Desk type coal quality analysis spectrometer and coal quality analysis method

Abstract

The present application provides a desktop coal quality analysis spectrometer and a coal quality analysis method, which comprises a shell, a sample cup, a light source assembly, a spectrum generation assembly, a temperature control assembly and a humidity control assembly, the sample cup and the light source assembly are respectively connected to two faces of the shell, the spectrum generation assembly, the temperature control assembly and the humidity control assembly are connected in the shell, the sample cup is used for placing a coal sample, the light source assembly is used for emitting light to the coal sample, the spectrum generation assembly is used for converting the reflected light of the coal sample into the spectrum of the coal sample, the temperature control assembly is used for keeping the temperature in the shell constant in a set temperature range, and the humidity control assembly is used for keeping the humidity in the shell constant in a set humidity range. The light source assembly and the spectrum generation assembly of the present application are respectively located outside and inside the shell, the heat dissipation of the light source assembly does not affect the spectrum generation assembly, and the influence of temperature and humidity on the spectrum is reduced through the temperature control assembly and the humidity control assembly.

Description

Technical Field

[0001] This invention relates to the field of coal quality analysis technology, specifically to a benchtop coal quality analysis spectrometer and coal quality analysis methods. Background Technology

[0002] Coal is widely used as fuel in various industries. The standard indicators for coal include moisture, ash, volatile matter, total sulfur, total moisture, and calorific value. Different coal qualities result in different prices and applications, so coal quality analysis is necessary before coal is distributed and used.

[0003] Current coal quality analysis techniques involve on-site sampling and manual sample preparation and testing in the laboratory, which is time-consuming and inefficient.

[0004] Laser-induced breakdown spectroscopy (LIBS) has been widely applied to the direct measurement of coal powder in granular flow. The working principle of LIBS direct measurement of granular flow involves focusing a pulsed laser beam onto the center of a freely falling granular flow, ablating the coal particles within a certain range and exciting the generation of plasma. A spectrometer then detects the spectral signal emitted by the plasma during its decay and cooling process. By analyzing the spectrum with specific wavelengths and intensities, the types and concentrations of coal powder are obtained. While LIBS direct measurement of granular flow has the advantage of not requiring sample preparation for coal powder detection, numerous studies have found that the spectral signal stability of this measurement method is poor. Because the number, size, and spatial distribution of particles near the laser focus vary randomly, the interaction between the laser and the particles is highly complex. The generated plasma not only exhibits significant differences in morphology, but its center also drifts before and after the laser focus.

[0005] Near-infrared spectroscopy, as a rapid and non-destructive testing technique, is gaining increasing attention in many fields. Two commonly used near-infrared spectrometers are: one is a grating-based spectrometer, where light emitted from the light source is absorbed by the sample, and the remaining light is diffracted by a diffraction grating in the spectrometer. After being received by the detector and processed, the spectrum is obtained. Although grating-based near-infrared spectrometers have good anti-interference capabilities, they suffer from low accuracy and low signal-to-noise ratio. Optical fibers are typically used for light propagation between the light source and the diffraction grating, and between the diffraction grating and the detector, causing light signal attenuation. The other type is a Fourier transform-based near-infrared spectrometer, where light emitted from the light source is split into two beams by an interferometer's beam splitter. One beam is transmitted to the moving mirror of the interferometer, and the other is reflected to the fixed mirror. The two beams are reflected back to the beam splitter by the fixed and moving mirrors, respectively. The moving mirror moves in a straight line at a constant speed, thus creating an optical path difference between the two beams after splitting, resulting in interference. After the interference light is combined by the beam splitter, it passes through the sample cup. After passing through the sample, the interference light containing sample information reaches the detector. Then, the signal is processed by Fourier transform to finally obtain the near-infrared absorption spectrum of transmittance or absorbance as a function of wavenumber or wavelength.

[0006] Existing near-infrared spectrometers based on Fourier transform technology have the following problems:

[0007] First, existing near-infrared spectrometers based on Fourier transform are mostly used in laboratories with relatively constant temperature and humidity to reduce the influence of temperature and humidity on the spectrum, which has high requirements for the measurement environment.

[0008] Secondly, existing near-infrared spectrometers based on Fourier transform rely solely on simple fans for cooling, failing to achieve temperature control. Consequently, the spectral measurement results of coal samples are significantly affected by temperature.

[0009] Third, the light source and the interferometer are integrated. The light emitted by the light source first passes through the interferometer before passing through the sample. The light source and the interferometer are fixed. When the sample signal intensity is too low, the light source cannot be replaced with a high-power light source. Therefore, if the light source is to be changed, the position between the lenses of the interferometer needs to be adjusted, or even replaced with lenses with other parameters.

[0010] Fourth, the optical path transformation between the light source and the interferometer, and between the interferometer and the detector, is performed using off-axis parabolic mirrors, which result in low optical conversion efficiency and large size. Summary of the Invention

[0011] To address one or more of the problems existing in the prior art, the present invention provides a system comprising a shell, a sample cup, a light source assembly, a spectrum generation assembly, a temperature control assembly, and a humidity control assembly. The sample cup and the light source assembly are respectively connected to two surfaces of the shell. The spectrum generation assembly, temperature control assembly, and humidity control assembly are connected inside the shell. The sample cup is used to hold a coal sample. The light source assembly is used to emit light onto the coal sample. The spectrum generation assembly is used to convert the reflected light from the coal sample into the spectrum of the coal sample. The temperature control assembly is used to keep the temperature inside the shell constant within a set temperature range. The humidity control assembly is used to keep the humidity inside the shell constant within a set humidity range.

