Cider fermentation process monitoring by Vis-NIR sensor system and chemometrics

- Optimization process of a multivariate calibration of a Vis-NIR sensor system.
- The sensor was developed for cider fermentation process monitoring.
- Most suitable pre-processing strategies selection.
- Variable selection methods were applied to obtain the finest calibration model.

Optimization of a multivariate calibration process has been undertaken for a Visible-Near Infrared (400-1100 nm) sensor system, applied in the monitoring of the fermentation process of the cider produced in the Basque Country (Spain). The main parameters that were monitored included alcoholic proof, l-lactic acid content, glucose + fructose and acetic acid content. The multivariate calibration was carried out using a combination of different variable selection techniques and the most suitable pre-processing strategies were selected based on the spectra characteristics obtained by the sensor system. The variable selection techniques studied in this work include Martens Uncertainty test, interval Partial Least Square Regression (iPLS) and Genetic Algorithm (GA). This procedure arises from the need to improve the calibration models prediction ability for cider monitoring.