Laser induced fluorescence and machine learning: a novel approach to microplastic identification

Abstract

Identifying the types of materials such as plastics, microplastics, and oil pollutants is essential for understanding their effects on marine life. We propose a new methodology for the real-time detection and identification of microplastics in aquatic environments. Our experiments are based on a compact Laser Induced Fluorescence (LIF) device, with machine learning techniques applied to classify the materials. A 405 nm CW laser excitation source effectively induces fluorescence spectra in the visible spectrum from material samples that are either floating or submerged in water. We examine known plastic pollutants in seawater, including polyethylene (PE), polypropylene (PP), polystyrene (PS) and polyethylene terephthalate (PET), as well as maritime fuels, lubricating oils, and other organic substances that are abundant in the marine environment. Our two-step identification process first employs machine learning algorithms to distinguish microplastics from other organic materials with a high degree of accuracy (97.6%). Subsequently, the type of plastic is determined with an accuracy of 88.3% in a second application of machine learning techniques.