Scientists from Aarhus University have developed a camera technology that can provide significant benefits in the field of plastic recycling. With this camera, which can distinguish between 12 different types of plastic, it can facilitate the classification and separation of waste plastics.
The study, published in the scientific journal Vibrational Spectroscopy, used unsupervised machine learning on shortwave infrared hyperspectral data to create a model for classifying plastics.
The model can successfully distinguish 12 different plastics commonly found in homes, such as PE, PP, PET, PS, PVC, PVDF, POM, PEEK, ABS, PMMA, PC, and PA12. The technology was also able to identify three more unknown samples, further proving its usefulness, the researchers said.
Generally, the plastic recycling process involves reducing the material and then mechanically separating it with density tests or using near-infrared technology. However, the plastic obtained in this way is between 75-and 95% pure. However, a plastic purity of at least 96% is expected in the industry.
The new technology, led by Associate Professor Mogens Hinge, can identify a wider range of plastic types than NIR technology and also classifies chemical purity in the composition, a promising development for the plastic recycling industry.
Hinge summed up the tech’s process in an interview with Resource: Basically, it’s a camera viewing a conveyor belt. The plastics are then transported from the side of the camera. When the camera takes the images, we use unsupervised machine learning to analyze the images and detect and distinguish individual types of plastic.
“The camera is special because it records images within the infrared range and with multiple channels. For reference, a mobile phone has three channels that make up an image – red, blue, and green. Our IR camera has 90 channels.
“I just want to point out that we use industrial components consciously – this includes all the components and systems in our installation. This means that we do not use highly sensitive, dedicated, specialized, and sensitive research equipment for our work. This restriction is made to ensure the industrial suitability of our work.”
He then explains that the system is “directly transferable to industries.” “Apart from the camera, all you need is a stand to place it on a conveyor belt on the production line. Due to its ability to detect a range of plastic types, the technology allows for the separation of unwanted impurities or unwanted materials from the plastic waste stream.
“This will offer higher purity recycled plastic fractions that can be applied to more demanding products. Hence, ensuring that plastic waste is further recycled (and in some cases fully recycled). We are focusing specifically on the treatment of home-collected plastics and ghost nets from the fishing industry.”
In the next steps of the technology, Mogens said: “We will install the cameras at two plastic recycling companies, Plastix and Dansk Affaldsminimering, which we then need to adjust and adapt the source code and machine learning training algorithms accordingly. ensuring maximum performance in the new environment.”
The research was developed as part of the Re-Plast project, which was funded by the Danish Innovation Fund with DKK 22.7 million.