[0012] According to one aspect of the present invention, the temperature control assembly includes a temperature sensor, a heat-conducting plate, a first fan, a cooling plate, a heat sink, and a second fan. The temperature sensor is used to measure the temperature inside the housing. The heat-conducting plate is used to dissipate heat generated inside the housing. The first fan is used to form a circulating air-cooling channel surrounding the housing. The cooling plate is used to cool the housing. The heat sink is used to dissipate heat from the cooling plate. The second fan is used to dissipate heat from the heat sink.

[0013] According to one aspect of the invention, a sealed box is further included, the sealed box being fixed inside the outer casing. The spectral generation assembly includes a coaxial optical path, an integrating sphere, and a spectrometer. The integrating sphere and the spectrometer are fixed inside the sealed box, and the coaxial optical path is fixed outside the sealed box. The coaxial optical path is used to focus the light emitted by the light source assembly onto the sample cup. The integrating sphere is used to diffusely scatter the reflected light from the coal sample in the sample cup before it enters the spectrometer. The spectrometer is used to generate the spectrum of the coal sample. A temperature sensor, a heat-conducting plate, and a first fan are also fixed inside the sealed box. The cooling plate, the heat sink, and the second fan are sequentially arranged outside the sealed box in a direction away from the sealed box.

[0014] According to one aspect of the invention, the humidity control assembly includes a connecting cylinder and a drying cylinder, the connecting cylinder being detachably connected to the housing, the drying cylinder being detachably connected to the connecting cylinder, and a desiccant being placed inside the drying cylinder.

[0015] According to one aspect of the present invention, the light source assembly includes a light source, a light source bracket, a bracket heat sink, a light source heat sink, and a light source fan. The light source is mounted on the light source bracket, the light source bracket is mounted on the bracket heat sink, the light source heat sink surrounds the light source and the light source bracket, and the air outlet of the light source fan faces the light source heat sink.

[0016] According to one aspect of the present invention, the spectrometer includes an interferometer, a detector, and a data processing component. The integrating sphere includes a light source inlet, a sample reflection port, and a diffuse reflection light outlet. The coaxial optical path is disposed between the light source component and the integrating sphere and is coaxial with the integrating sphere. The distances between the coaxial optical path and the light source component and between the coaxial optical path and the integrating sphere are configured such that light emitted from the light source component enters from the light source inlet and is diffusely reflected directly without passing through the integrating sphere, but instead first illuminates the sample at the sample reflection port. The relative positions of the sample reflection port and the diffuse reflection light outlet of the integrating sphere are configured such that the reflected light from the sample undergoes multiple diffuse reflections within the integrating sphere before entering the interferometer through the diffuse reflection light outlet. The interferometer is used to cause diffraction of the diffuse reflection light emitted from the integrating sphere. The detector is used to convert the diffraction into an electrical signal. The data processing component is used to convert the electrical signal into spectral data.

[0017] According to one aspect of the present invention, the coaxial optical path includes a first collimating lens and a first converging lens, wherein the first collimating lens is used to convert the light emitted by the light source assembly into a parallel beam, and the first converging lens is used to directly converge the parallel beam emitted by the first collimating lens through the light source inlet to the sample reflection port.

[0018] According to one aspect of the present invention, the spectrum generation component further includes a second collimating lens and a second converging lens, the second converging lens being disposed between the diffuse reflection light outlet of the integrating sphere and the entrance of the spectrometer, the second collimating lens being disposed between the second converging lens and the interferometer, the second converging lens being used to converge the diffuse reflection light emitted from the diffuse reflection light outlet of the integrator to the second collimating lens, and the second collimating lens being used to convert the light beam converged by the second converging lens into a parallel light beam.

[0019] According to one aspect of the invention, a window is also included, the window being disposed above the sample reflection port.

[0020] According to one aspect of the invention, a coal quality analysis component is further included, which is wired or wirelessly connected to the spectrum generation component. The coal quality analysis component is used to convert the spectral data of the spectrum generation component into a spectrum and perform coal quality analysis through the spectrum.

[0021] According to one aspect of the present invention, the coal quality analysis component includes:

[0022] The input module converts the spectral data into matrix form to obtain a spectral data matrix, wherein the spectral data is a multidimensional matrix composed of the absorbance of each spectrum of the coal sample.

[0023] The covariance matrix construction module constructs the covariance matrix of the spectral data matrix input by the input module.

[0024] The principal component analysis module performs principal component analysis on the covariance matrix constructed by the covariance matrix construction module to obtain the principal component spectral matrix composed of the principal components of each spectrum.

[0025] The coal index matrix construction module constructs a one-dimensional or multi-dimensional coal index matrix of one or more coal indices. The coal indices include one or more of the following: dry ash-free basis, dry basis, air-dried basis, and received basis.

[0026] The convolutional neural network construction module takes the principal component spectral matrix from the principal component analysis module as input and the coal index matrix constructed by the coal index matrix construction module as output to build a convolutional neural network.

