A general detection method suitable for multi-granularity coal quality
By employing the spectral preprocessing method of Kubelka-Munk transform and Savitzky-Golay second-order derivative, the problem of the diffuse reflectance spectrum of coal samples being affected by physical state was solved, and high-precision quantitative analysis of coal quality with multiple particle sizes was achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- BEIJING UNIV OF CHEM TECH
- Filing Date
- 2026-03-09
- Publication Date
- 2026-06-09
AI Technical Summary
In existing technologies, the diffuse reflectance spectrum of coal samples is affected by the physical state of the samples, resulting in low accuracy of quantitative analysis.
A spectral preprocessing method combining Kubelka-Munk transform and Savitzky-Golay second-order derivative is adopted. By determining the reference wavelength and scattering correction factor, the spectral scattering effect is eliminated, a property analysis model is constructed, and the accuracy of quantitative analysis is improved.
It significantly improves the accuracy of quantitative analysis of coal with multiple particle sizes, and is applicable to coal samples of various particle sizes, especially showing high robustness and accuracy on medium particle size sample sets.
Smart Images
Figure CN122171406A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of spectral analysis and coal quality testing technology, and specifically relates to a universal testing method applicable to coal of various particle sizes. Background Technology
[0002] Near-infrared spectroscopy is widely used in the coal industry for rapid coal quality testing (such as the determination of volatile matter and ash content) due to its advantages of fast analysis speed, non-destructive and pollution-free operation.
[0003] Traditional preprocessing methods typically convert diffuse reflectance to absorbance, based on the Beer-Lambert Law. However, Beer's Law strictly applies to the transmission spectrum of homogeneous media. For coal, a dark-colored, high-absorption, and strongly scattering solid particle mixture, photons primarily undergo diffuse reflection internally. From a physical optics perspective, the Kubelka-Munk theory more accurately describes the diffuse reflection process. Furthermore, traditional absorbance-based preprocessing methods (such as Standard Normal Variable Transform (SNV) or Multivariate Scatter Correction (MSC)) are usually based on statistical assumptions and utilize the entire wavelength range for calculations, which can only eliminate scattering effects to a certain extent. The accuracy of the model's quantitative analysis is far from meeting the requirements for rapid coal quality analysis.
[0004] Therefore, a universal method for coal quality testing is needed. Summary of the Invention
[0005] The purpose of this invention is to provide a universal detection method applicable to coal of various particle sizes, aiming to solve the problem of low quantitative analysis accuracy caused by the diffuse reflectance spectrum of coal samples being severely affected by the physical state of the samples in the existing technology.
[0006] To achieve the above objectives, the present invention adopts the following technical solution: I. A universal detection method applicable to coal with multiple particle sizes S1: Obtain the coal quality sample set under the target scenario and the near-infrared diffuse reflectance spectrum and chemical analysis properties of each coal sample, wherein the coal samples in the coal quality sample set have the same particle size; S2: Determine the property analysis model and reference wavelength based on the near-infrared diffuse reflectance spectrum and laboratory analysis properties of each coal sample in the coal quality sample set; S3: Collect the near-infrared diffuse reflectance spectrum of the unknown coal sample and perform Kubelka-Munk transformation. Then, preprocess the Kubelka-Munk transformed spectrum with reference wavelength to obtain the corresponding standardized spectrum and input it into the property analysis model. The model outputs the analytical properties of the unknown coal sample.
[0007] S2 includes: S21: Perform Kubelka-Munk transform on the near-infrared diffuse reflectance spectrum of each coal sample in the coal sample set to obtain the absorption-scattering ratio function corresponding to the current coal sample; determine the initial reference wavelength; S22: Based on the current reference wavelength and absorption-scattering ratio function, preprocess the near-infrared diffuse reflectance spectrum of the current coal sample to obtain the corresponding standardized spectrum; S23: Repeat S22 to iterate through the near-infrared diffuse reflectance spectra of other coal samples in the coal sample set and obtain the standardized spectra corresponding to all coal samples. S24: Construct a property analysis model based on the standardized spectral and analytical properties of all coal samples in the coal quality sample set, and calculate the accuracy of the property analysis model; S25: Change the reference wavelength, repeat S22-S24, generate several reference wavelengths and corresponding property analysis models, select the property analysis model with the best accuracy as the final property analysis model, and use the reference wavelength corresponding to the property analysis model as the final reference wavelength.
