AI Based Classification of Microplastics Libs Spectra

Abstract

The increasing accumulation of microplastics in the environment has become a major global concern due to their persistence, widespread distribution, and potential impacts on ecosystems and human health. This study investigates the classification of microplastics using Laser Induced Breakdown Spectroscopy (LIBS) combined with Artificial Intelligence (AI) techniques. Six common polymer types, including Polyethylene (PE), Polypropylene (PP), Polystyrene (PS), Polyethylene Terephthalate (PET), Polyvinyl Chloride (PVC), and Polylactic Acid (PLA), were selected for analysis. LIBS was employed to generate characteristic elemental emissions for each polymer. The acquired spectral data were preprocessed through background correction, noise filtering, wavelength calibration, and intensity normalization to improve data quality. Principal Component Analysis (PCA) was applied for dimensionality reduction and feature extraction. Three machine learning models, namely Support Vector Machine (SVM), Random Forest (RF), and Artificial Neural Network (ANN), were developed and evaluated for microplastic classification. The results demonstrated excellent classification performance, with ANN achieving the highest accuracy of 98.2%, followed by RF 96.5% and SVM 94.8%. The study revealed that LIBS spectra contain distinctive elemental fingerprints that enable accurate differentiation of polymer types. Compared with conventional techniques such as FTIR and Raman spectroscopy, the AI-assisted LIBS approach offers faster analysis, minimal sample preparation, and a higher degree of automation. The findings highlight the potential of integrating LIBS and AI as a rapid, reliable, and cost effective tool for automated microplastic identification and environmental monitoring applications.