Spectral preprocessing combined with feature selection improve model robustness for plastics samples classification by LIBS

Introduction: Nowadays, the widespread use of plastic products has significantly contributed towards environmental pollution caused by waste plastics. Laser-induced breakdown spectroscopy (LIBS), an emerging spectroscopic technology, shown great potential for rapid sorting and recycling However, poor robustness classification model severely limits large-scale application LIBS technology in recycling. Methods: In this research, we used spectral preprocessing combined with feature selection to improve support vector machine (SVM) four typical samples (ABS, nylon, 3240, its modified product FR-4). data were collected under different experimental conditions, then defined over time (ROT), focusing lenses (ROT&RFL), manufacturers (ROT&RDM) assess performance. The importance preprocessed spectra was evaluated using Relief-F algorithm, maximum accuracy validation set 92.6% when inputting first 19 most important features. Eventually, optimal prediction test set. Results discussion: ROT original spectrum, spectrum preprocessing, 58.4%, 79.1%, 98.47%, respectively. Similarly, ROT&RFL same methods 65.54%, 75%, 95.25%, ROT&RDM 65.5%, 67%, 93.92%, results demonstrate that can model, proposed method is feasible