Method for analyzing chemical components of geological mineral products by fusing laser-induced breakdown spectroscopy and Raman spectrum

Abstract

The invention relates to a laser-induced breakdown spectroscopy and Raman spectrum fused geological mineral chemical component analysis method, which comprises the following steps: selecting geological mineral samples with chemical component reference labels and classification labels, and collecting LIBS and Raman spectrum data of each sample; establishing a qualitative analysis model based on a data fusion strategy by taking the original spectral data as input and the classification label as response; segmenting the original spectral data set and the component label data corresponding to the modeling sample through classification labels to form a plurality of sub-training sets; establishing a chemical component quantitative analysis model based on different data fusion strategies for each sub-training set; combining the qualitative analysis model and a plurality of quantitative analysis models, and establishing a final chemical component quantitative analysis model fused by LIBS-Raman spectral data through a decision-level fusion strategy; in actual chemical component analysis application, LIBS and Raman spectrum data of a to-be-detected sample are collected under the same measurement condition and input into the quantitative analysis model, and chemical component content information is obtained. When the method is used for quantitative analysis of geological mineral samples with complex matrix composition, the mineral effect influence can be effectively eliminated, and more accurate element and compound composition information can be obtained.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of physical-measurement-spectral composition-scattering spectroscopy and emission spectroscopy, and particularly relates to a laser-induced breakdown spectroscopy and Raman spectroscopy fusion method for analyzing chemical composition of geological and mineral resources. BACKGROUND

[0002] Accurate and rapid chemical composition analysis of geological and mineral resources plays a crucial role in the fields of mineral resource exploration, evaluation, development and utilization, and environmental assessment. Traditional laboratory analysis methods for chemical composition of geological samples, such as X-ray fluorescence spectroscopy (XRF) and inductively coupled plasma mass spectrometry (ICP-MS), have the advantages of high analysis accuracy and low detection limit, but generally have problems such as complex sample pretreatment, long analysis period, and high cost, which makes it difficult to meet the urgent needs of rapid and in-situ analysis in the field environment and production site, as well as high-efficiency screening of large sample quantities.

[0003] In recent years, laser-induced breakdown spectroscopy (LIBS) technology, as a new emerging atomic spectroscopic analysis method, has shown great application prospects in the field of geological and mineral resources due to its advantages such as simple sample preparation, in-situ online analysis, and simultaneous detection of multiple elements. However, LIBS technology also has certain limitations: on the one hand, the analysis results are easily affected by the matrix effect, which poses a challenge to the accuracy of quantitative analysis; on the other hand, LIBS mainly provides information on the types and contents of elements, and cannot directly obtain key information such as molecular structure, crystal form, and phase composition of the material. For example, for carbon elements, LIBS cannot distinguish whether they exist in the form of graphite, diamond, or carbonate, which is exactly the core basis for the qualitative analysis and economic value evaluation of mineral resources.

[0004] Raman spectroscopy is a powerful molecular structure analysis technology that can non-destructively identify molecular bonds, functional groups, and crystal structures of a material through fingerprinting, thereby accurately determining the phase of a compound. However, Raman spectroscopy is generally difficult to accurately quantify element content, and has insufficient analysis sensitivity for certain samples with strong fluorescence effect.

[0005] Currently, there have been preliminary attempts to combine LIBS and Raman spectroscopy technology in analytical instruments, but most of them are limited to simple integration or sequential measurement at the hardware level. At the information processing level, existing technologies usually interpret and simply compare the data obtained from the two technologies independently, without achieving deep fusion and collaborative analysis of spectral information and chemical composition data, and without fully utilizing the complementary "element-structure" information of the two technologies to achieve "1+1>2" fusion analysis potential. Therefore, developing an analysis method that can deeply fuse LIBS and Raman spectroscopy information to achieve more rapid, comprehensive, and accurate identification of the chemical composition of geological and mineral resources has become a clear demand and key problem to be solved in the field of technology development. Summary of the Invention

[0006] In view of the above-mentioned shortcomings of the prior art, the purpose of this invention is to solve the problems that single LIBS or Raman spectra lack comprehensive information reflecting the elemental and compound components in geological and mineral samples, and that simple and primary dual-mode spectral fusion has limited effect and cannot achieve high-precision quantitative analysis of elemental and compound components.

