Plastic waste categorisation using machine learning methods-metals contaminations

conference paper

English

The global plastic consumption is consistently increasing, and plastic recycling is still a crucial issue to be solved due to the depletion of fossil resources. The polymers in the plastic can be chemically enhanced with materials such as colourants or metal fillers. This work aims to analyse the metal contamination data in the virgin plastic and plastic waste to derive a general categorisation rule for different plastic polymers (PET, PE, PP, PS). The metals contamination in plastics can be accumulated during use or waste management practices during recycling, which can be harmful for application. The metal concentrations for plastic streams are sampled from different origins: virgin plastic, household waste, and reprocessed household and industrial waste. Machine Learning methods, specifically the tree-based classification models, are used to derive a series of 'if-then' rules for classifying the plastic waste based on the sampled data. This helps the identification of the data patterns on the plastic streams and aids in deriving a categorisation rule for any plastic. The data-driven plastic categorisation could help in understanding the current waste practices and determining a proper recycling plan for local or even global plastic waste. © 2021 University of Split, FESB.

Circular Economy; Classification; Machine Learning; Plastic recycling; Plastic waste; Tree-based model

Chin, H.H., Varbanov, P.S., Klemeš, J.J.

8. 9. 2021

Institute of Electrical and Electronics Engineers Inc.

9789532901122

2021 6th International Conference on Smart and Sustainable Technologies (SpliTech)

173101

173101

13