Laser-Induced Breakdown Spectroscopy and Hyperspectral Imaging Data Fusion for improved Mineralogical Analysis of Copper Concentrates. Issue 21 (2022)

• Record Type: Journal Article
• Authors: Fuentes, Rodrigo; Luarte, Danny; Sandoval, Claudio; Myakalwar, Ashwin Kumar; Alvarez, Jonnathan; Yáñez, Jorge; Sbarbaro, Daniel
• Is Part Of: IFAC-PapersOnLine. Volume 55:Issue 21(2022)
• Page Start: 85
• Page End: 90
• Publication Date: 2022
• Subjects: Data fusion -- LIBS -- hyperspectral imaging -- copper concentrate -- mineralogical analysis -- neural networks; Automatic control -- Periodicals; 629.805
• DOI: 10.1016/j.ifacol.2022.09.248
• Languages: English
• ISSNs: 2405-8963

Abstract

Abstract: The mineralogical analyses of copper concentrates are very important not only for technical reasons, to monitor and control the smelting processes, but also for commercial purposes, since copper concentrates are sold as an intermediate product. In order to analyze copper concentrate samples in real-time it is necessary to develop fast and efficient measuring methods. This work proposes a simple strategy to combine the information provided by laser-induced breakdown spectroscopy (LIBS) and hyperspectral imaging (HSI) to quantify several mineral species compositions in pellet samples. Since both methods provide complementary information low-level and mid-level data fusion strategies are proposed to combine their measurements and improve the predictions. A data-set of LIBS and HSI obtained from 77 samples of copper concentrates is used for regression tests. A nonlinear model, artificial neural network (ANN), is proposed to approximate the relationship between the mineralogical concentration and the spectral information. In order to reduce the number of inputs of the regression model and improve its generalization capabilities variable selection is performed for LIBS and HSI. The results obtained with mid-level data fusion are encouraging and outperform the ones obtained by using solely the individual sources. Further work is underway to take into account the spatial variability of the minerals in the sample and the detection of minor elements.