Spectral classification analysis of recycling plastics of small household appliances based on infrared spectroscopy

Abstract

The recycling of plastics from small household appliances is of great significance in improving the environment and addressing resource shortages, and has gradually become a focus of attention in various countries. Firstly, spectra were collected from samples with different colors, oxidation levels, and flame retardants. It was found that samples with different colors and oxidation levels exhibited different reflectivity, while samples with flame retardants showed smaller absorption peaks. Subsequently, the spectrum was preprocessed and analyzed, and the results showed that the samples collected under different conditions had little effect on plastic classification. Finally, plastic spectral classification was carried out using algorithms such as support vector machine (SVM), backpropagation neural network (BP), k-nearest neighbor (k-NN), partial least squares discriminant analysis (PLS-DA), and linear discriminant analysis (LDA). Overall, the classification accuracy of each algorithm exceeds 92 %, with SVM and PLS-DA having the best classification performance, while K-NN has relatively poor classification performance. In summary, the plastic classification algorithm for small household appliance recycling based on infrared spectroscopy can meet the actual plastic classification needs of plastic recycling plant production lines.