[0027] The training module trains the convolutional neural network using the training set, including: a training set construction unit to construct the training set; and a network training unit that sequentially passes the training set through the input module, the covariance matrix construction module, and the principal component analysis module to obtain the principal component spectral matrix of the training set, and obtains the coal index matrix of the training set through the coal index matrix construction module. The convolutional neural network constructed by the convolutional neural network construction module is then trained using the principal component spectral matrix and the coal index matrix of the training set.

[0028] The analysis module inputs the principal component spectral matrix of the coal sample into the trained convolutional neural network to obtain the corresponding coal index matrix.

[0029] According to one aspect of the present invention, the coal quality analysis component further includes an interpolation module for interpolating the principal component spectral matrix of the principal component analysis module.

[0030] According to a second aspect of the present invention, a method for coal quality analysis using the above-described benchtop coal quality analysis spectrometer is provided, comprising:

[0031] Place the coal sample in the sample cup;

[0032] Check if the temperature inside the casing reaches the set temperature range;

[0033] Check if the humidity inside the casing reaches the set humidity range;

[0034] When the temperature inside the casing reaches the set temperature range and the humidity inside the casing reaches the set humidity range.

[0035] The light emitted by the light source component is reflected by the coal sample in the sample cup and then passes through the spectral generation component to generate the spectrum of the coal sample.

[0036] According to a second aspect of the invention, it further includes:

[0037] The spectral data is converted into matrix form to obtain a spectral data matrix, wherein the spectral data is a multidimensional matrix composed of the absorbance of each spectrum of the coal sample.

[0038] Construct the covariance matrix of the spectral data matrix;

[0039] Principal component analysis is performed on the covariance matrix to obtain the principal component spectral matrix composed of the principal components of each spectrum;

[0040] Construct a one-dimensional or multi-dimensional coal index matrix for one or more coal indices, where the coal indices include one or more of the following: dry ash-free basis, dry basis, air-dried basis, and received basis.

[0041] Using the principal component spectral matrix as input and the coal index matrix as output, a convolutional neural network is constructed:

[0042] Training a convolutional neural network using a training set includes: constructing a training set; obtaining the principal component spectral matrix of the training set; obtaining the coal index matrix of the training set; and training the convolutional neural network using the principal component spectral matrix and the coal index matrix of the training set.

[0043] The principal component spectral matrix of the coal sample is input into the trained convolutional neural network to obtain the corresponding coal index matrix.

[0044] According to a second aspect of the present invention, before the step of constructing a convolutional neural network by taking the principal component spectral matrix as input and the coal index matrix as output, the method further includes: interpolating the principal component spectral matrix.

[0045] The benchtop coal quality analysis spectrometer of this invention uses temperature control and humidity control components to keep the spectrometer's generating components in a constant temperature and humidity environment, reducing the influence of temperature and humidity on the spectrum and ensuring the stability of the results.

[0046] The temperature control component of this invention includes a first fan for circulating heat dissipation inside the housing, a heat-conducting plate for dissipating heat from inside the housing, and a cooling plate for cooling the housing. It ensures a constant temperature environment for all components inside the housing from the inside out, which is beneficial to the stability and speed of temperature control.

[0047] This invention can achieve constant temperature control of the sealed box from the inside out through a temperature control component. The sealed box is equipped with a spectrometer and an integrating sphere, which makes the temperature control of the spectrometer faster and more accurate.

[0048] The light source assembly of this invention is located on the outside of the housing. The heat emitted by the light source does not affect the spectrum generation assembly inside the housing. Furthermore, the light source is cooled by overlapping heat dissipation through the support heat sink, the light source heat sink, and the light source fan, which ensures the stability of the light source.

[0049] This invention controls humidity through a drying cylinder, which is conveniently installed, disassembled, and replaced via a connecting cylinder.

[0050] The optical path of the present invention is a light source component-sample cup-spectrum generation component, which has the following advantages compared with the prior art: the light source component and the spectrum generation component are separate, making it convenient to replace different light source components according to different sample signal intensities.

[0051] The light emitted by the light source component of this invention undergoes sequential reflection from the sample and multiple diffuse reflections within the integrating sphere before being diffracted by the spectrometer, eliminating the need for an off-axis parabolic mirror and reducing the size of the instrument. Compared to light entering the spectrometer through optical fiber, spatial light is more stable and enters the spectrometer directly through the integrating sphere, improving the signal-to-noise ratio. This also prevents situations where the light emitted by the light source component undergoes diffuse reflection from the integrating sphere before being reflected by the sample, or where the light emitted by the light source component enters the spectrometer directly without undergoing sample reflection after diffuse reflection from the integrating sphere. This eliminates the influence of stray light generated by the above situations on the spectral data and improves the signal-to-noise ratio. Attached Figure Description

[0052] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0053] Figure 1 This is an overall schematic diagram of the benchtop coal quality analysis spectrometer described in this invention;

[0054] Figure 2 This is a schematic diagram of the structural block diagram of the benchtop coal quality analysis spectrometer described in this invention;

[0055] Figure 3 This is a three-dimensional schematic diagram of an embodiment of the internal configuration of the benchtop coal quality analysis spectrometer described in this invention;

[0056] Figure 4 This is a schematic diagram of the temperature control component and humidity control component of the benchtop coal quality analysis spectrometer described in this invention;

