Using hyperspectral imaging to identify and classify large microplastic contamination in industrial composting processes

Compostable plastics are used as alternatives to conventional (non-compostable) plastics due to their ability to decompose through industrial composting comingled with food waste. However conventional (non-compostable) plastics sometimes contaminate this industrial composting process resulting in the formation of microplastics in the end compost. Therefore, it is crucial to effectively identify the types of plastics entering industrial composters to improve composting rates and enhance compost quality. In this study, we applied Hyperspectral Imaging (HSI) with various pre-processing techniques in the short-wave infrared (SWIR) region to develop an efficient model for identifying and classifying plastics and large microplastics during the industrial composting process. The materials used in the experimental analysis included compostable plastics such as PLA and PBAT, and conventional (non-compostable) plastics including PP, PET, and LDPE. Chemometric techniques, namely Partial Least Squares Discriminant Analysis (PLS-DA), was applied to develop a classification model. The Partial Least Squares Discriminant Analysis (PLS-DA) model effectively distinguished between virgin PP, PET, PBAT, PLA, and PHA plastics and soil-contaminated plastics measuring larger than 20 mm × 20 mm, achieving accuracy of 100%. Furthermore, it demonstrated a 90% accuracy rate in discriminating between pristine large microplastics and those contaminated with soil. When we tested our model on plastic samples during industrial composting we found that the accuracy of identification depended on parameters such as darkness, size, color, thickness and contamination level. Nevertheless, we achieved 85% for plastics and large microplastics detected within compost.