[0007] The technical solution adopted by the present invention to achieve the above objectives is as follows:

[0008] A method for analyzing the chemical composition of geological and mineral resources by fusing laser-induced breakdown spectroscopy and Raman spectroscopy establishes an accurate quantitative analysis model for elemental or compound composition through the following measurement and analysis steps. This method is applicable to the analysis and measurement of geological and mineral samples with complex matrix compositions. The method includes the following steps:

[0009] In the modeling phase, step 1) collects LIBS and Raman spectral data for each sample to obtain the original spectral dataset for modeling; the samples are geological and mineral samples with chemical composition content labels and classification labels; step 2) uses the original spectral data as input and the classification labels as response to establish a qualitative analysis model based on a data fusion strategy; step 3) uses the classification labels to segment the original spectral dataset and composition content labels corresponding to the modeling samples to form multiple sub-training sets containing the original spectral and composition content label data; step 4) for each sub-training set, establishes several quantitative analysis models of chemical composition based on different data fusion strategies; step 5) combining the qualitative analysis model and multiple quantitative analysis models, and through a decision-level fusion strategy, establishes the final ideal quantitative analysis model of chemical composition based on LIBS-Raman spectral data fusion.

[0010] In practical chemical composition analysis applications, LIBS and Raman spectral data of the sample to be tested are collected and input into the ideal quantitative analysis model obtained in step 5) to obtain the chemical composition content information of the unknown sample.

[0011] The geological and mineral samples include, but are not limited to, at least one of the following: preparation directly from standard substances with content certificates, preparation from a mixture of high-purity compounds, and preparation from geological and mineral samples collected on-site with content information obtained after laboratory analysis. The components and contents are classified and labeled according to their physicochemical properties.

[0012] The LIBS and Raman spectra acquired in the modeling and actual chemical composition analysis applications are either acquired using a single LIBS-Raman combined system or separate LIBS and Raman measurement systems. All spectral data in the modeling and applications are acquired using the same measurement system and under the same measurement conditions. The detection wavelength range of the LIBS system and the detection wavenumber range of the Raman system are determined by the specific geological and mineral samples being analyzed, as well as the elements and compounds of interest.

[0013] Step 2) The classification label is an N-dimensional logical vector K, where each dimension represents the classification status of the sample under different classification rules.

[0014] Step 2) describes a qualitative analysis model based on a data fusion strategy, which fuses LIBS and Raman spectra at the data level, feature level, or decision level, and models it using supervised learning. The model expression is as follows:

[0015]

[0016] Where K Ref For N-dimensional classification label vectors, , ; Qualitative analysis models for LIBS and Raman spectral data fusion established for each classification rule A set of.

[0017] Step 3) involves dividing the original spectral data and the corresponding chemical component content label data into subsets Dataset0 according to the sample's classification label. i , Each sample's data is simultaneously distributed across multiple subsets. This involves performing N rounds of sampling with replacement from the original dataset based on the N classification methods included in the classification label, resulting in N subsets that serve as sub-training sets. .

[0018] Step 4) describes a quantitative chemical composition analysis model based on different data fusion strategies, which uses LIBS and Raman spectral data from each sub-training set (I...). i_LIBS I i_Raman ) as input, chemical component content label information C Ref In response, a quantitative characterization model for chemical composition content information is established by fusing LIBS and Raman spectral data at the data level, feature level, or decision level, using supervised learning and combining the category characteristics of samples from different sub-training sets. The model is expressed as follows:

[0019]

[0020] Where C Ref For chemical component content labeling, For the sub-training set Dataset i LIBS Spectrum I i_LIBS and Raman spectrum I i_Raman A quantitative analysis model established through data fusion.