[0057] Figure 5 This is a schematic diagram of an embodiment of the spectral generation component and the coal quality analysis component of the benchtop coal quality analysis spectrometer described in this invention;

[0058] The components include: outer shell 1, sample cup 2, light source assembly 3, light source 31, light source support 32, support heat sink 33, light source heat sink 34, light source fan 35, spectrum generation assembly 4, coaxial optical path 41, first collimating lens 411, first converging lens 412, integrating sphere 42, light source inlet 421, sample reflection port 422, diffuse reflection light outlet 423, light shield 424, spectrometer 43, interferometer 431, moving mirror 4311, beam splitter 4312, fixed mirror 4313, detector 432, data processing assembly 433, second collimating lens 44, and second converging lens 45. 5. Window plate 46. Temperature control component 5. Temperature sensor 51. Heat-conducting plate 52. First fan 53. Cooling plate 54. Heat sink 55. Second fan 56. Temperature control circuit 57. Humidity control component 6. Connecting cylinder 61. Drying cylinder 62. Sealed box 7. Coal quality analysis component 8. Input module 81. Covariance matrix construction module 82. Principal component analysis module 83. Coal index matrix construction module 84. Interpolation module 85. Convolutional neural network construction module 86. Training module 87. Training set construction unit 871. Network training unit 872. Analysis module 88. Upper bracket 9. Detailed Implementation

[0059] In the following description, only certain exemplary embodiments are briefly described. As those skilled in the art will recognize, the described embodiments can be modified in various ways without departing from the spirit or scope of the invention. Therefore, the drawings and description are considered to be exemplary in nature and not restrictive.

[0060] The following disclosure provides many different implementations or examples for carrying out different structures of the present invention. Of course, these are merely examples and are not intended to limit the invention. Preferred embodiments of the invention are described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the invention.

[0061] Figure 1 This is an overall schematic diagram of the benchtop coal quality analysis spectrometer described in this invention. Figure 2 This is a schematic diagram of the structural block diagram of the benchtop coal quality analysis spectrometer described in this invention, as shown below. Figure 1 and 2As shown, the benchtop coal quality analysis spectrometer includes a shell 1, a sample cup 2, a light source assembly 3, a spectrum generation assembly 4, a temperature control assembly 5, and a humidity control assembly 6. The sample cup 2 and the light source assembly 3 are respectively located on two sides of the shell 1. The shell 1 is a hollow cavity. The spectrum generation assembly 4, the temperature control assembly 5, and the humidity control assembly 6 are located in the hollow cavity. The sample cup 2 is used to hold a coal sample. The spectrum generation assembly 4 is used to generate the spectrum of the coal sample. The temperature control assembly 5 is used to control the temperature in the hollow cavity to be constant within a set temperature range. The humidity control assembly 6 is used to control the humidity in the hollow cavity to be constant within a set humidity range.

[0062] In this invention, the light source component 3 and the spectrum generation component 4 are located outside and inside the housing, respectively. The heat dissipation of the light source component 3 does not affect the spectrum generation component 4, and the influence of temperature and humidity on the spectrum is reduced by the temperature control component 5 and the humidity control component 6.

[0063] In one embodiment, such as Figure 3 and 4 As shown, the temperature control component 5 includes a temperature sensor 51, a heat-conducting plate 52, a first fan 53, a cooling plate 54, a heat sink 55, and a second fan 56. The temperature sensor 51 is used to measure the temperature inside the housing. The heat-conducting plate 52 is used to dissipate the heat generated inside the housing. The first fan 53 is used to form a circulating air-cooling channel around the housing. The cooling plate 54 is used to cool the housing. The heat sink 55 is used to dissipate heat from the cooling plate. The second fan 56 is used to dissipate heat from the heat sink.

[0064] Preferably, such as Figure 3 and 4 As shown, it also includes a sealed box 7, which is fixed inside the outer shell 1. The spectrum generation component 4 includes a coaxial optical path 41, an integrating sphere 42, and a spectrometer 43. The integrating sphere 42 and the spectrometer 43 are fixed inside the sealed box 7. The coaxial optical path 41 is fixed outside the sealed box 7. The coaxial optical path 41 is used to focus the light emitted by the light source component 3 onto the sample cup 2. The integrating sphere 42 is used to diffusely scatter the reflected light from the coal sample in the sample cup 2 before it enters the spectrometer 43. The spectrometer 43 is used to generate the spectrum of the coal sample. A temperature sensor 51, a heat-conducting plate 52, and a first fan 53 are also fixed inside the sealed box 7. The cooling plate 54, the heat sink 55, and the second fan 56 are arranged sequentially outside the sealed box 7 in a direction away from the sealed box 7.

[0065] The spectrometer and integrating sphere are both housed in a sealed enclosure, making the integration sphere and spectrometer more compact. At the same time, the temperature control component ensures that the temperature is constant before the light shines on the coal sample. This allows the reflected light from the coal sample to undergo multiple diffuse reflections within the integrating sphere and the temperature of the diffusely reflected light exiting the spectrometer to remain constant, thus greatly reducing the influence of temperature on the coal sample spectrum.

[0066] In one embodiment, such as Figure 3 and 4 As shown, the humidity control component 6 includes a connecting cylinder 61 and a drying cylinder 62. The connecting cylinder 61 is detachably connected to the outer shell 1, and the drying cylinder 62 is detachably connected to the connecting cylinder 61. A desiccant is placed inside the drying cylinder 62.

