Mushroom Preservation with Ai-Optimized Ultrasound-Assisted Freezing

Abstract

This study investigates ultrasound-assisted freezing (UIF) optimization for button mushrooms using probe-type ultrasonication, integrating machine learning (ML) and statistical analysis to improve freezing efficiency and quality. Ultrasound power levels (50–400 W) in continuous and pulse modes were evaluated for their effects on freezing time, pH, total soluble solids (TSS), drip loss, and textural properties (hardness, gumminess, chewiness, resilience). ML models—Support Vector Regression (SVR), Random Forest Regressor (RFR), and Gradient Boosting Regressor (GBR)—predicted outcomes, with SVR excelling in freezing time (R² = 0.9986, MSE = 0.0099), RFR in pH (R² = 0.9136, MSE = 0.0242), and GBR in TSS (R² = 0.8087, MSE = 0.0536). SHAP analysis highlighted ultrasound power as the primary factor (72% contribution to freezing time reduction) and sample weight for resilience (31%). ANOVA (p < 0.05) and Tukey’s HSD test confirmed significant treatment differences, with UC200 reducing freezing time by 8.213 seconds and UC400 minimizing drip loss by 4.411%. Hardness peaked at 400 W, resilience at 200 W. AI-driven UIF optimization offers an energy-efficient, precise freezing solution, advancing food preservation technologies for the frozen food industry.

Keywords: Machine Learning Prediction, Ultrasound Freezing Technology, Food Quality Preservation, Button Mushroom Processing