LAUSR

The burning of incense produces toxic and harmful gases and particulate matters, posing a tremendous threat to both human health and the atmospheric environment. As a consequence, online in situ detection, classification, and traceability of burnt incense are of vital importance. In this paper, taking ambergris, musk, and Tibetan incense as examples, laser-induced breakdown spectroscopy (LIBS) is applied to the online detection of smoke and ash from the burning of three common types of incenses. It is found that metallic elements such as K, Mg, and Ca are present in the smoke. In contrast, more complex metallic elements, such as Fe, Al, Mn, Sr, etc., are present in the incense ash. By comparing the smoke and ash spectra of three different incenses, the feature spectra with large differences are selected, and the data are dimensionality reduced using the principal component analysis. Combined with error back propagation training artificial neural networks, the classification and traceability models of the smoke and ash from different incenses are performed, and the final recognition accuracies are 93.24% and 96.33%, respectively. All the results indicate that the combination of LIBS and machine learning has good application prospects for detecting and online tracing different incense smoke and ash and is also beneficial for human health and the natural environment.