[0067] Preferably, the drying cylinder 62 is disposed inside the sealed box 7.

[0068] In one embodiment, the humidity control component 6 further includes a humidity sensor for measuring the humidity inside the housing 1 or the hollow cavity inside the sealed box 7.

[0069] In one embodiment, the humidity sensor and the temperature sensor 51 are integrated into one unit.

[0070] In one embodiment, the temperature control component 5 further includes a temperature control circuit 57, and the humidity component further includes a humidity control circuit. Preferably, the temperature control circuit 57 and the humidity control circuit can be integrated into one unit.

[0071] In one embodiment, such as Figure 3 and Figure 4 As shown, the light source assembly 3 includes a light source 31, a light source bracket 32, a bracket heat sink 33, a light source heat sink 34, and a light source fan 35. The light source 31 is mounted on the light source bracket 32, and the light source bracket 32 ​​is mounted on the bracket heat sink 33. The light source heat sink 34 surrounds the light source 31 and the light source bracket 32. The air outlet of the light source fan 35 faces the light source heat sink 34. The bracket heat sink 33, the light source heat sink 34, and the light source fan 35 provide all-round heat dissipation for the light source 31, maintaining the stability of the light source 31.

[0072] This invention uses a heat-conducting plate 52, a first fan 53, a cooling plate 54, a heat sink 55, and a second fan 56 to cool the sealed box 7 from the inside out. The light source 31 is cooled from all directions by the bracket heat sink 33, the light source heat sink 34, and the light source fan 35. The spectrometer 43 is located inside the sealed box 7, while the light source 31 is located outside the outer shell. The heat dissipation of the light source 31 does not affect the spectrometer 43. This invention ensures the temperature stability of the spectrometer 43 and the light source 31 from multiple dimensions, reducing the impact of temperature on the coal sample spectrum.

[0073] In one embodiment, such as Figure 3 and Figure 5 As shown, the spectrometer 43 includes an interferometer 431, a detector 432, and a data processing component 433. The integrating sphere 42 includes a light source inlet 421, a sample reflection port 422, and a diffuse reflection light outlet 423. The coaxial optical path 41 is disposed between the light source component 3 and the integrating sphere 42 and is coaxial with the integrating sphere 42. The distances between the coaxial optical path 41 and the light source component 3, and between the coaxial optical path 41 and the integrating sphere 42, are set such that the light emitted from the light source component 3 enters through the light source inlet 421 without passing through the integrating sphere 423. Instead of direct diffuse reflection, the light is first irradiated onto the sample at the sample reflection port 422. The relative positions of the sample reflection port 422 and the diffuse reflection light outlet 423 of the integrating sphere 42 are set so that the reflected light from the sample undergoes multiple diffuse reflections within the integrating sphere 42 and then enters the interferometer 431 through the diffuse reflection light outlet 423. The interferometer 431 is used to cause the diffuse reflection light emitted from the integrating sphere 42 to diffract. The detector 432 is used to convert the diffraction into an electrical signal. The data processing component 433 is used to convert the electrical signal into spectral data.

[0074] In one embodiment, the coaxial optical path 41 includes a first collimating lens 411 and a first converging lens 412. The first collimating lens 411 is used to convert the light emitted by the light source assembly 3 into a parallel beam, and the first converging lens 412 is used to directly converge the parallel beam emitted by the first collimating lens 411 through the light source inlet 421 to the sample reflection port 422.

[0075] In one embodiment, the smaller the collimating focal length of the first collimating lens 411, the smaller the distance between the first collimating lens 411 and the light source assembly 3, and the higher the light intensity of the parallel beam passing through the first collimating lens 411.

[0076] In one embodiment, the integrating sphere 42 further includes a light-blocking plate 424, which is used to increase the number of diffuse reflections of the reflected light from the sample within the integrating sphere 42;

[0077] Preferably, one end of the light-blocking plate 424 is fixed to the inner wall of the integrating sphere 42 of the diffuse reflection light outlet 423 facing the sample reflection port 422, and the other end of the light-blocking plate 424 is inclined at an acute angle relative to the inner wall of the integrating sphere 42 along the optical axis of the diffuse reflection light outlet 423.

[0078] In one embodiment, the spectrum generation component 4 further includes a second collimating lens 44 and a second converging lens 45. The second converging lens 45 is disposed between the diffuse reflection light outlet 423 of the integrating sphere 42 and the entrance of the spectrometer 43. The second collimating lens 44 is disposed between the second converging lens 45 and the interferometer 431. The second converging lens 45 is used to converge the diffuse reflection light emitted from the diffuse reflection light outlet 423 of the integrator to the second collimating lens 44. The second collimating lens 44 is used to convert the light beam converged by the second converging lens 45 into a parallel light beam.

[0079] In one embodiment, a window 46 is also included, which is disposed above the sample reflection port 422.

[0080] In one embodiment, the benchtop coal quality analysis spectrometer further includes an upper bracket 9 for mounting a sample cup 2 and a window 46. The upper bracket 9 is connected to a sealed box 7, which is connected to the outer casing 1 via one or more support frames.