[0021] The chemical composition quantitative analysis model based on LIBS-Raman spectral data fusion described in step 5) is established through supervised learning and is expressed as follows:

[0022]

[0023] Where C Ref For chemical ingredient labeling, For the sub-training set Dataset i LIBS Spectrum I i_LIBS and Raman spectrum I i_Raman Quantitative analysis models established through data fusion These are the weighting coefficients of the qualitative analysis result vector K.

[0024] The specific implementation process of the aforementioned practical chemical composition analysis application is as follows:

[0025] 1) Obtain the LIBS and Raman spectra measured under the same conditions as the modeled sample. and ;

[0026] 2) and Substitute into the qualitative analysis model and calculate ;

[0027] 3) Substitute into the weighting coefficient calculation model, calculate , ;

[0028] 4) , and Substitute the values ​​into each quantitative analysis model to calculate the predicted values ​​of chemical components. .

[0029] A geological and mineral chemical composition analysis system that integrates laser-induced breakdown spectroscopy and Raman spectroscopy includes: spectral acquisition equipment and host computer;

[0030] The spectral acquisition device adopts a LIBS-Raman combined system that simultaneously acquires LIBS and Raman spectra, or separate LIBS and Raman measurement systems. All spectral data in modeling and application are acquired using the same measurement system and under the same measurement conditions.

[0031] The host computer includes a control backend and a frontend interface. The frontend interface is used for human-computer interaction to collect the operating parameters and instructions of the spectral acquisition device input by the user, send the qualitative classification and quantitative identification algorithm parameters to the control backend, and visualize the results of qualitative classification and quantitative identification of samples. The control backend is equipped with a memory and a processor. The memory stores the program, LIBS and Raman spectral data collected by the spectral measurement system, chemical composition content labels and classification labels of the modeled samples, and the processor synchronously controls the acquisition device, loads the program and executes the methods described above to realize the modeling of the ideal quantitative analysis model and the identification of the chemical composition content information of unknown samples in actual chemical composition analysis applications.

[0032] The present invention has the following advantages and beneficial effects:

[0033] 1. The online chemical composition analysis method proposed in this invention achieves deep fusion and synchronous analysis of elemental composition information characterized by atomic spectroscopy and compound composition information characterized by molecular spectroscopy based on different LIBS and Raman spectral fusion strategies. It breaks through the limitations of traditional single atomic or molecular spectroscopy techniques in chemical composition analysis and is widely applicable to the detection of elemental and compound composition in various geological and mineral resources.

[0034] 2. The online chemical composition analysis method proposed in this invention can effectively eliminate the influence of differences in different physical and chemical matrices on spectral measurements and improve the accuracy of chemical composition analysis by using multi-dimensional classification modeling and decision-level weighted fusion. Attached Figure Description

[0035] Figure 1 A schematic diagram of a geological and mineral chemical composition analysis method that combines laser-induced breakdown spectroscopy and Raman spectroscopy.

[0036] Figure 2 The key chemical components and classification labels of 10 simulated iron ore samples.

[0037] Figure 3 This is a diagram illustrating the effect of an example application. Detailed Implementation

[0038] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, a detailed description of the specific implementation method of the present invention is given below using a simulated quantitative analysis process of iron ore chemical composition as an example. Many specific details are set forth in the following description to provide a thorough understanding of the present invention. However, the present invention can be implemented in many other ways different from those described herein, and those skilled in the art can make similar modifications without departing from the spirit of the invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0039] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.

