIRS) provides fast, cost-effective and in situ analytical measurements without sample preparation, which makes it an ideal technology towards this goal. The purpose of this work is to assess and optimize the spectral acquisition with a fiber-optic probe, especially designed to sample and analyze tank trucks. This probe features two fiber optic bundles, one of measurement and one of illumination. First of all, two olive oils samples, one of extra virgin and one of lampante category, were used to evaluate the noise and repeatability of the NIRS spectral signals of the target probe. A set of 20 spectra were recorded for each sample, and the sequence of measurements lampante-extra virgin was repeated three times per day for seven days spread over five weeks. The noise level was evaluated using a first derivative pre-treatment, and the optimum working range (after removing the noisy regions) was 1150–2149 nm. The Root Mean Square (RMS) statistic and the MEAN values were calculated in each case, and the spectral repeatability results showed that the RMS (MEAN) values for lampante were higher than for extra virgin olive oil (12,630 and 2998 μlog(1/R), respectively). Furthermore, a standardization procedure was performed to make a comparative study with a laboratory probe. Both probes were connected to a MATRIX-F Fourier transform-NIR instrument. A set of 42 samples of different nature were used for standardization purposes. Two matrices of standardization were developed, selecting 1 (STD1) and 10 (STD2) samples, and a validation set of 32 samples was used. The spectral adjustment was evaluated using the validation set by calculating the RMS(c) statistic. The RMS(c) values obtained in the master (laboratory probe, LP) versus satellite (in situ analysis probe, IAP) before standardization showed important differences that were significantly corrected by the best standardization matrix (STD1).