[0081] In one embodiment, such as Figure 3 and Figure 5 As shown, the interferometer 431 of the present invention consists of a moving mirror 4311, a beam splitter 4312, and a fixed mirror 4313. The beam splitter 4312 is used to split a parallel beam into two beams. The two beams are reflected by the fixed mirror 4313 and the moving mirror 4311 and return to the beam splitter to meet again, forming an interference beam. The interference beam converges to the detector 432. The movement of the moving mirror 4311 along the incident light direction changes the optical path difference between the two reflected beams with different paths, generating a time-series interference signal. The detector 432 receives the interference signal and converts it into an electrical signal for output.

[0082] In one embodiment, such as Figure 3 As shown, it also includes a coal quality analysis component 8, which is connected to the spectrum generation component 4 via wired or wireless means. The coal quality analysis component 8 is used to convert the spectral data of the spectrum generation component 4 into a spectrum, and to perform coal quality analysis through the spectrum.

[0083] The coal quality analysis component 8 mentioned above can be a terminal device such as a computer, or other electronic devices including a processor and memory, or a computer-readable storage medium storing computer programs.

[0084] In one embodiment, the coal quality analysis component 8 includes:

[0085] Input module 81 converts spectral data into matrix form to obtain a spectral data matrix, wherein the spectral data is a multidimensional matrix composed of each absorbance of each spectrum of the coal sample;

[0086] Covariance matrix construction module 82 constructs the covariance matrix of the spectral data matrix input by input module 81;

[0087] The principal component analysis module 83 performs principal component analysis on the covariance matrix constructed by the covariance matrix construction module 82 to obtain the principal component spectral matrix composed of the principal components of each spectrum.

[0088] The coal index matrix construction module 84 constructs one-dimensional or multi-dimensional coal index matrices for one or more coal indices. Coal indices include dry ash-free basis, dry basis, air-dried basis, and as-received basis, etc. Different bases are used for different applications. The as-received basis of coal is a classification based on all components of the received coal, such as as-received calorific value, air-dried basis calorific value or as-received volatile matter, dry ash-free basis calorific value, etc.

[0089] The convolutional neural network construction module 86 takes the principal component spectral matrix of the principal component analysis module 83 as input and the coal index matrix constructed by the coal index matrix construction module 84 as output to construct a convolutional neural network.

[0090] Training module 87 trains a convolutional neural network using a training set, including: training set construction unit 871, which constructs a training set; network training unit 872, which sequentially passes the training set through input module 81, covariance matrix construction module 82, and principal component analysis module 83 to obtain the principal component spectral matrix of the training set, and through coal index matrix construction module 84 to obtain the coal index matrix of the training set, and trains the convolutional neural network constructed by convolutional neural network construction module 86 using the principal component spectral matrix and coal index matrix of the training set;

[0091] Analysis module 88 inputs the principal component spectral matrix of the coal sample into the trained convolutional neural network to obtain the corresponding coal index matrix.

[0092] In one embodiment, the coal quality analysis component 8 further includes an interpolation module 85 for interpolating the principal component spectral matrix of the principal component analysis module 83. More preferably, Hermite interpolation is used to perform multiple interpolations in the principal component spectral matrix.

[0093] In one embodiment, the convolutional neural network construction module 86 uses the MobileNetV2 model to construct the convolutional neural network. The network input feature of the MobileNetV2 model is a single-channel feature with a height of 1 and a width of 3. It is processed using two-dimensional convolution. All bottleneck layers are inverted residual bottleneck layers (PW+DW+PW). The convolution kernel K is a non-PW convolution kernel on each layer. To adapt to the special characteristics of the height and width of the near-infrared spectrum, the size of the convolution kernel is set to 1X3.

[0094] The multiplication factor t is only used in the inverse residual bottleneck layer, setting the middle part of the layer to be t times the number of output channels C. When the number of repetitions n≠1, the inverse residual bottleneck layer execution ends, and then the inverse residual bottleneck layer is executed (n-1) times. The step size S of each repetition is 1, which means moving S points each time. The padding is always (0, 1), that is, 0 in the height direction and 1 in the width direction.

[0095] The Global Average Pooling (GAP) layer is used to obtain the output features corresponding to the processed principal component spectral matrix. After passing through GAP, the reshape function is used for transformation, and linear regression is performed to obtain the property values ​​of the corresponding sample.

[0096] This invention uses the covariance matrix to reflect the degree of linear correlation between any two spectral data. By using the principal component analysis module 83 to perform feature dimensionality reduction on the covariance matrix, not only can useless noise be removed, but also the amount of computation can be reduced. Furthermore, by combining convolutional neural networks, a lightweight, fast-running quantitative model that can be run on embedded devices is established, providing a new approach for the generalization of near-infrared spectral data and the establishment of quantitative models.

[0097] The present invention also provides a method for coal quality analysis using the above-mentioned benchtop coal quality analysis spectrometer, comprising:

[0098] Place the coal sample in sample cup 2;

[0099] Detect whether the temperature inside the outer casing reaches the set temperature range; preferably, detect whether the temperature inside the sealed box reaches the set temperature range.

[0100] The humidity inside the casing is checked to see if it reaches the set humidity range. Preferably, the humidity inside the casing is checked to see if it reaches the set humidity range.

[0101] When the temperature of the spectrometer 43 reaches the set temperature range and the humidity of the spectrometer 43 reaches the set humidity range,

[0102] The light emitted by the light source component 3 is reflected by the coal sample in the sample cup 2 and then passes through the spectral generation component to generate the spectrum of the coal sample.

