Differentiation of Plastics by Combining Raman Spectroscopy and Machine Learning

We combined Raman spectroscopy with machine learning for the classification of 11 plastic samples. A confocal Raman system with an excitation wavelength of 532 nm was used to collect the Raman spectral data of plastic samples and principal component analysis was used for feature extraction. The prediction models of plastic classification based on three machine learning algorithms are compared. The results show that all three machine learning algorithms are able to classify 11 plastics well. This indicates that the combination of Raman spectroscopy and machine learning has great potential in the rapid and nondestructive classification of plastics.