A coal quality industrial analysis method based on laser-induced breakdown spectroscopy characteristic peak selection set

Abstract

The present application relates to the technical field of coal quality industrial analysis, and provides a coal quality industrial analysis method based on laser-induced breakdown spectroscopy characteristic peak selection set, comprising the following steps: data acquisition, using laser-induced breakdown experiment to collect coal sample spectrum data; data preprocessing, sequentially completing spectrum continuous radiation removal, local maximum value peak searching, characteristic peak selection, spectrum peak value fitting, and solving electron density and plasma electron temperature; quantitative analysis, taking standard coal sample test results as a training set, and using partial least square regression method to realize quantitative calculation of coal quality industrial analysis components of the sample; the present application can realize coal quality industrial quantitative analysis based on laser-induced breakdown spectroscopy, is suitable for the characteristic peak set of coal quality industrial analysis, and the integrated application of the method can achieve fitting degrees of sample ash content, volatile matter and fixed carbon all above 0.85.

Description

Technical Field

[0001] This invention relates to the field of coal quality industrial analysis technology, and in particular to a coal quality industrial analysis method based on the selection set of characteristic peaks in laser-induced breakdown spectroscopy. Background Technology

[0002] Current coal quality analysis techniques primarily rely on traditional laboratory analysis methods, which are cumbersome and typically take several hours from sampling and sample preparation to obtaining results. The delay in coal quality data cannot be matched with the current combustion status of the boiler, failing to meet the timeliness requirements for supervision and inspection. Furthermore, traditional laboratory methods usually require an instrument that can only measure one or two items, resulting in large footprints, high labor costs, and poor portability. Therefore, they are not suitable for in-situ, remote, and rapid detection of coal entering the boiler. In addition, current rapid coal quality analysis technologies generally suffer from drawbacks such as high installation and maintenance costs and strict safety supervision, making it difficult to promote their widespread use. For example, the neutron source in instantaneous gamma neutron activation analysis technology poses potential radiation hazards, X-ray fluorescence technology cannot analyze low atomic number elements such as C and H, and inductively coupled plasma atomic emission spectrometry requires a large amount of argon gas.

[0003] In summary, the aforementioned technologies all have their limitations when applied to rapid coal quality testing. However, LIBS technology, with its significant advantages such as simple sample pretreatment, near-non-destructive testing, simultaneous multi-element measurement, and rapid measurement, has great potential in the field of rapid testing technology. In the future, based on the rational development of LIBS technology, it can be applied to in-situ and remote measurements, such as real-time monitoring of boiler thermal efficiency using LIBS quantitative technology.

[0004] Therefore, to address the above problems, a coal quality industrial analysis method based on the selection set of characteristic peaks in laser-induced breakdown spectroscopy is proposed to solve these problems. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention develops a coal quality industrial analysis method based on the selection set of characteristic peaks in laser-induced breakdown spectroscopy. The integrated application of this invention can achieve a goodness of fit of over 0.85 for the ash content, volatile matter, and fixed carbon of the test sample.

[0006] The technical solution to the problem solved by this invention is as follows: This invention provides a coal quality industrial analysis method based on the selection set of characteristic peaks in laser-induced breakdown spectroscopy, comprising the following steps:

[0007] Step 1: Data Acquisition: Acquire spectral data of coal samples using laser-induced breakdown experiments;

[0008] Step 2: Data preprocessing: Sequentially complete the removal of continuous spectral radiation, local maximum peak finding, characteristic peak selection, spectral peak fitting, and solve for electron density and plasma electron temperature;

[0009] Step 3: Quantitative analysis: Using the test results of standard coal samples as the training set, partial least squares regression is used to quantitatively calculate the industrial analytical components of the coal in the samples.

[0010] As an optimization, in step one, the laser energy for the laser-induced breakdown experiment was 90 mJ, the delay time was 1.2 μs, the focal depth was 2 mm, the sample pressing pressure was 30 MPa, and the test method involved 10 ablation cycles at the same location. The local thermal equilibrium of the plasma was verified using the following expression.

[0011]

[0012] Where, N c It is the minimum electron density required for local thermal equilibrium, T ex It is the electron temperature, and ΔE is the energy level difference of the AlⅠ emission line transition.

[0013] As an optimization, in step two, the spectral continuous radiation removal is performed using the ZhangFit method of coal quality analysis.

[0014] As an optimization, in step two, the characteristic peak selection is based on the data collected from 100 samples, the baseline-removed spectral intensity threshold is set to 2000, and the peak wavelength repetition count is set to 50.

