Detection of K in soil using time-resolved laser-induced breakdown spectroscopy based on convolutional neural networks

Abstract

One of the technical bottlenecks of traditional laser-induced breakdown spectroscopy (LIBS) is the difficulty in quantitative detection caused by the matrix effect. To troubleshoot this problem, this paper investigated a combination of time-resolved LIBS and convolutional neural networks (CNNs) to improve K determination in soil. The time-resolved LIBS contained the information of both wavelength and time dimension. The spectra of wavelength dimension showed the characteristic emission lines of elements, and those of time dimension presented the plasma decay trend. The one-dimensional data of LIBS intensity from the emission line at 766.49 nm were extracted and correlated with the K concentration, showing a poor correlation of R 2 c =0.0967, which is caused by the matrix effect of heterogeneous soil. For the wavelength dimension, the two-dimensional data of traditional integrated LIBS were extracted and analyzed by an artificial neural network (ANN), showing R 2 v =0.6318 and the root mean square error of validation (RMSEV)=0.6234. For the time dimension, the two-dimensional data of time-decay LIBS were extracted and analyzed by ANN, showing R 2 v =0.7366 and RMSEV=0.7855. These higher determination coefficients reveal that both the non-K emission lines of wavelength dimension and the spectral decay of time dimension could assist in quantitative detection of K. However, due to limited calibration samples, the two-dimensional models presented over-fitting. The three-dimensional data of time-resolved LIBS were analyzed by CNNs, which extracted and integrated the information of both the wavelength and time dimension, showing the R 2 v =0.9968 and RMSEV=0.0785. CNN analysis of time-resolved LIBS is capable of improving the determination of K in soil.

Quantitative Analysis of Major Metals in Agricultural Biochar Using Laser-Induced Breakdown Spectroscopy with an Adaboost Artificial Neural Network Algorithm

“…The reason may have been that the concentration of Zn was “blank” in the content range of 92.3762 mg·kg −1 –313.2861 mg·kg −1 , weakening the prediction performance of the entire detection range. As shown in Table 5, the prediction abilities for K, Ca, Mg, and Cu in this work were all superior to those in the related literature [31,33,34,35,36,37], and the performance for Zn was comparable to that of Kim et al [38]. The reason may have been that the BP-Adaboost algorithm has a better adaptive ability than traditional neural networks and PLSR, avoiding the local optimum and over-fitting phenomenon.…”

Simultaneous determination of lithology and major elements in rocks using laser-induced breakdown spectroscopy (LIBS) coupled with a deep convolutional neural network

“…Learning from continuous layers to extract important features, the CNN has achieved great success in solving various computer vision problems, especially for tasks with a small training dataset, and has been applied in the spectroscopy data analysis in recent years. As a pioneering attempt in LIBS spectral analysis coupled with the CNN, Lu et al 27 improved potassium determination in soils by using a simple-structured 2D CNN, in which the core layers consisted of a single convolutional layer and a single pooling layer. Chen et al 28 developed a more deep 2D CNN model to identify different rock samples, including dolomites, granites, limestones, mudstones and shales.…”

Deep Learning Models for Data-Driven Laser Induced Breakdown Spectroscopy (LIBS) Analysis of Interstitial Oxygen Impurities in Czochralski-Si Crystals

“…Among the few significant attempts, deep belief networks have been employed to classify lead-contaminated soil samples that demonstrated the outperformance of deep learning over SVM and PLS techniques for data analysis; 38 and, in more recent years, minor metal elements in steel were determined by combining LIBS with machine learning algorithms using back-propagation neural networks (BPNN) for building multivariate calibration models. 39 Other efforts using neural network models for various spectroscopy data analyses include the use of convolutional neural networks (CNN) for predicting potassium concentration in soil with significantly improved accuracy, 40 as well as soil sensing techniques using 1D CNN on visible-near-IR spectroscopy, 41 and quantitative LIBS analyses of elemental compositions of bimetallic targets using shallow neural network models for multi-output regression. 42 In this study, we follow-up our previous investigation 23 to bridge the void between LIBS experimental data and data analytics using one-dimensional (1D) CNN to predict the concentration of interstitial oxygen dissolved in non-doped commercial grade Si wafers based on LIBS spectral data.…”