A method for correlating and predicting low-rank coal infrared spectrum parameters and their calorific values

Abstract

The application discloses a kind of low rank coal infrared spectrum parameter and its correlation prediction method of calorific value, belong to chemical analysis technical field.The present application includes the following steps: (1) with low rank coal as research object, it is crushed, dried, and obtains coal sample;(2) coal sample is analyzed by Fourier-infrared spectrum, and its infrared spectrum diagram is obtained, and infrared spectrum parameter is calculated;(3) the low calorific value of air dry base of coal sample is determined;(4) based on infrared spectrum parameter, coal sample is classified, and correlation model is respectively constructed using polynomial regression method and verified.The present application takes the low rank coal of relatively abundant resources as research object, proposes the method for using polynomial regression to correlate the relationship between low rank coal infrared spectrum functional group structure parameter and coal calorific value, which is relatively simple to calculate, targeted, can quickly and efficiently estimate low rank coal low calorific value, and has good application prospect.

Description

Technical Field

[0001] This invention relates to the field of chemical analysis technology, and in particular to a method for predicting the correlation between infrared spectral parameters of low-rank coal and its calorific value. Background Technology

[0002] In the current context of scarce fossil resources, studying the combustion characteristics of various resources is of great significance for providing a foundation for the rational and efficient utilization of fuels. Calorific value is one of the main parameters for measuring fuel combustion performance, reflecting the usable energy content of the fuel. However, testing calorific value using the traditional oxygen bomb analysis method is time-consuming and labor-intensive, necessitating a rapid and accurate method for assessing fuel calorific value.

[0003] Current research on the correlation between infrared spectroscopy and coal largely focuses on systems and devices for detecting the quality of thermal coal using infrared spectroscopy. These systems utilize all spectral data to predict coal elemental analysis, industrial analysis, and calorific value. While applicable to all types of thermal coal, and predicting parameters including elemental analysis, industrial analysis, and calorific value, their accuracy in predicting specific parameters for particular coal types is relatively low.

[0004] For example, Chinese invention patent CN118379474A discloses a method and system for detecting the calorific value of coal based on single-wavelength measurement. This method includes collecting coal samples and determining the chemical groups in the coal samples; dividing the coal samples into a first data sample and a second data sample, and acquiring near-infrared spectral images of the first and second data samples respectively; establishing a near-infrared molecular spectral library based on the near-infrared spectral image of the first data sample; preprocessing the near-infrared spectral image of the second data sample based on the near-infrared spectral image of the second data sample; screening and determining relevant chemical groups related to the calorific value of coal based on the preprocessed near-infrared spectral image of the second data sample and the near-infrared molecular spectral library; and establishing a coal calorific value prediction model based on the relevant chemical groups related to the calorific value of coal and their corresponding wavelengths. However, this method only utilizes the wavelength parameters of functional groups in the infrared spectrum. Research shows that functional groups on the coal surface have a significant impact on the oxidation characteristics of coal, and ignoring the influence of parameters such as infrared spectral peak intensity and peak width will lead to a decrease in prediction accuracy.

[0005] Low-rank coal accounts for over 50% of coal resources. Due to its higher volatile matter content and lower calorific value, low-rank coal has a relatively low utilization rate, and research on its calorific value is relatively limited. Therefore, providing suitable detection methods to improve the accuracy of predicting the calorific value of low-rank coal is an urgent technical problem to be solved. Summary of the Invention

[0006] In view of the above-mentioned deficiencies of the prior art, the present invention provides a fast, efficient, and highly accurate method for predicting the correlation between low-order coal infrared spectral parameters and their calorific value, comprising the following steps:

[0007] (1) Using low-rank coal as the research object, it was crushed and dried to obtain coal samples;

[0008] (2) The coal sample was subjected to Fourier transform infrared spectroscopy analysis to obtain its infrared spectrum and the infrared spectral parameters were calculated;

[0009] (3) Determine the lower heating value of the coal sample on an air-dried basis;

[0010] (4) Based on infrared spectral parameters, coal samples are classified, and correlation models are constructed and verified using multinomial regression.

[0011] Preferably, in step (1), according to the classification of GB / T 5751-2009, low-rank coal includes lignite, long-flame coal, non-caking coal, and weakly caking coal.

[0012] Preferably, in step (2), the sample preparation for Fourier transform infrared spectroscopy analysis is performed using the pelleting method.

[0013] More preferably, the tableting method includes mixing the diluent with the coal sample and then pressing it into tablets.

