Quantitative analysis of sedimentary rocks using laser-induced breakdown spectroscopy: comparison of support vector regression and partial least squares regression chemometric methods

Abstract

Laser Induced Breakdown Spectroscopy (LIBS) is attracting more and more attention in geology fields for its unique adventages of on-line and in-situ analysis and the portable even handheld instruments due to the development of laser source and mini-spectrometers. However, parameters such as accuracy and precision of the instrument is still essential for field application.In this paper, two algorithm to determine the concentrations of five main elements (Si, Ca, Mg, Fe and Al) in sedimentary rock samples are proposed based on support vector regression (SVR) and partial least squares regression (PLSR). The proposed comparison demonstrates that the SVR model performed better with more satisfied accuracy and precision under the optimized conditions.For SVR quantitative analysis, the spectral features (20 lines) without principal component analysis (PCA) were selected as input variables. The optimized penalty parameter C and the key parameter of radial basis function (RBF)-σ obtained by genetic algorithmare (GA) were 4.63 and 0.9159, respectively. As well, The best number of the best principal components of PLSR was 2 optimized to be 8 by 10-fold cross-validation(CV) testing. Furthermor, the accuracy corresponding to the average relative standard deviations (RSDs) and the precision related to the root mean square error (RMSE) were calculated according to the two regression models performance. A significant enhancement of accuracy up to 43.50 times and the precision of 7.19 times for SVR model was obtained, which can eliminate the self-absorbtion of plasma efficiently compared with linear machine learning method PLSR. In conclusion, the chemometric method of SVR with better accuracy and precision can be successfully applied for quantitative analysis of complex geological samples using LIBS technique.

Analyses of Li-Rich Minerals Using Handheld LIBS Tool

“…More recently, portable LIBS tools have been widely used for the analysis of rocks, minerals, economic elements, and light elements [18,[27][28][29][30][31][32][33][34]. Although quantitative data are mainly obtained from metal studies [30,35], for more complex materials such as geological mineral mixtures or rocks, standards and calibrations need to be designed to obtain consistent data when using the LIBS technique and, thus, the handheld LIBS tools [6,17,[36][37][38][39][40][41][42][43]. The LIGHTS (Lightweight Integrated Ground and airborne Hyperspectral Topological Solutions) project aims at facilitating the exploration of new lithium (Li) resources, which mainly consist of pegmatites (http://lights.univ-lorraine.fr accessed on 1 May 2021).…”

Hackem-LIBS: An Heterogeneous Stacking Ensemble Model for Laser-Induced Breakdown Spectroscopy Elemental Quantitative Analysis

“…To demonstrate the superiority of our model against individual learners. We conduct experiments to compare Hackem-LIBS with several popular individual learners for regression learning tasks including Support Vector Regression (SVR) [14], Random Forest Regression (RFR) [12], Gradient Boosting (GBoost), Lasso Regression (Lasso) and Elastic Net (ENet) [37]. For each learner, we repeated the experiments multiple times and reported the average results.…”

“…Zhang T et al [13] combined Support Vector Regression (SVR) with Partial Least Square model to conduct quantitative and classification analysis of slag samples and proved that an integration of two models yields higher prediction accuracy than a single PLSR model. Shi Q et al [14] proved that incorporating SVR to LIBS quantitative analysis could provide a better prediction result. Yan C et al [15] improved the traditional BP-ANNs and introduced the Extreme Kernel Machine (K-ELM) to achieve better generalization ability and more accurate predictions.…”

Calibration Methods of Laser-Induced Breakdown Spectroscopy

“…Generally, the most common chemometric technique applied to concentration measurement by LIBS is PLS. It has been applied to many fields of analysis, such as soil [55,65,66], steel [67][68][69], glass [50], rock [70], iron ore [63] and coal [71,72]. The rest of the analysis methods are PCR [50,73,74], LASSO [75,76], kNN [77], ANN [78][79][80][81], SVM [82,83] and so on.…”