Machine Learning Allows Calibration Models to Predict Trace Element Concentration in Soils with Generalized LIBS Spectra

Chen Sun,Jin Yu,Yishuai Niu,Ye Tian,Hua Li,Nicole Delepine-Gilon,Zengqi Yue,Liang Gao,Yuqing Zhang,Tianlong Zhang

doi:10.1038/s41598-019-47751-y

Abstract

Determination of trace elements in soils with laser-induced breakdown spectroscopy is significantly affected by the matrix effect, due to large variations in chemical composition and physical property of different soils. Spectroscopic data treatment with univariate models often leads to poor analytical performances. We have developed in this work a multivariate model using machine learning algorithms based on a back-propagation neural network (BPNN). Beyond the classical chemometry approach, machine learning, with tremendous progresses the last years especially for image processing, is offering an ensemble of powerful and constantly renewed algorithms and tools efficient for the different steps in the construction of a spectroscopic data treatment model, including feature selection and neural network training. Considering the matrix effect as the focus of this work, we have developed the concept of generalized spectrum, where the information about the soil matrix is explicitly included in the input vector of the model as an additional dimension. After a brief presentation of the experimental procedure and the results of regression with a univariate model, the development of the multivariate model will be described in detail together with its analytical performances, showing average relative errors of calibration (REC) and of prediction (REP) within the range of 5–6%.

Highlights

Www.nature.com/scientificreports such as spark-induced breakdown spectroscopy (SIBS), have been developed to enhance the analytical capability of light elements like carbon[10]
Multivariate regressions based on chemometry, principally partial least-squares regression (PLSR) and neuronal networks analysis (NNA), have been demonstrated being able to provide robust calibration models for soil samples, with a reduced dependence of such models on the specific soil physical and chemical properties[34,35,36,37,38,39]
We used machine learning approach to significantly improve the data processing of laser-induced breakdown spectroscopy (LIBS) spectra of soils, with a particular concern in the establishment of a soil-independent calibration model able to efficiently take into account samples from different types of soil, and in the same time to significantly reduce the influence of emission source noise

Summary

Introduction

Www.nature.com/scientificreports such as spark-induced breakdown spectroscopy (SIBS), have been developed to enhance the analytical capability of light elements like carbon[10]. Multivariate regressions based on chemometry, principally partial least-squares regression (PLSR) and neuronal networks analysis (NNA), have been demonstrated being able to provide robust calibration models for soil samples, with a reduced dependence of such models on the specific soil physical and chemical properties[34,35,36,37,38,39] These demonstrations certainly leave rooms for improvements. We used machine learning approach to significantly improve the data processing of LIBS spectra of soils, with a particular concern in the establishment of a soil-independent calibration model able to efficiently take into account samples from different types of soil, and in the same time to significantly reduce the influence of emission source noise. We emphasize the satisfactory and impressive reductions of the matrix effect and the emission source noise allowed by the developed machine learning-based multivariate calibration model, before we deliver the conclusions of the paper

Methods

Results

Conclusion