[0015] As an optimization, in step two, Lorentz fitting is used for spectral peak fitting. The function fitting expression for the Lorentz line shape is as follows:

[0016]

[0017] Where y0 is the baseline intensity; x0 is the peak position; w L is the half-width at half-maximum of the Lorentz line; a is a constant.

[0018] As an optimization, in step two, the electron density is calculated using Stark broadening at HI 656.28nm. The Stark broadening calculation formula is as follows:

[0019]

[0020] Where, Δλ Hα It is the full width at half maximum (FWHM) of HI 656.28nm, and α is the reduced Stark linewidth.

[0021] As an optimization, in step two, the Boltzmann method is used to solve for the plasma electron temperature.

[0022] As an optimization, in step three, the partial least squares regression method sets the number of factors k for partial least squares to 12 according to the quantitative analysis objective, and calculates R based on the five-fold cross-validation results of ash content. 2 The maximum value is 0.942, the minimum RMSECV value is 1.349%, and R is used. 2 The maximum value and the minimum value of RMSECV are used for modeling. The spectral data of the prediction set are input into the existing model to obtain the predicted value of ash content, and the prediction set R is obtained. 2 The value is 0.970, and the RMSEP value is 1.012%. According to the RMSEP calculation formula and its meaning, the predicted value fluctuates by an average of 1.012% around the level of its true value.

[0023] As an optimization, in step three, the volatile matter is evaluated using five-fold cross-validation based on partial least squares regression. When k = 17, R0 2 The value is the largest, R 2 The value is 0.882, and the RMSECV value is the smallest, at 0.803%.

[0024] As an optimization, in step three, the five-fold cross-validation of the partial least squares regression with fixed carbon is performed. When k=7, R0 2 The value is largest when k=10, at which point the RMSECV value is 1.443%. However, when k=10, the RMSECV value is smallest at 1.396%. Modeling with k=7 is chosen, and the prediction set R... 2 The values ​​and RMSEP values ​​were 0.956 and 1.409%, respectively.

[0025] The effects described in the invention are merely those of the embodiments, and not all the effects of the invention. The above technical solutions have the following advantages or beneficial effects:

[0026] This invention enables quantitative analysis of coal quality in industrial applications based on laser-induced breakdown spectroscopy. Key operational parameters for sample testing were determined through extensive data analysis. It provides a set of characteristic peaks suitable for industrial coal quality analysis and identifies the corresponding partial least squares regression factors for quantitative analysis of ash, volatile matter, and fixed carbon. The integrated application of this method achieves a goodness of fit of over 0.85 for the ash, volatile matter, and fixed carbon of the tested samples. Attached Figure Description

[0027] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof.

[0028] Figure 1 This is a comparison of the LIBS spectra before and after baseline removal.

[0029] Figure 2 This is a schematic diagram showing the wavelength positions of the characteristic peaks.

[0030] Figure 3 This is a comparison chart of the results of Lenz fitting and Voigt fitting.

[0031] Figure 4 The plasma electron density frequency distribution diagram for 100 coal samples.

[0032] Figure 5 The fitting results are for the Boltzmann plot method.

[0033] Figure 6 The histogram of plasma electron temperature and frequency distribution for 100 coal samples.

[0034] Figure 7 The figure shows the results of five-fold cross-validation for partial least squares regression ash content quantification.

[0035] Figure 8 This is a graph showing the prediction results of partial least squares regression ash quantification.

[0036] Figure 9 The figure shows the results of the five-fold cross-validation of partial least squares regression for volatile matter quantification.

[0037] Figure 10 This is a graph showing the prediction results of partial least squares regression for volatile matter quantification.

[0038] Figure 11 The figure shows the results of five-fold cross-validation for partial least squares regression with a fixed amount of carbon.

[0039] Figure 12 The figure shows the prediction results for partial least squares regression fixed carbon quantification. Detailed Implementation

[0040] To clearly illustrate the technical features of this solution, the invention will be described in detail below through specific embodiments and in conjunction with the accompanying drawings. The following disclosure provides many different embodiments or examples for implementing different structures of the invention. To simplify the disclosure of the invention, components and arrangements of specific examples are described below. Furthermore, reference numerals and / or letters may be repeated in different examples. This repetition is for simplification and clarity and does not in itself indicate a relationship between the various embodiments and / or arrangements discussed. It should be noted that the components illustrated in the drawings are not necessarily drawn to scale. Descriptions of well-known components and processing techniques and processes are omitted to avoid unnecessarily limiting the invention. Terms such as "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance. In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0041] A coal quality industrial analysis method based on the selection set of characteristic peaks in laser-induced breakdown spectroscopy includes the following steps:

[0042] Step 1: Data Acquisition: Acquire spectral data of coal samples using laser-induced breakdown experiments;

[0043] Step 2: Data preprocessing: Sequentially complete the removal of continuous spectral radiation, local maximum peak finding, characteristic peak selection, spectral peak fitting, and solve for electron density and plasma electron temperature;

[0044] Step 3: Quantitative analysis: Using the test results of standard coal samples as the training set, partial least squares regression is used to quantitatively calculate the industrial analytical components of the coal in the samples.