[0008] The preprocessing includes: First, the absorption-scattering ratio function value corresponding to the current reference wavelength in the absorption-scattering ratio function for each coal sample is used as a scattering correction factor. Then, the near-infrared diffuse reflectance spectrum of each coal sample is corrected using the current scattering correction factor to obtain the corrected spectrum. Next, the Savitzky-Golay second derivative is taken from the corrected spectrum to obtain the derivative spectrum. Finally, the derivative spectrum is standardized to obtain the standardized spectrum.
[0009] The property analysis model includes partial least squares method, support vector regression, Gaussian process regression or neural network.
[0010] The target scenarios include pulverized coal boilers, flotation clean coal, laboratory standard chemical analysis, coking coal blending processes, circulating fluidized bed boiler fuel, and online monitoring of belt conveyors.
[0011] The process of correcting the near-infrared diffuse reflectance spectrum of each coal sample using the current scattering correction factor to obtain the corrected spectrum includes: The corrected spectrum is obtained by dividing the full-band Kubelka-Munk function value in the near-infrared diffuse reflectance spectrum of each coal sample by the scattering correction factor.
[0012] II. A universal testing device suitable for coal with multiple particle sizes The data input unit is used to acquire a coal quality sample set under the target scenario, as well as the near-infrared diffuse reflectance spectrum and laboratory analysis properties of each coal sample. The transformation unit is used to perform Kubelka-Munk transformation on the near-infrared diffuse reflectance spectrum to obtain the absorption-scattering ratio function corresponding to the current coal sample. Reference wavelength generation unit, used to select a reference wavelength; The spectral preprocessing unit is used to preprocess the near-infrared diffuse reflectance spectrum based on the reference wavelength to obtain a standardized spectrum; The model building unit is used to construct a property analysis model and its corresponding accuracy based on the standardized spectral and analytical properties of each coal sample in the coal quality sample set. The optimal parameter selection unit is used to select the optimal property analysis model and reference wavelength.
[0013] The spectral preprocessing unit includes: The scattering correction factor determination unit is used to determine the absorption-scattering ratio function value corresponding to each coal sample based on the reference wavelength and use it as the scattering correction factor. The spectral correction unit is used to correct the near-infrared diffuse reflectance spectrum of each coal sample using the current scattering correction factor to obtain the corrected spectrum; The spectral differentiation unit is used to perform Savitzky-Golay second-order differentiation on the corrected spectrum to obtain the derivative spectrum. The spectral normalization unit is used to normalize the derivative spectrum to obtain the normalized spectrum.
[0014] III. A computer device The computer device includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps of the general detection method applicable to coal of multiple particle sizes.
[0015] IV. A computer-readable storage medium The medium stores a computer program, which, when executed by a processor, implements the steps of a general detection method applicable to coal of various particle sizes.
[0016] The beneficial effects of this invention are as follows: This invention performs a Kubelka-Munk transform on the original diffuse reflectance of coal samples to obtain a function of apparent absorption-scattering ratio. Based on a sample set with a unified particle size standard and similar scattering wavelength ranges, a property-based quantitative analysis verification method is used to determine the scattering reference wavelength of the sample set and the scattering correction factor for each sample spectrum. The spectrum after scattering factor correction is then subjected to Savitzky-Golay second-order derivative and normalization. The above preprocessing can significantly eliminate the spectral scattering effect caused by the physical state of the sample, improve the accuracy of quantitative analysis of coal quality, and is applicable to coal samples of various particle size modalities (ultrafine powder, medium particles, coarse particles, in-situ rough surface, in-situ smooth surface) with a unified particle size standard in various application scenarios. Attached Figure Description
[0017] Figure 1 This is a flowchart of the method of the present invention.