[0103] In one embodiment, it also includes:

[0104] The spectral data is converted into matrix form to obtain a spectral data matrix, wherein the spectral data is a multidimensional matrix composed of the absorbance of each spectrum of the coal sample.

[0105] Construct the covariance matrix of the spectral data matrix;

[0106] Principal component analysis is performed on the covariance matrix to obtain the principal component spectral matrix composed of the principal components of each spectrum;

[0107] Construct a one-dimensional or multi-dimensional coal index matrix for one or more coal indices, where the coal indices include one or more of the following: dry ash-free basis, dry basis, air-dried basis, and received basis.

[0108] Using the principal component spectral matrix as input and the coal index matrix as output, a convolutional neural network is constructed:

[0109] Training a convolutional neural network using a training set includes: constructing a training set; obtaining the principal component spectral matrix of the training set; obtaining the coal index matrix of the training set; and training the convolutional neural network using the principal component spectral matrix and the coal index matrix of the training set.

[0110] The principal component spectral matrix of the coal sample is input into the trained convolutional neural network to obtain the corresponding coal index matrix.

[0111] In one embodiment, before the step of constructing a convolutional neural network by taking the principal component spectral matrix as input and the coal index matrix as output, the method further includes: interpolating the principal component spectral matrix.

[0112] In a specific embodiment of the present invention, Example 1: Coal samples were analyzed using the benchtop coal quality analysis spectrometer of the present invention; Comparative Example 1: Coal quality analysis was performed using a prior art grating-based near-infrared spectrometer and the coal quality analysis component 8 of the present invention; Comparative Example 2: Coal quality analysis was performed using a prior art Fourier transform-based near-infrared spectrometer and the coal quality analysis component 8 of the present invention; Comparative Example 3: Coal samples were analyzed using the benchtop coal quality analysis spectrometer of the present invention (excluding the coal quality analysis component 8) and the partial least squares (PLS) method; Comparative Example 4: Coal quality analysis was performed using a prior art grating-based near-infrared spectrometer and the partial least squares (PLS) method; Comparative Example 5: Coal quality analysis was performed using a prior art Fourier transform-based near-infrared spectrometer and the partial least squares (PLS) method; The received basis calorific value data obtained from coal quality analysis of the same coal samples in Examples 1-Comparative Examples 5 are shown in the following table:

[0113] Table 1

[0114]

[0115] As can be seen from the table above, the absolute value of the deviation in coal quality analysis of this invention is the smallest, and the coal quality analysis results are the most accurate.

[0116] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A benchtop coal quality analysis spectrometer, characterized in that, The device includes a shell, a sample cup, a light source assembly, a spectrum generation assembly, a temperature control assembly, and a humidity control assembly. The sample cup and the light source assembly are respectively connected to two sides of the shell. The spectrum generation assembly, temperature control assembly, and humidity control assembly are connected inside the shell. The sample cup is used to hold a coal sample. The light source assembly is used to emit light to the coal sample. The spectrum generation assembly is used to convert the reflected light from the coal sample into the spectrum of the coal sample. The temperature control assembly is used to keep the temperature inside the shell constant within a set temperature range. The humidity control assembly is used to keep the humidity inside the shell constant within a set humidity range. The temperature control component includes a temperature sensor, a heat-conducting plate, a first fan, a cooling plate, a heat sink, and a second fan. The temperature sensor is used to measure the temperature inside the housing. The heat-conducting plate is used to dissipate the heat generated inside the housing. The first fan is used to form a circulating air-cooling channel around the housing. The cooling plate is used to cool the housing. The heat sink is used to dissipate heat from the cooling plate. The second fan is used to dissipate heat from the heat sink. The benchtop coal quality analysis spectrometer further includes a sealed enclosure fixed inside the outer shell. The spectral generation component includes a coaxial optical path, an integrating sphere, and a spectrometer. The integrating sphere and the spectrometer are fixed inside the sealed enclosure, while the coaxial optical path is fixed outside. The coaxial optical path is used to focus the light emitted by the light source component onto the sample cup. The integrating sphere is used to diffusely scatter the reflected light from the coal sample in the sample cup before it enters the spectrometer. The spectrometer is used to generate the spectrum of the coal sample. A temperature sensor, a heat-conducting plate, and a first fan are also fixed inside the sealed enclosure. The cooling plate, heat sink, and second fan are arranged sequentially outside the sealed enclosure in a direction away from it. The spectrometer includes an interferometer, a detector, and a data processing component. The integrating sphere includes a light source inlet, a sample reflection port, and a diffuse reflection light outlet. The coaxial optical path is disposed between the light source component and the integrating sphere and is coaxial with the integrating sphere. The distances between the coaxial optical path and the light source component, and between the coaxial optical path and the integrating sphere, are configured such that the light emitted from the light source component enters from the light source inlet and is diffusely reflected directly without passing through the integrating sphere, but instead first illuminates the sample at the sample reflection port. The relative positions of the sample reflection port and the diffuse reflection light outlet of the integrating sphere are configured such that the reflected light from the sample undergoes multiple diffuse reflections within the integrating sphere before entering the interferometer through the diffuse reflection light outlet. The interferometer is used to cause diffraction of the diffuse reflection light emitted from the integrating sphere. The detector is used to convert the diffraction into an electrical signal. The data processing component is used to convert the electrical signal into spectral data. The coaxial optical path includes a first collimating lens and a first converging lens. The first collimating lens is used to convert the light emitted by the light source assembly into a parallel beam, and the first converging lens is used to directly converge the parallel beam emitted by the first collimating lens through the light source inlet to the sample reflection port. The spectrum generation component further includes a second collimating lens and a second converging lens. The second converging lens is disposed between the diffuse reflection light outlet of the integrating sphere and the entrance of the spectrometer. The second collimating lens is disposed between the second converging lens and the interferometer. The second converging lens is used to converge the diffuse reflection light emitted from the diffuse reflection light outlet of the integrating sphere to the second collimating lens. The second collimating lens is used to convert the beam converged by the second converging lens into a parallel beam.