[0045] This invention enables quantitative analysis of coal quality in industrial applications based on laser-induced breakdown spectroscopy. Key operational parameters for sample testing were determined through extensive data analysis, and a characteristic peak set suitable for industrial coal quality analysis was established. The corresponding partial least squares regression factors were determined for the quantitative analysis of ash, volatile matter, and fixed carbon. The integrated application of this method achieves a goodness of fit of over 0.85 for the ash, volatile matter, and fixed carbon values ​​of the tested samples.

[0046] In this embodiment, in step one, the laser energy used in the laser-induced breakdown experiment is 90 mJ, the delay time is 1.2 μs, the focal depth is 2 mm, and the sample pressing pressure is 30 MPa. The test method involves 10 ablation cycles at the same location. The average relative standard deviation (RSD) of the main peaks of the spectra collected after 10 ablation cycles at the same location is less than 10%, improving the repeatability of the LIBS technique. Furthermore, it passes the local thermal equilibrium verification of the plasma. The local thermal equilibrium verification of the plasma is performed using the following expression.

[0047]

[0048] Where, N c It is the minimum electron density required for local thermal equilibrium, T ex It is the electron temperature, and ΔE is the energy level difference of the AlⅠ emission line transition.

[0049] In this embodiment, step two involves using the Zhang-Fit method for coal quality analysis to remove continuous radiation from the spectrum. The baseline of the average spectrum is prone to drift, primarily due to continuous radiation (i.e., the baseline) caused by the acquisition environment and equipment. This blurs the signal, deteriorating the analysis results. Removing spectral noise also improves the signal-to-noise ratio of the spectral lines. The baseline removal effects of three basic methods (Modpoly, IModPoly, and Zhang-Fit) are compared. Figure 1 As shown, the ZhangFit method for coal quality analysis can pull the intensity of the background spectrum back to near zero to the greatest extent, and has the best effect.

[0050] In this embodiment, in step two, the characteristic peak selection is based on data collected from 100 samples. The baseline-removed spectral intensity threshold is set to 2000, and the peak wavelength repetition count is set to 50. That is, firstly, all peak wavelengths with an intensity higher than 2000 are selected, and then the peak wavelength must simultaneously satisfy the condition that the repetition count exceeds 50. Finally, 54 characteristic peaks are obtained, and the characteristic peak wavelengths are as follows: Figure 2 As shown in Table 1, the characteristic peaks used for classification and quantification are listed. This process first selects the spectral peaks of the transition based on physical meaning, filtering out background noise and continuous radiation. Secondly, it performs data cleaning based on quantitative and classification models to remove low-intensity or nearly non-intensity missing values.

[0051] Table 1: Characteristic peaks used for classification and quantification

[0052]

[0053] In this embodiment, in step two, both Lorentz fitting and Voigt fitting can achieve a fitting accuracy of over 99%, and their fitting results are almost identical. Figure 3 As shown,

[0054] The function fitting expression for the Lorentz line shape is as follows:

[0055]

[0056] Where y0 is the baseline intensity; x0 is the peak position; w L is the half-width at half-maximum of the Lorentz line; a is a constant.

[0057] The function fitting expression for the Voigt linear pattern is as follows:

[0058]

[0059] Where A is the peak area; x0 is the peak position; w G The half-width at half-height of the Gaussian line; w L The half-width of the Lorentz line.

[0060] Since the Lorentz function has only 4 initial parameters while the Voigt function has 5 initial parameters, Lorentz fitting is chosen for spectral peak fitting.

[0061] In this embodiment, step two uses the Stark broadening at HI 656.28 nm to calculate the electron density. This is because Stark broadening and shift are primarily caused by collisions between electrons and atoms or ions, and are also influenced by the micro-electrostatic field of ions in a small amount of static plasma. This paper uses the Stark broadening at HI 656.28 nm to calculate the electron density because HI spectral lines have advantages such as low signal-to-noise ratio, low interference, and long duration, which can reduce calculation errors. The Stark broadening calculation formula is as follows:

[0062]

[0063] Where, Δλ Ha The full width at half maximum (FWHM) of HⅠ656.28 nm can be obtained by Lorentz fitting, and the result of Lorentz fitting is as follows: Figure 3 As shown, α is the reduced Stark line half-width, and the electron density frequency distribution results of the selected 100 coal samples are as follows. Figure 4 As shown, it conforms to the typical LIBS electron density order of magnitude (10). 16 ).