[0014] The pellet method for sample preparation in Fourier transform infrared spectroscopy offers numerous advantages, including enhanced spectral quality, ease of operation, low cost, high repeatability, and wide applicability. The pellet method typically involves mixing the sample with a diluent (such as potassium bromide) at a mass ratio of 50-200:1, and then pressing the mixture under high pressure to form transparent or translucent pellets.

[0015] Preferably, in step (2), the infrared spectra are grouped according to wavenumber, and then grouped by hydroxyl group (3700-3100 cm⁻¹). -1 ), aliphatic hydrocarbons (3000-2800 cm⁻¹) -1 ), oxygen-containing functional groups (1800-1000 cm) -1 Aromatic hydrocarbons (900-700 cm) -1 It is divided into four parts.

[0016] Preferably, in step (2), the infrared spectral parameters are calculated using the following method:

[0017] Calculate the characteristic peak areas of hydroxyl groups, aliphatic hydrocarbons, oxygen-containing functional groups, and aromatic hydrocarbons, and denote them as follows: A 1. A 2. A 3. A 4; Calculate the sum of the areas of the four characteristic peaks, denoted as . A tot The relative peak area is obtained by using the ratio of the characteristic peak area to the total peak area. X i :

[0018]

[0019] In the formula, A tot Represents the total peak area. X i Represents the relative area of ​​each characteristic peak. X 1. X 2. X 3. X 4 represents the infrared spectral parameters.

[0020] Considering that the absorption peak intensity of each functional group in Fourier transform infrared spectroscopy is affected by factors such as the intensity of the detected infrared light and the thickness of the pressed material, it is difficult to directly use the peak area to represent the number of functional groups in different samples. This invention addresses this issue by considering that for the same set of data, the testing conditions are identical, allowing for the acquisition of the proportion of each peak area. After removing the dimensions, the data sets can be compared longitudinally. Therefore, this invention solves the aforementioned technical problem by using the ratio of the characteristic peak area to the total peak area to obtain the relative peak area.

[0021] Those skilled in the art can measure the lower heating value of a coal sample on an air-dried basis using common instruments or methods. As presented in the specific embodiments of the present invention, this data can be measured using a calorific value analyzer.

[0022] More preferably, in step (4), the low-rank coal is divided into categories based on the relative relationship between hydroxyl groups and oxygen-containing functional groups. X , Y Two groups:

[0023] .

[0024] In this invention, the characteristic peak areas of the aliphatic hydrocarbon band and the aromatic hydrocarbon band are negligible compared to other bands. Therefore, based on the relative relationship between the hydroxyl functional group band and the oxygen-containing functional group band, this invention classifies low-rank coal into the two groups mentioned above.

[0025] More preferably, in step (4), the highest order of the polynomial regression method is set to quadratic.

[0026] Underfitting can lead to a decrease in fitting performance and lower accuracy. For example, a binomial function may have a poor fit when linearly fitted. Overfitting, on the other hand, strictly satisfies every calculation point, and the samples used for calculation can satisfy this requirement well, but its robustness is poor, and it is only applicable to the data used for fitting. In this invention, setting the highest degree of the polynomial to quadratic can avoid underfitting or overfitting of the model, thereby improving prediction accuracy.

[0027] Furthermore, in step (4), pure quadratic polynomial regression models are established using the polynomial regression method:

[0028]

[0029] In the formula, Q The lower heating value of the coal sample is MJ / kg; a 0、 d 0 is a constant coefficient; b i , c j , e k , f l These are coefficients to be determined;

[0030] When solving for the constant coefficients and undetermined coefficients of a polynomial, the goal is to minimize the sum of squared sample errors. The optimal solution for each coefficient is obtained by using the least squares method, and the correlation model can be obtained by substituting the solutions into the model.

[0031] Furthermore, in step (4), the coal sample parameters that were not involved in solving the polynomial coefficients are substituted into the correlation model to obtain the predicted value of the lower heating value, and the root mean square error and relative error of the model are calculated to verify the accuracy of the model.

[0032] In verifying the model's accuracy, the original training data can be substituted into the model, and the error can be calculated to obtain the model's training error; the model can be used to predict new coal samples, and the error can be analyzed to obtain the model's prediction error.

[0033] Currently, most infrared spectroscopy applications for coal quality detection are industrial, involving calculations (such as convolution) of the entire infrared spectrum to obtain elemental analysis, industrial analysis, and calorific value of the coal. This process is relatively complex and yields results with significant errors. This invention focuses on relatively abundant low-rank coal and proposes a method using polynomial regression to correlate the functional group structure parameters of the infrared spectrum of low-rank coal with its calorific value. This method can quickly and efficiently predict the lower heating value of low-rank coal. The calculation is relatively simple and highly targeted.