[0018] Figure 2 The original near-infrared diffuse reflectance spectrum of the coal sample set (particle size 0.045 mm).
[0019] Figure 3 The spectrum is the result of Kubelka-Munk transformation for the coal sample set (particle size 0.045 mm).
[0020] Figure 4 The spectrum is the result of Kubelka-Munk correction only for the coal sample set (particle size 0.045 mm).
[0021] Figure 5 The spectrum is the result of Kubelka-Munk and Savitzky-Golay second-order derivative correction for the coal sample set (particle size 0.045 mm).
[0022] Figure 6 The results of quantitative analysis of fixed carbon were obtained by using the preprocessing method of this invention for spectral sets of coal samples of different particle sizes.
[0023] Figure 7 The results of quantitative analysis of volatile matter in coal sample spectral sets of different particle sizes were obtained by using the processing method of the present invention.
[0024] Figure 8 The results of quantitative analysis of ash content were obtained by using the preprocessing method of this invention on the spectral sets of coal samples of different particle sizes.
[0025] Figure 9 The results of quantitative moisture analysis of coal sample spectral sets of different particle sizes using the preprocessing method of this invention are presented. Detailed Implementation
[0026] To enable those skilled in the art to better understand the present invention and to more clearly define the scope of protection of the present invention, the present invention will be described in detail below with reference to certain specific embodiments and accompanying drawings. It should be noted that the following are only some specific embodiments of the present invention, and are merely a part of the embodiments of the present invention. The specific and direct descriptions of related structures are only for the convenience of understanding the present invention, and the specific features do not necessarily or directly limit the scope of the present invention. Conventional selections and substitutions made by those skilled in the art under the guidance of the present invention, as well as reasonable arrangements and combinations of several technical features under the guidance of the present invention, should all be considered within the scope of protection of the present invention.
[0027] like Figure 1 As shown, the universal detection method for coal of various particle sizes proposed in this invention includes the following steps: S1: Obtain the coal quality sample set (generally a representative sample set) under the target scenario (i.e. the scenario to be analyzed) and the near-infrared diffuse reflectance spectrum and analytical properties (i.e. standard properties) of each coal sample in the coal quality sample set. The coal samples in the coal quality sample set have the same particle size. Optionally, the standard properties include industrial analytical indicators (moisture, volatile matter, ash, fixed carbon), calorific value, etc.
[0028] Optionally, target scenarios include pulverized coal boilers, flotation clean coal, laboratory standard chemical analysis, coking coal blending processes, circulating fluidized bed boiler fuel, and online monitoring of belt conveyors. Coal sample particle sizes include ultrafine powder, medium particles, coarse particles, in-situ rough surfaces, and in-situ smooth surfaces. Specifically, the particle size for pulverized coal boilers and flotation clean coal is ultrafine powder (0.045mm, 0.074mm), the particle size for laboratory standard chemical analysis is micro powder (≤0.2mm), the particle size for coking coal blending processes is fine particles (≤3mm), and the particle size for circulating fluidized bed boiler fuel and online monitoring of belt conveyors is coarsely broken particles (4.75mm, 8mm). Various other scenarios, such as in-situ rough surfaces and in-situ smooth surfaces, are also included. The method of this invention is applicable to various types of coal.
