Modelling Milk Lactic Acid Fermentation Using Multivariate Curve Resolution-Alternating Least Squares (MCR-ALS)

The purpose of the current work was to investigate the capability of multivariate curve resolution-alternating least squares (MCR-ALS) to extract relevant information from Fourier transform near-infrared (FT-NIR) spectra acquired on-line with a fibre probe during milk lactic acid fermentation. The fermentation trials were conducted replicating twice a factorial design with three different starter cultures (Streptococcus thermophilus and Lactobacillus bulgaricus alone or as 1:1 mixed culture) and three different incubation temperatures (37, 41 and 45 °C), for a total of 18 experiments. The runs were monitored for 7.5 h through pH measurement, dynamic oscillatory test for rheological properties evaluation and FT-NIR spectra acquisition. The obtained MCR-ALS models successfully described the experimental FT-NIR spectra recorded (99.9 % of explained variance, 0.63665 % lack of fit, and standard deviation of the residuals lower than 0.0072). The three spectral profiles obtained by MCR-ALS pointed to the characteristic coagulation phases of milk lactic acid fermentation. The concentration profiles defined as a function of time for each run were strongly dependent on starter and temperature tested, in agreement with pH and rheological results. MCR-ALS applied to FT-NIR spectroscopy provides to the dairy industry a control system which could be implemented in-line as reliable management method for monitoring fermentation processes and to define the coagulation profile no matter the operative conditions adopted for the process.