A Machine Learning System for Universal Primer Coating Optimal Formula Prediction

Traditional computer-aided formulation for coatings applies Design of Experiment principles to efficiently sample a broad formula space, but this approach distributes experiments evenly across the entire space. In contrast, the machine learning system predicts and suggests new formulas in unexplored areas, reducing the need for extensive experimentation. This machine learning system predicts optimal formulas for universal primer coatings using experimental data as its learning base. The dataset can be expanded for improved accuracy or adapted to other coating types. The system employs multiple models trained on experimental results to deliver highly accurate performance predictions. Additional experiments can be incorporated to further enhance accuracy if required.

 Data-driven framework for universal primer formulation optimization

Benefits

• Combines the strengths of different models, reducing errors and increasing prediction accuracy.
• Makes the model focus on meaningful patterns grounded in chemistry and physics. Reduces the risk of the model learning noise or irrelevant correlations.
• Efficiently finds optimal formulations with fewer experiments.

Applications

• Industrial coating optimization
• Material discovery and customization
• Green and sustainable coating development

Patents