[0034] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0035] This invention provides a method for predicting the correlation between infrared spectral parameters of low-order coal and its calorific value. It has the advantages of being fast, efficient, and having high prediction accuracy, and has good application prospects. Attached Figure Description

[0036] Figure 1 A flowchart illustrating the method for predicting the correlation between infrared spectral parameters of low-rank coal and its calorific value.

[0037] Figure 2 For 41 coal samples X Group infrared spectral parameters and calorific value;

[0038] Figure 3 For 41 coal samples Y Group infrared spectral parameters and calorific value;

[0039] Figure 4 For 41 coal samples X Fitting error of the group model;

[0040] Figure 5 For 41 coal samples X Prediction error of the group model;

[0041] Figure 6 For 41 coal samples Y Fitting error of the group model;

[0042] Figure 7 For 41 coal samples Y The prediction error of the group model. Detailed Implementation

[0043] The present invention is further illustrated below by way of embodiments, but the invention is not limited to the scope of the embodiments described herein. Experimental methods in the following embodiments that do not specify specific conditions were performed according to conventional methods and conditions, or as selected according to the product instructions.

[0044] Methods for predicting the correlation between infrared spectral parameters of low-rank coal and its calorific value, such as... Figure 1 As shown, it includes the following steps:

[0045] (1) 41 kinds of low-rank coal from different sources were selected as research objects, ground to 90 mesh or above, and then dried in an oven at 105 ℃ for 24 h to obtain coal samples;

[0046] (2) Potassium bromide and coal sample were mixed at a mass ratio of 100:1 and then pressed into sheets. The infrared spectrum was obtained by Fourier transform infrared spectroscopy analysis, and the infrared spectral parameters were calculated.

[0047] Fourier transform infrared spectroscopy was used to analyze the infrared spectrum of low-rank coal, and the spectrum was divided according to hydroxyl groups (3700-3100 cm⁻¹). -1 ), aliphatic hydrocarbons (3000-2800 cm⁻¹) -1 ), oxygen-containing functional groups (1800-1000 cm) -1 ) and aromatic hydrocarbons (900-700 cm -1 The diagram is divided into four parts; the characteristic peak areas of hydroxyl groups, aliphatic hydrocarbons, oxygen-containing functional groups, and aromatic hydrocarbons are calculated and denoted as follows: A 1. A 2. A 3. A 4; Calculate the sum of the areas of the four characteristic peaks, denoted as . A totThe relative peak area is obtained by using the ratio of the characteristic peak area to the total peak area. X i :

[0048]

[0049] In the formula, A tot Represents the total peak area. X i Represents the relative area of ​​each characteristic peak. X 1. X 2. X 3. X 4 represents the infrared spectral parameters;

[0050] (3) The lower heating value of the coal sample on an air-dried basis was determined using a calorific value analyzer;

[0051] (4) Based on oxygen-containing functional groups (1800-1000 cm) -1 ) and hydroxyl groups (3700-3100 cm) -1 The ratio of ) is divided into X Group (ratio less than 3) of 21 low-rank coals and Y There are 20 groups in total (with a ratio greater than or equal to 3).

[0052] Groups X and Y respectively construct pure quadratic polynomials relating low-order coal infrared spectral parameters to coal lower heating value:

[0053]

[0054] Where Q is in MJ / kg;

[0055] X Groups and Y Each group used 10 groups of low-rank coal, aiming to minimize the sum of squared errors of all samples. The optimal solutions for each coefficient were obtained using the least squares method, and substituted into the regression equation:

[0056] Q X =1.196279734×10 9 -1.19627977×10 9 X 1-1.19627969×10 9 X 2-1.196279703×10 9 X 3-1196279789×10 9 X 4+84.65 X 1 2 +101.75 X 22 +14.98 X 3 2 +218.28 X 4 2 ;

[0057] Q Y =1.170063642×10 9 -1.170064175×10 9 X 1-1170063608×10 9 X 2-1170063076×10 9 X 3-1.170064219×10 9 X 4+858.62 X 1 2 -4195.98 X 2 2 -553.37 X 3 2 +3257.50 X 4 2 ;

[0058] The fitting accuracy of the equation is obtained by substituting the data set involved in the multinomial regression, and the prediction accuracy of the equation is obtained by substituting the data set not involved in the multinomial regression.