[0029] For example, the coal sample has 9 particle sizes and 2 roughnesses. The 9 particle sizes are 0.045 mm, 0.074 mm, 0.1 mm, 0.21 mm, 0.5 mm, 1 mm, 2.5 mm, 4.75 mm and 8 mm. The 2 roughnesses are a rough surface with in-situ fracture (rough block) and a smooth surface with a roughness of less than 0.3 mm after polishing (smooth block), covering particle size scenarios of coal samples from ultrafine powder to coarse particles and in-situ rough and smooth states under various application scenarios. Twelve types of coal were included at each particle size: 0.3 coking coal, anthracite coal grade 1, anthracite coal grade 2, coking coal, fat coal, gas coal, gas-fat coal, lean coal, lean-thin coal, lignite coal grade 1, lignite coal grade 2, and thin coal, covering a variety of coal types for different uses. Near-infrared diffuse reflectance spectra of each coal were collected at each particle size. Fixed carbon (%), volatile matter (%), ash (%), and moisture (%) were analyzed for each coal sample on an air-dried basis, totaling four industrial analytical indicators. The purpose of this invention is to effectively correct the near-infrared spectra of coal samples with a uniform particle size standard at any particle size, thereby improving the accuracy of quantitative analysis.
[0030] S2: Determine the property analysis model and the reference wavelength for each model based on the near-infrared diffuse reflectance spectrum and analytical properties (i.e., standard properties) of each coal sample in the coal quality sample set; S2 specifically refers to: S21: Perform Kubelka-Munk transform on the near-infrared diffuse reflectance spectrum (i.e., spectral diffuse reflectance) of each coal sample in the coal sample set to obtain the absorption-scattering ratio function corresponding to the current coal sample; determine the initial reference wavelength; Figure 2 The original near-infrared diffuse reflectance spectra of 12 different coal samples with a particle size of 0.045 mm are shown. Ultrafine powder is the most mainstream combustion method in large-scale thermal power plants. In order to achieve instantaneous and complete combustion, coal must be ground to an extremely fine state. Figure 2 The results show that the ultrafine powder has a huge surface area, extremely strong diffuse reflection, and a very high spectral baseline. Figure 3 It shows the Figure 2 The spectrum of the coal sample set after Kubelka-Munk transformation shows that the absorption peaks of the sample become more obvious.
[0031] The Kubelka-Munk transformation formula is: in, It is a full-band wavelength. It is the full-band absorption coefficient. It is the full-band scattering coefficient. It is diffuse reflectance across the entire wavelength range. The value of the Kubelka-Munk function.
[0032] S22: Based on the current reference wavelength and absorption-scattering ratio function, preprocess the near-infrared diffuse reflectance spectrum of the current coal sample to obtain the corresponding standardized spectrum; Figure 4 It shows the Figure 3 The coal sample set uses the Kubelka-Munk function value at 1270 nm. The corrected spectrum shows that the baseline of the sample has been largely eliminated and the absorption peaks have become more pronounced. Figure 5 It shows the Figure 4 The spectrum of the coal sample set after Savitzky-Golay second derivative is shown in the window with a width of 9 data points, a polynomial order of 3, and a derivative order of 2. It can be seen that the sample spectrum no longer has baseline drift and the detailed features are more obvious.
[0033] S23: Repeat S22 to iterate through the near-infrared diffuse reflectance spectra of other coal samples in the coal sample set and obtain the standardized spectra corresponding to all coal samples. S24: Construct a property analysis model based on the standardized spectral and analytical properties of all coal samples in the coal quality sample set, and calculate the accuracy of the property analysis model. Optionally, a property analysis model can be constructed based on multiple analytical properties, or it can be constructed based on a single analytical property. There can be multiple property analysis models. Each property analysis model has its corresponding reference wavelength.
[0034] Optionally, the property analysis model includes partial least squares (PLS), support vector regression (SVR), Gaussian process regression (GPR), or neural network (ANN).
[0035] In this embodiment, only 12 samples are used at each granularity, a very small number. Partial Least Squares (PLS) is employed for quantitative property analysis, with two principal components. PLS models are built for all 12 samples, using the root mean square error of the training set. and coefficient of determination Two statistical indicators, namely the square of the correlation coefficient between the predicted and actual values, are used as indicators of the accuracy of quantitative analysis. in, It is the sample size. It is the first The true property value of a coal sample It is the first Predicted value for a coal sample; in, It is the average of all true values. These are fluctuations that the model cannot explain (i.e., prediction errors). It is the original fluctuation of the data itself (a manifestation of variance). The closer the model is to 1, the more accurate it is.
