Machine learning driven multi-objective optimization of cellulose acetate film composition for food packaging...

Area of Science:

• Materials Science
• Polymer Chemistry
• Biotechnology

Background:

• Biopolymer films require precise formulation for optimal performance.
• Cellulose acetate is a promising biopolymer for film applications.
• Controlling film properties necessitates understanding the role of additives.

Purpose of the Study:

• To optimize cellulose acetate bioplastic film properties.
• To model film characteristics using artificial intelligence.
• To determine ideal additive concentrations for specific film attributes.

Main Methods:

• Cellulose acetate films were formulated with varying concentrations of polyethylene glycol (PEG), malic acid (MA), and hexadecanoic acid (HAD).
• Film properties including thickness, water absorbency, transparency, and moisture content were characterized.
• Artificial neural networks (ANN) and ensemble regression trees were employed for property modeling and optimization.

Main Results:

• ANN models demonstrated high accuracy in predicting film properties (R²: 0.981-0.999).
• Individual optimization identified conditions for minimum thickness and maximum transparency.
• Multi-objective optimization proposed a specific blend of PEG, MA, and HAD for balanced film properties.

Conclusions:

• Predictive modeling accurately guides bioplastic film formulation.
• Optimized cellulose acetate films exhibit desirable characteristics for various applications.
• The study provides a framework for tailoring biopolymer film properties through controlled additive incorporation.