[0064] As an optimization, in step two, the Boltzmann method is used to solve for the plasma electron temperature. The plasma electron temperature can reflect the excitation and ionization state inside the plasma. When solving for the plasma electron temperature, the assumption that the plasma is in a local thermal equilibrium state is first proposed. The atomic non-resonant lines Al (308.22nm, 309.27nm, 394.40nm and 396.15nm) are selected. These four spectral lines have high intensity, strong reproducibility, do not overlap with other spectral lines, and do not reabsorb, making them suitable for calculating the electron temperature. Detailed calculation parameters are shown in Table 2.

[0065] Table 2: Parameters for calculating plasma electron temperature

[0066] Atomic spectral lines wavelength λ <![CDATA[Transition probability A ki > <![CDATA[Upper level excitation energy E k > <![CDATA[Upper level degeneracy g k > unit nm <![CDATA[10 8 ·s -1 ]]> eV -- Al(I) 308.22 0.59 4.02 4.00 Al(I) 309.27 0.73 4.02 6.00 Al(I) 394.40 0.50 3.14 2.00 Al(I) 396.15 0.99 3.14 2.00

[0067] Data source: National Institute of Standards and Technology (NIST)

[0068] The Boltzmann plot formula is as follows:

[0069]

[0070] In the formula, I represents the intensity of the four spectral lines, h is Planck's constant, c is the speed of light, λ is the wavelength of the spectral line, N(T) and U(T) are the total particle number density and partition function of the plasma, and g k It is the degeneracy of the upper energy level, A ki It is the transition probability, E k It is the energy of the higher energy level, k B It is the Boltzmann constant, T ex It represents the electron temperature.

[0071] Through logarithmic transformation, it can be transformed into,

[0072]

[0073] Transform the equation into a linear function form (y = kx + b).

[0074]

[0075] In the formula, y and x can be obtained from Table 2. Then, the four sets of independent variables (x) and dependent variables (y) are plotted on a scatter plot and linear regression is performed to obtain the slope of the equation. The goodness of fit Ri of the regression is also calculated. 2 Greater than 0.95, the result is as follows Figure 5 As shown in the figure. The electron temperature of the plasma can be calculated from this, and the frequency distribution of the electron temperature of 100 coal samples is shown in the figure. Figure 6 As shown, it meets the typical LIBS electronic temperature order of magnitude (10). 3 -104 K).

[0076] In this embodiment, in step three, the partial least squares regression method sets the number of partial least squares factors k to 12 according to the quantitative analysis objective, and calculates based on the five-fold cross-validation results of ash content, such as... Figure 7 As shown, R 2 The maximum value is 0.942, the minimum RMSECV value is 1.349%, and R is used. 2 Modeling the maximum value and the minimum value of RMSECV, such as... Figure 8 As shown, the spectral data of the prediction set is input into the existing model to obtain the predicted value of ash content, and the prediction set R is obtained. 2 The value is 0.970, and the RMSEP value is 1.012%. According to the RMSEP calculation formula and its meaning, the predicted value fluctuates by an average of 1.012% around the level of its true value.

[0077] In this embodiment, in step three, the volatile matter is validated using a five-fold cross-validation based on partial least squares regression, and the validation results are as follows: Figure 9 As shown, when k = 17, R 2 The value is the largest, R 2 The value is 0.882, and the RMSECV value is the smallest, at 0.803%. Based on this parameter, the modeling process would be as follows: Figure 10 As shown, the R of the prediction set 2 The value was 0.934 and the RMSECV was 0.878%, indicating that although the fit of the training set was generally poor, it was a problem between the data itself and the model features. However, the model did not overfit, demonstrating its applicability.

