Machine Learning-Based Prediction of Textural Properties and Nonlinear Regulatory Pattern Analysis of 3D-Printed...

Area of Science:

• Food Science and Technology
• Materials Science
• Chemical Engineering

Background:

• 3D food printing precision relies on dough textural stability during extrusion.
• Understanding ingredient and process variable interactions is crucial for consistent food printing.

Purpose of the Study:

• To investigate the nonlinear effects of Konjac Glucomannan (KGM) concentration and printing pressure on 3D-printed dough texture.
• To develop predictive models for optimizing 3D-printed dough formulation and processing.

Main Methods:

• Employed a space-filling experimental design with 30 trials.
• Utilized Support Vector Regression (SVR) and Gaussian Process Regression (GPR) for modeling.
• Performed 4-fold cross-validation and hyperparameter optimization.

Main Results:

• KGM concentration and printing pressure showed significant nonlinear coupling effects on dough hardness, cohesiveness, and chewiness.
• SVR models achieved high predictive performance (Rp2 > 0.987 for gumminess and chewiness).
• A favorable processing region (0.5-0.8% KGM, 4.0-4.6 bar) was identified.

Conclusions:

• A quantitative, data-driven framework for 3D-printed dough formulation pre-optimization was established.
• The study provides insights into controlling textural properties for improved 3D food printing.
• Predictive modeling aids in achieving desired food product characteristics.