[0040] Please see the appendix Figure 1 - Appendix Figure 3 This invention provides a method for quantitative analysis of iron ore chemical composition based on LIBS-Raman spectral multi-model fusion, the specific implementation steps of which are as follows:

[0041] (1-1) Modeling samples made from m simulated iron ore pellets with known contents of total iron (TFe), quartz (SiO2), hematite (Fe2O3) and magnetite (Fe3O4) are classified according to the total set, whether they contain quartz, whether they contain hematite, and whether they contain magnetite, and n-bit binary classification labels are obtained;

[0042] (1-2) All m modeling samples were sequentially placed into the Raman and LIBS measurement systems, and multiple raw spectra were acquired and averaged to obtain representative LIBS and Raman spectral data (as shown in the appendix). Figure 3 (As shown), together with the chemical composition content data, they form the original training set;

[0043] (1-3) Principal component analysis (PCA) was performed on the LIBS and Raman spectra in the original training set to obtain the principal component coefficient matrix Coeff. LIBS and Coeff Raman The first p1 principal components of the LIBS spectral data and the first p2 principal components of the Raman spectrum were extracted to form the spectral feature matrix.

[0044] (1-4) Using the spectral feature matrix Features as input and n classification labels as responses, n binary linear classifiers are trained to obtain the iron ore sample classification model;

[0045] (1-5) Based on the classification labels, n sub-training sets are established through the original training set. In each sub-training set, the PLS regression model is trained with the spectral feature matrix Features as input and the contents of TFe, SiO2, Fe2O3 and Fe3O4 as response.

[0046] (1-6) For samples with unknown chemical composition and classification: a) Obtain representative LIBS and Raman spectral data under the same measurement conditions as in (1-2); b) Analyze the principal component coefficient matrix Coeff... LIBS and Coeff Raman Extract the first p1 LIBS spectral principal components and the first p2 Raman spectral principal components to form spectral feature vectors. c) will Input the classification model and each regression model to obtain an n-digit predicted classification label and n groups of predicted contents of TFe, SiO2, Fe2O3 and Fe3O4. Based on the classification of the predicted contents of the classification label, calculate the mean of the corresponding x groups of predicted contents (x≤n) as the final chemical composition analysis result.

[0047] Table 1

[0048]

[0049] The chemical composition and classification labels of 10 simulated iron ore samples, as shown in Table 1, were modeled and predicted using the above method to analyze the contents of TFe, SiO2, Fe2O3, and Fe3O4. The classification labels were set to 4 digits (n = 4) for the entire set, whether it contains quartz, whether it contains hematite, and whether it contains magnetite. The number of LIBS and Raman principal components extracted was p1 = 10 and p2 = 5. Samples 1-3 and 5-10 of the 10 simulated samples were used as the original modeling samples, and sample 4 was used as the validation sample. With the number of principal components in PLS regression also set to 5, the comparison between the predicted value and the reference value of sample 4 is shown in Table 2.

[0050] Table 2

[0051]

[0052] The LIBS-Raman fusion quantitative analysis model established according to the method proposed in this invention has better prediction results for the contents of TFe, SiO2, Fe2O3 and Fe3O4 than the quantitative analysis model established solely based on LIBS spectroscopy, with the prediction error of each component reduced by more than 30%.

[0053] The present invention also provides another example: a geological and mineral chemical composition analysis system that integrates laser-induced breakdown spectroscopy and Raman spectroscopy, comprising: a spectral acquisition device and a host computer;

[0054] The spectral acquisition equipment uses a LIBS-Raman combined system that simultaneously acquires LIBS and Raman spectra, or separate LIBS and Raman measurement systems. All spectral data used in modeling and applications are acquired using the same measurement system and under the same measurement conditions.

[0055] The host computer includes a control backend and a frontend interface. The frontend interface is used for human-computer interaction to collect the operating parameters and instructions of the spectral acquisition device input by the user, send the qualitative classification and quantitative identification algorithm parameters to the control backend, and visually display the results of qualitative classification and quantitative identification of samples. The control backend is equipped with a memory and a processor. The memory stores the program, LIBS and Raman spectral data collected by the spectral measurement system, chemical composition content labels and classification labels of the modeled samples, and the processor synchronously controls the acquisition device, loads the program and executes the method steps described above to realize the modeling of the ideal quantitative analysis model and the identification of the chemical composition content information of unknown samples in actual chemical composition analysis applications.