[0078] In this embodiment, in step three, the five-fold cross-validation of partial least squares regression with fixed carbon is performed, and the validation structure is as follows: Figure 11 As shown, when k = 7, R 2 The value is largest when k=10, at which point the RMSECV value is 1.443%. However, when k=10, the RMSECV value is smallest at 1.396%. Therefore, modeling with k=7 is chosen. Figure 12 As shown, the prediction set R at this time 2 The values ​​and RMSEP values ​​were 0.956 and 1.409%, respectively. Since ash, volatile matter, and fixed carbon satisfy the physical relationship of adding up to 100% on a dry basis, and given the relatively ideal cross-validation and prediction results for ash and volatile matter, the predicted value for fixed carbon should theoretically not show significant deviation. The results indicate that the values ​​for ash, volatile matter, and fixed carbon all exhibit high spectral fit on the training, testing, and prediction sets. Therefore, the characteristic peaks selected in Table 1 are considered suitable for partial least squares quantification, meaning that the intensity of these characteristic peaks corresponds well with the ash, volatile matter, and fixed carbon content of the coal.

[0079] Although the specific embodiments of the invention have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of the invention. Based on the technical solutions of the invention, various modifications or variations that can be made by those skilled in the art without creative effort are still within the scope of protection of the invention.

Claims

1. A coal quality industrial analysis method based on the selection set of characteristic peaks in laser-induced breakdown spectroscopy, characterized by: Includes the following steps: Step 1: Data Acquisition: Acquire spectral data of coal samples using laser-induced breakdown experiments; Step 2: Data preprocessing: Sequentially complete the removal of continuous spectral radiation, local maximum peak finding, characteristic peak selection, spectral peak fitting, and solve for electron density and plasma electron temperature; In step two, the spectral continuous radiation removal is performed using the ZhangFit method for coal quality analysis; In step two, the characteristic peak selection is based on the data collected from 100 samples. The threshold of the spectral intensity after baseline removal is set to 2000, and the number of times the peak wavelength is repeated is set to 50. In step two, Lorentz fitting is used for spectral peak fitting. The function fitting expression for the Lorentz line shape is as follows: in, Baseline intensity; This is the peak position; The half-width of the Lorentz line; It is a constant; In step two, the electron density is calculated using Stark broadening at HI 656.28 nm. The formula for calculating Stark broadening is as follows: in, It is HⅠ656.28nm's full width at half maximum (FWHM). To reduce the half-width of the Stark line; Step 3: Quantitative analysis: Using the test results of standard coal samples as the training set, partial least squares regression is used to quantitatively calculate the industrial analytical components of the coal in the samples.

2. The coal quality industrial analysis method based on the selection set of characteristic peaks in laser-induced breakdown spectroscopy according to claim 1, characterized in that: In step one, the laser energy used in the laser-induced breakdown experiment was 90 mJ, and the delay time was 1.2 seconds. The focal depth was 2 mm, the sample pressing pressure was 30 MPa, and the test method involved 10 ablation cycles at the same location. The local thermal equilibrium of the plasma was verified using the following expression. in, It is the minimum electron density required for local thermal equilibrium. It is the temperature of the electrons. It is the upper energy level difference of the AlⅠ emission spectral line transition.

3. The coal quality industrial analysis method based on the selection set of characteristic peaks in laser-induced breakdown spectroscopy according to claim 1, characterized in that: In step two, the plasma electron temperature is solved using the Boltzmann method.

4. The coal quality industrial analysis method based on the selection set of characteristic peaks in laser-induced breakdown spectroscopy according to claim 1, characterized in that: In step three, the partial least squares regression method sets the number of factors k for partial least squares regression to 12 according to the quantitative analysis objective, and calculates the result based on the five-fold cross-validation of ash content. The maximum value is 0.942, the minimum RMSECV value is 1.349%, and it utilizes... The maximum value and the minimum value of RMSECV are used for modeling. The spectral data of the prediction set are input into the existing model to obtain the predicted value of ash content, and the prediction set is obtained. The value is 0.970, and the RMSEP value is 1.012%. According to the RMSEP calculation formula and its meaning, the predicted value fluctuates by an average of 1.012% around the level of its true value.

5. The coal quality industrial analysis method based on the selection set of characteristic peaks in laser-induced breakdown spectroscopy according to claim 4, characterized in that: In step three, the volatile matter is evaluated using five-fold cross-validation based on partial least squares regression. When k=17, The value is the largest. The value is 0.882, and the RMSECV value is the smallest, with an RMSECV of 0.803%.

6. The coal quality industrial analysis method based on the selection set of characteristic peaks in laser-induced breakdown spectroscopy according to claim 5, characterized in that: In step three, the five-fold cross-validation of the partial least squares regression with fixed carbon is performed when k=7. The value is largest when k=10, at which point the RMSECV value is 1.443%. However, when k=10, the RMSECV value is smallest at 1.396%. Modeling with k=7 is chosen, and the prediction set... The values ​​and RMSEP values ​​were 0.956 and 1.409%, respectively.

Images