2. The benchtop coal quality analysis spectrometer according to claim 1, characterized in that, The humidity control component includes a connecting cylinder and a drying cylinder. The connecting cylinder is detachably connected to the outer casing, and the drying cylinder is detachably connected to the connecting cylinder. A desiccant is placed inside the drying cylinder.

3. The benchtop coal quality analysis spectrometer according to claim 1, characterized in that, The light source assembly includes a light source, a light source bracket, a bracket heat sink, a light source heat sink, and a light source fan. The light source is mounted on the light source bracket, the light source bracket is mounted on the bracket heat sink, the light source heat sink surrounds the light source and the light source bracket, and the air outlet of the light source fan faces the light source heat sink.

4. The benchtop coal quality analysis spectrometer according to claim 1, characterized in that, It also includes a window, which is disposed above the sample reflection port.

5. The benchtop coal quality analysis spectrometer according to claim 1, characterized in that, It also includes a coal quality analysis component, which is connected to the spectrum generation component by wire or wireless means. The coal quality analysis component is used to convert the spectral data of the spectrum generation component into a spectrum, and to perform coal quality analysis through the spectrum.

6. The benchtop coal quality analysis spectrometer according to claim 5, characterized in that, The coal quality analysis component includes: The input module converts the spectral data into matrix form to obtain a spectral data matrix, wherein the spectral data is a multidimensional matrix composed of the absorbance of each spectrum of the coal sample. The covariance matrix construction module constructs the covariance matrix of the spectral data matrix input by the input module. The principal component analysis module performs principal component analysis on the covariance matrix constructed by the covariance matrix construction module to obtain the principal component spectral matrix composed of the principal components of each spectrum. The coal index matrix construction module constructs a one-dimensional or multi-dimensional coal index matrix of one or more coal indices. The coal indices include one or more of the following: dry ash-free basis, dry basis, air-dried basis, and received basis. The convolutional neural network construction module takes the principal component spectral matrix from the principal component analysis module as input and the coal index matrix constructed by the coal index matrix construction module as output to build a convolutional neural network. The training module trains the convolutional neural network using the training set, including: a training set construction unit to construct the training set; and a network training unit that sequentially passes the training set through the input module, the covariance matrix construction module, and the principal component analysis module to obtain the principal component spectral matrix of the training set, and obtains the coal index matrix of the training set through the coal index matrix construction module. The convolutional neural network constructed by the convolutional neural network construction module is then trained using the principal component spectral matrix and the coal index matrix of the training set. The analysis module inputs the principal component spectral matrix of the coal sample into the trained convolutional neural network to obtain the corresponding coal index matrix.

7. The benchtop coal quality analysis spectrometer according to claim 6, characterized in that, The coal quality analysis component also includes an interpolation module for interpolating the principal component spectral matrix of the principal component analysis module.

8. A method for coal quality analysis using a benchtop coal quality analysis spectrometer according to any one of claims 1-7, characterized in that, include: Place the coal sample in the sample cup; Check if the temperature inside the casing reaches the set temperature range; Check if the humidity inside the casing reaches the set humidity range; When the temperature inside the casing reaches the set temperature range and the humidity inside the casing reaches the set humidity range. The light emitted by the light source component is reflected by the coal sample in the sample cup and then passes through the spectral generation component to generate the spectrum of the coal sample.

9. The coal quality analysis method according to claim 8, characterized in that, Also includes: The spectral data is converted into matrix form to obtain a spectral data matrix, wherein the spectral data is a multidimensional matrix composed of the absorbance of each spectrum of the coal sample. Construct the covariance matrix of the spectral data matrix; Principal component analysis is performed on the covariance matrix to obtain the principal component spectral matrix composed of the principal components of each spectrum; Construct a one-dimensional or multi-dimensional coal index matrix for one or more coal indices, where the coal indices include one or more of the following: dry ash-free basis, dry basis, air-dried basis, and received basis. Using the principal component spectral matrix as input and the coal index matrix as output, a convolutional neural network is constructed: Training a convolutional neural network using a training set includes: constructing a training set; obtaining the principal component spectral matrix of the training set; obtaining the coal index matrix of the training set; and training the convolutional neural network using the principal component spectral matrix and the coal index matrix of the training set. The principal component spectral matrix of the coal sample is input into the trained convolutional neural network to obtain the corresponding coal index matrix.

10. The coal quality analysis method according to claim 9, characterized in that, Before the step of constructing a convolutional neural network by taking the principal component spectral matrix as input and the coal index matrix as output, the method further includes: interpolating the principal component spectral matrix.

Images