[0056] The operating parameters and commands of the spectral acquisition device input by the user during human-computer interaction include:

[0057] The operating commands include: trigger commands for a LIBS-Raman combined system that simultaneously acquires LIBS and Raman spectra, or synchronous trigger commands for separate LIBS and Raman measurement systems; the operating parameters include: LIBS system laser pulse repetition frequency and energy, spectrometer delay and integration time, Raman system laser power, spectrometer integration time, etc.

[0058] The present invention has been described above by way of example with reference to the accompanying drawings. Obviously, the present invention is not limited to the above embodiments and there can be many variations. Any non-substantial improvements made using the inventive concept and technical solution of the present invention, or the direct application of the inventive concept and technical solution to other situations without modification, are all within the protection scope of the present invention.

Claims

1. A method for analyzing the chemical composition of geological and mineral resources by fusing laser-induced breakdown spectroscopy and Raman spectroscopy, characterized in that, An accurate quantitative analysis model for elemental or compound composition is established through the following measurement and analysis steps. This model is used for the analysis and measurement of geological and mineral samples with complex matrix compositions. The method includes the following steps: In the modeling phase, step 1) collects LIBS and Raman spectral data for each sample to obtain the original spectral dataset for modeling; the samples are geological and mineral samples with chemical composition content labels and classification labels; step 2) uses the original spectral data as input and the classification labels as response to establish a qualitative analysis model based on a data fusion strategy; step 3) uses the classification labels to segment the original spectral dataset and composition content labels corresponding to the modeling samples to form multiple sub-training sets containing the original spectral and composition content label data; step 4) for each sub-training set, establishes several quantitative analysis models of chemical composition based on different data fusion strategies; step 5) combining the qualitative analysis model and multiple quantitative analysis models, and through a decision-level fusion strategy, establishes the final ideal quantitative analysis model of chemical composition based on LIBS-Raman spectral data fusion. In practical chemical composition analysis applications, LIBS and Raman spectral data of the sample to be tested are collected and input into the ideal quantitative analysis model obtained in step 5) to obtain the chemical composition content information of the unknown sample.

2. The method for analyzing the chemical composition of geological and mineral resources by fusing laser-induced breakdown spectroscopy and Raman spectroscopy according to claim 1, characterized in that, The geological and mineral samples include, but are not limited to, at least one of the following: preparation directly from standard substances with content certificates, preparation from a mixture of high-purity compounds, and preparation from geological and mineral samples collected on-site with content information obtained after laboratory analysis. The components and contents are classified and labeled according to their physicochemical properties.

3. The method for analyzing the chemical composition of geological and mineral resources by fusing laser-induced breakdown spectroscopy and Raman spectroscopy according to claim 1, characterized in that, The LIBS and Raman spectra acquired in the modeling and actual chemical composition analysis applications are either acquired using a single LIBS-Raman combined system or separate LIBS and Raman measurement systems. All spectral data in the modeling and applications are acquired using the same measurement system and under the same measurement conditions. The detection wavelength range of the LIBS system and the detection wavenumber range of the Raman system are determined by the specific geological and mineral samples being analyzed, as well as the elements and compounds of interest.

4. The method for analyzing the chemical composition of geological and mineral resources by fusing laser-induced breakdown spectroscopy and Raman spectroscopy according to claim 1, characterized in that, Step 2) The classification label is an N-dimensional logical vector K, where each dimension represents the classification status of the sample under different classification rules.

