[Modeling of sugar content based on NIRS during cider-making fermentation]

Area of Science:

• Food Science & Technology
• Analytical Chemistry
• Spectroscopy

Context:

• Cider fermentation involves dynamic changes in sugar content and matrix composition.
• Accurate and rapid monitoring of sugar levels is essential for process control and quality assurance.
• Traditional methods for sugar analysis can be time-consuming and labor-intensive.

Purpose:

• To develop and validate robust calibration models for predicting sugar content during cider fermentation using near-infrared spectroscopy (NIRS).
• To investigate the effectiveness of different spectral data pretreatment methods, specifically straight-line subtraction (SLS), for optimizing model performance.
• To establish models applicable to different fermentation stages (Stage I and Stage II).

Summary:

• Near-infrared spectroscopy (NIRS) data, collected between 12,000-4,000 cm⁻¹, was utilized with artificial neural networks to build sugar content calibration models.
• Optimal models were achieved using SLS pretreatment in specific spectral ranges (7,502–6,472.1 cm⁻¹ for Stage I; 6,102–5,446.2 cm⁻¹ for Stage II).
• The developed models demonstrated high accuracy, with correlation coefficients (R²) of 98.93% (Stage I) and 99.34% (Stage II), and low prediction errors (RMSEP of 4.07 g/L and 1.13 g/L, respectively).

Impact:

• The established NIRS models provide a rapid and accurate method for real-time monitoring of sugar content during cider fermentation.
• This technology can significantly improve process efficiency, consistency, and quality control in cider production.
• The findings offer a valuable tool for enologists and researchers in optimizing fermentation processes.