Online in situ Diagnosis and Traceability of Rubber Combustion: Utilizing LIBS, Mass Spectrometry, and Machine Learning

Abstract

An online in situ detection system based on laser-induced breakdown spectroscopy (LIBS) is developed for the diagnosis and tracing of the rubber combustion process. The feasibility and accuracy of the system are verified by taking styrene–butadiene rubber, fluoro rubber, silicone rubber, chloroprene rubber, and natural rubber as samples. Metallic elements such as Ca, Mg, and Na are detected in the LIBS spectra of rubber combustion. The results show that the smoke produced contains different elements for different kinds of rubbers. Based on the self-developed single particle aerosol mass spectrometry system, the smoke produced by rubber combustion is detected by mass spectrometry. The mass spectrum data are compared and supplemented for LIBS spectral data. Then, the system, combining LIBS with principal component analysis (PCA) and backpropagation artificial neural network BP-ANN, achieves efficient detection and identification of different rubber smoke, as well as tracking of the rubber combustion process. The identification accuracy reached 94.00%. The combination of LIBS and algorithm helps to improve the efficiency of processing a large amount of spectral information data. Furthermore, the element differences among various types of rubber slabs are analyzed, thereby validating the accuracy of the diagnosis and traceability system. The aforementioned results indicate that the online in situ diagnosis and traceability of the rubber in the combustion process with LIBS is feasible, and using spectral detection for different types of rubber recycling is also promising.