[0036] S25: Change the reference wavelength, repeat S22-S24, generate several reference wavelengths and corresponding property analysis models, select the property analysis model with the best accuracy as the final property analysis model, and use the reference wavelength corresponding to the property analysis model as the final reference wavelength.
[0037] In one feasible implementation, reference wavelengths are selected at certain wavelength intervals within a wavelength range. For example, for Figure 3 For the coal sample set, quantitative analysis was performed on the properties of fixed carbon, volatile matter, ash, and moisture in the wavelength range of 1050 nm to 1500 nm at 20 nm intervals. The selected reference wavelengths were 1270 nm, 1930 nm, 1350 nm, and 1370 nm. Reference wavelengths are generally wavelengths in the near-infrared spectrum where the chemical components have extremely weak absorption and the signal is mainly dominated by the scattering coefficient.
[0038] In one feasible implementation, the preprocessing includes: First, the absorption-scattering ratio function value corresponding to the current reference wavelength in the absorption-scattering ratio function for each coal sample is used as a scattering correction factor. Then, the near-infrared diffuse reflectance spectrum of each coal sample is corrected using the current scattering correction factor to obtain the corrected spectrum. Next, the Savitzky-Golay second derivative is taken from the corrected spectrum to obtain the derivative spectrum. Finally, the derivative spectrum is standardized to obtain the standardized spectrum.
[0039] Optionally, in the Savitzky-Golay second derivative, the window width is 5-15 data points, the polynomial order is 3-5, and the derivative order is 2.
[0040] The near-infrared diffuse reflectance spectrum of each coal sample is corrected using the current scattering correction factor to obtain the corrected spectrum, including: The corrected spectrum is obtained by dividing the full-band Kubelka-Munk function value in the near-infrared diffuse reflectance spectrum of each coal sample by the scattering correction factor.
[0041] The specific scattering coefficient value for each sample is determined by the Kubelka-Munk function value at the reference wavelength. Differences in scattering coefficients (physical terms) between samples are offset by division, resulting in the corrected spectrum. It is directly proportional to absorbance (chemical property), as derived below: Assumptions: Sample sets with uniform granularity have similar purely physical scattering reference bands and wavelengths; the scattering coefficients of individual samples at different wavelengths are similar. It is constant, denoted as , ; Scattering reference wavelength Generally, chemical components in the near-infrared spectrum have extremely weak absorption. The Kubelka-Munk value measured at the location is mainly affected by the scattering coefficient. control, ; Kubelka-Munk function values before correction: Kubelka-Munk function value at the reference wavelength: Corrected Kubelka-Munk function values: Therefore, the spectrum after scattering factor correction eliminates the differences in the physical state of the sample, extracts the absorption coefficient that is linearly related to the coal quality concentration, and then combines the Savitzky-Golay second derivative and standardization in the subsequent steps, which can significantly improve the linear correlation between near-infrared spectra and coal quality indicators, and improve the prediction accuracy and robustness of the quantitative analysis model. S3: Acquire the near-infrared diffuse reflectance spectrum of the unknown coal sample in the target scene and perform Kubelka-Munk transform. Determine the corresponding reference wavelength for the analytical properties to be analyzed. Preprocess the Kubelka-Munk transform spectrum using the determined reference wavelength to obtain the corresponding standardized spectrum, which is then input into the property analysis model. The model outputs the analytical properties of the unknown coal sample. The particle size of the unknown coal sample is the same as the particle size of the coal samples in the coal sample set.