[0059] The above tests included 41 types of coal samples based on... X Group, Y The classification of the groups, their infrared spectral parameters and calorific value are as follows: Figure 2 , 3 As shown.

[0060] In step (4), X The fitting error of the group model is as follows Figure 4 As shown, X The prediction accuracy of the group model is as follows Figure 5 As shown; Y The fitting error of the group model is as follows Figure 6 As shown, Y The prediction accuracy of the group model is as follows Figure 7 As shown. X , Y The root mean square error of the group model is shown in Table 1.

[0061] Table 1: X , Y Root mean square error of the group model

[0062]

[0063] The results above show that the fitting and prediction errors of both models are controlled at low levels, demonstrating high precision and accuracy in the analysis of coal samples.

[0064] In summary, this invention focuses on relatively abundant low-rank coal and proposes a method using polynomial regression to correlate the relationship between the functional group structure parameters of the infrared spectrum of low-rank coal and its calorific value. The formula established according to this invention is simpler to calculate, more targeted, and can quickly and efficiently predict the lower heating value of low-rank coal, showing promising application prospects.

[0065] The preferred embodiments of the present invention have been described in detail above. It should be understood that those skilled in the art can make numerous modifications and variations based on the concept of the present invention without creative effort. Therefore, all technical solutions that can be obtained by those skilled in the art based on the concept of the present invention through logical analysis, reasoning, or limited experimentation on the basis of existing technology should be within the scope of protection defined by the claims.

Claims

1. A method for predicting the correlation between infrared spectral parameters of low-rank coal and its calorific value, characterized in that, Includes the following steps: (1) Using low-rank coal as the research object, it was crushed and dried to obtain coal samples; (2) The coal sample was subjected to Fourier transform infrared spectroscopy analysis to obtain its infrared spectrum, and the infrared spectral parameters were calculated; the calculation method of the infrared spectral parameters is as follows: Calculate the characteristic peak areas of hydroxyl groups, aliphatic hydrocarbons, oxygen-containing functional groups, and aromatic hydrocarbons, and denote them as follows: A 1. A 2. A 3. A 4; Calculate the sum of the areas of the four characteristic peaks, denoted as . A tot The relative peak area is obtained by using the ratio of the characteristic peak area to the total peak area. X i : In the formula, A tot Represents the total peak area. X i Represents the relative area of ​​each characteristic peak. X 1. X 2. X 3. X 4 represents the infrared spectral parameters; (3) Determine the lower heating value of the coal sample on an air-dried basis; (4) Based on infrared spectral parameters, coal samples are classified, and correlation models are constructed and verified using the multinomial regression method. In step (4), based on the relative relationship between hydroxyl groups and oxygen-containing functional groups, low-rank coal is divided into... X , Y Two groups: ； Establish pure quadratic polynomial regression models using the polynomial regression method: In the formula, Q The lower heating value of the coal sample is MJ / kg; a 0、 d 0 is a constant coefficient; b i , c j , e k , f l These are coefficients to be determined; When solving for the constant coefficients and undetermined coefficients of a polynomial, the goal is to minimize the sum of squared sample errors. The optimal solution for each coefficient is obtained by using the least squares method, and the correlation model can be obtained by substituting the solutions into the model.

2. The method for predicting the correlation between infrared spectral parameters of low-rank coal and its calorific value according to claim 1, characterized in that: In step (1), according to the classification of GB / T 5751-2009, low-rank coal includes lignite, long-flame coal, non-caking coal, and weakly caking coal.

3. The method for predicting the correlation between infrared spectral parameters of low-rank coal and its calorific value according to claim 1, characterized in that: In step (2), the sample preparation for Fourier transform infrared spectroscopy analysis is performed using the pelleting method.

4. The method for predicting the correlation between infrared spectral parameters of low-rank coal and its calorific value according to claim 3, characterized in that: The tableting method involves mixing a diluent with a coal sample and then pressing it into tablets.

5. The method for predicting the correlation between infrared spectral parameters of low-rank coal and its calorific value according to claim 1, characterized in that: In step (2), the infrared spectrum is divided into four parts according to wavenumber grouping: hydroxyl groups, aliphatic hydrocarbons, oxygen-containing functional groups, and aromatic hydrocarbons.

6. The method for predicting the correlation between infrared spectral parameters of low-rank coal and its calorific value according to claim 1, characterized in that: In step (4), the coal sample parameters that were not involved in solving the polynomial coefficients are substituted into the correlation model to obtain the predicted value of the lower heating value. The root mean square error and relative error of the model are calculated to verify the accuracy of the model.

Images