Accelerated optimization of PET PCR formulations using active learning

Overview

Accelerated optimization of PET PCR (post-consumer recycled) formulations is now achievable through the integration of machine learning (ML) and active learning techniques. This innovative solution targets the injection molding industry, where identifying optimal material properties like viscosity and tensile strength is crucial. By combining data from pristine and PCR content, the platform predicts material properties and suggests experiments to refine formulations, significantly reducing development time and enhancing product performance.

Technical specifications

Key features:

• Utilizes Citrine Informatics' platform for materials and chemicals informatics
• Integrates published and simulated data on pristine and PCR properties
• Employs active learning to iteratively improve ML model accuracy and predict material properties
• Suggests experiments with ranked importance to optimize formulations
• Capable of functioning with minimal data, starting with as few as 30 data points

Technology readiness level

This solution is at Technology Readiness Level 6, indicating it has been demonstrated in relevant environments and is ready for further validation with industry partners. The platform has been successfully used by major companies, highlighting its reliability and effectiveness in accelerating material development processes.