[0042] Figure 6 , Figure 7 , Figure 8 and Figure 9The results of quantitative analysis of fixed carbon, volatile matter, ash, and moisture were shown after applying the preprocessing method of this invention to the spectra of coal samples of different particle sizes for spectral scattering correction. The results were also compared with the quantitative analysis results of other preprocessing methods, namely: standardization only (SS), standard normal variable transformation and standardization (SNV+SS), multivariate scattering correction and standardization (MSC+SS), and Savitzky-Golay second derivative and standardization (SGD2+SS). The preprocessing method of this invention is represented in the legend as KM+SGD2+SS, marked with square lines. It can be seen that the SS, SNV+SS, and MSC+SS preprocessing methods have large quantitative analysis errors and are far from accurate. SGD2+SS is a relatively effective preprocessing method with smaller quantitative analysis errors. After preprocessing with the scattering correction method proposed in this invention, the quantitative analysis accuracy of the four properties analyzed for all particle size modes is further improved, especially for the medium-particle size sample set. Very small, except for a few properties measured in situ for ultrafine powders and in-situ roughness measurements. Less than 0.98 (greater than 0.94), the four properties of other particle size modes Most values were around 0.99, demonstrating strong robustness. This indicates that for sample sets with uniform granularity standards, regardless of granularity size, the method proposed in this invention largely eliminates scattering interference, preserves effective chemical information, and significantly improves the accuracy of quantitative analysis models.
[0043] The present invention proposes a universal detection device suitable for coal with multiple particle sizes, comprising: The data input unit is used to acquire the coal quality sample set (generally a representative sample set) under the target scenario (i.e. the scenario to be analyzed) and the near-infrared diffuse reflectance spectrum and analytical properties (i.e. standard properties) of each coal sample in the coal quality sample set. The transformation unit is used to perform Kubelka-Munk transformation on the near-infrared diffuse reflectance spectrum to obtain the absorption-scattering ratio function corresponding to the current coal sample. Reference wavelength generation unit, used to select a reference wavelength; The spectral preprocessing unit is used to preprocess the near-infrared diffuse reflectance spectrum based on the reference wavelength to obtain a standardized spectrum; The model building unit is used to construct a property analysis model and its corresponding accuracy based on the standardized spectral and analytical properties (i.e., standard properties) of each coal sample in the coal quality sample set. The optimal parameter selection unit is used to select the optimal property analysis model and reference wavelength.
[0044] The spectral preprocessing unit includes: The scattering correction factor determination unit is used to determine the absorption-scattering ratio function value corresponding to each coal sample based on the reference wavelength and use it as the scattering correction factor. The spectral correction unit is used to correct the near-infrared diffuse reflectance spectrum of each coal sample using the current scattering correction factor to obtain the corrected spectrum; The spectral differentiation unit is used to perform Savitzky-Golay second-order differentiation on the corrected spectrum to obtain the derivative spectrum. The spectral normalization unit is used to normalize the derivative spectrum to obtain the normalized spectrum.
[0045] This invention proposes a computer device, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps of a general detection method applicable to coal of various particle sizes.
[0046] The present invention proposes a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of a general detection method applicable to coal of various particle sizes.
[0047] The present invention proposes a computer program product, comprising a computer program / instructions, which, when executed by a processor, implements the steps of a general detection method applicable to coal of various particle sizes.
Claims
1. A universal detection method applicable to coal with multiple particle sizes, characterized in that, Includes the following steps: S1: Obtain the coal quality sample set under the target scenario and the near-infrared diffuse reflectance spectrum and chemical analysis properties of each coal sample, wherein the coal samples in the coal quality sample set have the same particle size; S2: Determine the property analysis model and reference wavelength based on the near-infrared diffuse reflectance spectrum and laboratory analysis properties of each coal sample in the coal quality sample set; S3: Collect the near-infrared diffuse reflectance spectrum of the unknown coal sample and perform Kubelka-Munk transformation. Then, preprocess the Kubelka-Munk transformed spectrum with reference wavelength to obtain the corresponding standardized spectrum and input it into the property analysis model. The model outputs the analytical properties of the unknown coal sample.