5. The method for analyzing the chemical composition of geological and mineral resources by fusing laser-induced breakdown spectroscopy and Raman spectroscopy according to claim 1, characterized in that, Step 2) describes a qualitative analysis model based on a data fusion strategy, which fuses LIBS and Raman spectra at the data level, feature level, or decision level, and models it using supervised learning. The model's expression is as follows: Where K Ref For N-dimensional classification label vectors, , ; Qualitative analysis model for LIBS and Raman spectral data fusion established for each classification rule A set of.

6. The method for analyzing the chemical composition of geological and mineral resources by fusing laser-induced breakdown spectroscopy and Raman spectroscopy according to claim 1, characterized in that, Step 3) involves dividing the original spectral data and the corresponding chemical component content label data into subsets Dataset0 according to the sample's classification label. i , Each sample's data is simultaneously distributed across multiple subsets. This involves performing N rounds of sampling with replacement from the original dataset based on the N classification methods included in the classification label, resulting in N subsets that serve as training sets. .

7. The method for analyzing the chemical composition of geological and mineral resources by fusing laser-induced breakdown spectroscopy and Raman spectroscopy according to claim 1, characterized in that, Step 4) describes a quantitative chemical composition analysis model based on different data fusion strategies, which uses LIBS and Raman spectral data from each sub-training set (I...). i_LIBS I i_Raman ) as input, chemical component content label information C Ref In response, a quantitative characterization model for chemical composition content information is established by fusing LIBS and Raman spectral data at the data level, feature level, or decision level, using supervised learning and combining the category characteristics of samples from different sub-training sets. The model is expressed as follows: Where C Ref For chemical component content labeling, For the sub-training set Dataset i LIBS Spectrum I i_LIBS and Raman spectrum I i_Raman A quantitative analysis model established through data fusion.

8. The method for analyzing the chemical composition of geological and mineral resources by fusing laser-induced breakdown spectroscopy and Raman spectroscopy according to claim 1, characterized in that, The chemical composition quantitative analysis model based on LIBS-Raman spectral data fusion described in step 5) is established through supervised learning and is expressed as follows: Where C Ref For chemical ingredient labeling, For the sub-training set Dataset i LIBS Spectrum I i_LIBS and Raman spectrum I i_Raman Quantitative analysis models established through data fusion These are the weighting coefficients for the qualitative analysis result vector K.

9. The method for analyzing the chemical composition of geological and mineral resources by fusing laser-induced breakdown spectroscopy and Raman spectroscopy according to claim 1, characterized in that, The specific implementation process of the aforementioned practical chemical composition analysis application is as follows: 1) Obtain the LIBS and Raman spectra measured under the same conditions as the modeled sample. and ; 2) and Substitute into the qualitative analysis model and calculate ; 3) Substitute into the weighting coefficient calculation model, calculate , ; 4) , and Substitute the values ​​into each quantitative analysis model to calculate the predicted values ​​of chemical components. .

10. A geological and mineral chemical composition analysis system integrating laser-induced breakdown spectroscopy and Raman spectroscopy, characterized in that, include: Spectral acquisition equipment and host computer; The spectral acquisition device adopts a LIBS-Raman combined system that simultaneously acquires LIBS and Raman spectra, or separate LIBS and Raman measurement systems. All spectral data in modeling and application are acquired using the same measurement system and under the same measurement conditions. The host computer includes a control backend and a frontend interface. The frontend interface is used for human-computer interaction to collect the working parameters and instructions of the spectral acquisition device input by the user, send the qualitative classification and quantitative identification algorithm parameters to the control backend, and visually display the results of qualitative classification and quantitative identification of samples. The control backend is equipped with a memory and a processor. The memory stores the program, LIBS and Raman spectral data collected by the spectral measurement system, chemical composition content labels and classification labels of the modeled samples, and the processor synchronously controls the acquisition device, loads the program to execute the method described in any one of claims 1-9, and realizes the modeling of the ideal quantitative analysis model and the identification of the chemical composition content information of unknown samples in actual chemical composition analysis applications.

Images