Plastics Analysis, Sorting & Recycling Technologies Through Intelligent Classification

Plastic remains an invaluable material for human society in many key areas, but the challenge of sustainable end-of-life disposal routes continues to exist. Accurate sorting and the production of high quality recyclate is a key target to ensure confidence and security in recycled plastic supply chains and subsequent manufacturing industries. This is essential for society to reach targets for the inclusion of recycled and recyclable content across sectors such as automotive and packaging.

To meet this challenge, an infrastructure of knowledge-led and digitally enabled systems that underpin future manufacturing needs to be developed.

Our previous work on AI & machine learning (ML) for sorting highlighted a need for practical solutions and we were the first to demonstrate the potential of deep learning methods to solve the sorting problem, followed up by demonstrating that using multiple data sources (e.g. IR, Raman and LIBS) can improve overall performance.

While classification accuracy has improved significantly with our work, the quality and quantity of recyclate use is still a challenging problem. The aim of this project is to enhance the previous work on sorting with an improved understanding of extrusion compounding through in-line rheometry for real-time monitoring of shear viscosity. The integration of real-time rheological data with ML principles will form the basis of an intelligent recycling system, backed up by measurable sustainability credentials with the aim of delivering a new approach that delivers on both recycled content and net zero targets.

This project is aligned with the EPSRC-funded project "PLASTIC - Plastics Analysis, Sorting & Recycling Technologies Through Intelligent Classification"