2. The universal detection method for coal of various particle sizes according to claim 1, characterized in that, S2 includes: S21: Perform Kubelka-Munk transform on the near-infrared diffuse reflectance spectrum of each coal sample in the coal sample set to obtain the absorption-scattering ratio function corresponding to the current coal sample; determine the initial reference wavelength; S22: Based on the current reference wavelength and absorption-scattering ratio function, preprocess the near-infrared diffuse reflectance spectrum of the current coal sample to obtain the corresponding standardized spectrum; S23: Repeat S22 to iterate through the near-infrared diffuse reflectance spectra of other coal samples in the coal sample set and obtain the standardized spectra corresponding to all coal samples. S24: Construct a property analysis model based on the standardized spectral and analytical properties of all coal samples in the coal quality sample set, and calculate the accuracy of the property analysis model; S25: Change the reference wavelength, repeat S22-S24, generate several reference wavelengths and corresponding property analysis models, select the property analysis model with the best accuracy as the final property analysis model, and use the reference wavelength corresponding to the property analysis model as the final reference wavelength.
3. The universal detection method for coal of various particle sizes according to claim 2, characterized in that, The preprocessing includes: First, the absorption-scattering ratio function value corresponding to the current reference wavelength in the absorption-scattering ratio function for each coal sample is used as a scattering correction factor. Then, the near-infrared diffuse reflectance spectrum of each coal sample is corrected using the current scattering correction factor to obtain the corrected spectrum. Next, the Savitzky-Golay second derivative is taken from the corrected spectrum to obtain the derivative spectrum. Finally, the derivative spectrum is standardized to obtain the standardized spectrum.
4. The universal detection method for coal of various particle sizes according to claim 1, characterized in that, The property analysis model includes partial least squares method, support vector regression, Gaussian process regression or neural network.
5. The universal detection method for coal of various particle sizes according to claim 1, characterized in that, The target scenarios include pulverized coal boilers, flotation clean coal, laboratory standard chemical analysis, coking coal blending processes, circulating fluidized bed boiler fuel, and online monitoring of belt conveyors.
6. The universal detection method for coal of various particle sizes according to claim 1, characterized in that, The process of correcting the near-infrared diffuse reflectance spectrum of each coal sample using the current scattering correction factor to obtain the corrected spectrum includes: The corrected spectrum is obtained by dividing the full-band Kubelka-Munk function value in the near-infrared diffuse reflectance spectrum of each coal sample by the scattering correction factor.
7. A universal testing device suitable for coal of various particle sizes, characterized in that, include: The data input unit is used to acquire a coal quality sample set under the target scenario, as well as the near-infrared diffuse reflectance spectrum and laboratory analysis properties of each coal sample. The transformation unit is used to perform Kubelka-Munk transformation on the near-infrared diffuse reflectance spectrum to obtain the absorption-scattering ratio function corresponding to the current coal sample. Reference wavelength generation unit, used to select a reference wavelength; The spectral preprocessing unit is used to preprocess the near-infrared diffuse reflectance spectrum based on the reference wavelength to obtain a standardized spectrum; The model building unit is used to construct a property analysis model and its corresponding accuracy based on the standardized spectral and analytical properties of each coal sample in the coal quality sample set. The optimal parameter selection unit is used to select the optimal property analysis model and reference wavelength.
8. A universal testing device for coal of various particle sizes according to claim 7, characterized in that, The spectral preprocessing unit includes: The scattering correction factor determination unit is used to determine the absorption-scattering ratio function value corresponding to each coal sample based on the reference wavelength and use it as the scattering correction factor. The spectral correction unit is used to correct the near-infrared diffuse reflectance spectrum of each coal sample using the current scattering correction factor to obtain the corrected spectrum; The spectral differentiation unit is used to perform Savitzky-Golay second-order differentiation on the corrected spectrum to obtain the derivative spectrum. The spectral normalization unit is used to normalize the derivative spectrum to obtain the normalized spectrum.
9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the universal detection method for multi-grain size coal as described in any one of claims 1 to 6.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the general detection method for multi-grained coal as described in any one of claims 1 to 6.