Particle sample transmission-type near-infrared analysis method for enhancing stability, medium and equipment

Abstract

The invention discloses a particle sample transmission-type near-infrared analysis method capable of enhancing stability, a medium and equipment. The method comprises the following steps: S1, incident light is focused and collimated, passes through an antireflection film glass baffle and a sample, and then receives transmission energy by 16 optical fibers; s2, a servo motor controls the distance H between a baffle and an optical fiber carrying plane baffle, and spectrums are collected at 1.5 cm and 2 cm respectively; s3, uniformly arranging the light receiving optical fibers, and measuring the energy intensity of the inner and outer rings of light receiving optical fibers; and S4, sampling at variable integration time, and collecting the spectrum at the integration time of 50ms and 200ms. By optimizing lighting and light receiving, spectrum deviation caused by sample density and light source intensity changes is eliminated, the problem of CCD uniformity difference is solved, and the model adaptability is enhanced.

Description

Technical Field

[0001] The invention relates to the field of analysis technology, and in particular to a particle sample transmission near-infrared spectroscopy analysis method, medium and equipment. Background Art

[0002] Near infrared spectroscopy (NIR) analysis technology has been widely used in process control in many fields such as medicine, grain and oil processing, and industrial production due to its advantages of non-pollution, non-destructive and rapid analysis. Under the premise of precise control of measurement conditions (such as light source intensity, ambient temperature, etc.), the quantitative analysis model constructed by this technology has excellent prediction accuracy for liquid and powder samples. However, when the collection conditions cannot be kept consistent, especially when the uniformity of granular samples is poorer than that of liquid and powder samples, the prediction stability of the model will be significantly reduced.

[0003] Currently, strategies to improve model stability can be divided into three categories. First, ensure the consistency between the modeling environment and the measurement environment to reduce the impact of environmental differences. Second, use spectral preprocessing techniques such as derivative processing and multivariate scattering correction to optimize spectral data to reduce data noise and improve signal quality. Finally, by actively introducing potential interference factors in the modeling process and simulating the interference that may be encountered in actual applications, the robustness of the model to these interference factors is enhanced, further improving the stability of the model. Summary of the invention

[0004] In the analysis of powder and solution samples, it is relatively easy to maintain the consistency of the modeling environment. However, for particle samples, a transmission measurement method is usually used. In view of the significant individual differences between particle samples, a single sample must be sampled multiple times in order to increase the amount of information. In conventional transmission near-infrared analysis systems, it is difficult to keep the density of the sample constant during the feeding process, which leads to poor repeatability of the spectral data. In addition, the energy intensity of the near-infrared spectrum in the short-wave near-infrared region varies greatly, especially the energy of the band before 800nm ​​is strong, and the instability of the signal after 800nm ​​will have a negative impact on the stability of the model. In practical applications, the measurement conditions will inevitably be disturbed by various factors such as light source fluctuations, temperature changes, and instrument system differences, all of which may lead to reduced model stability. In order to solve this technical problem, the present invention proposes a particle sample transmission near-infrared analysis method with enhanced stability, which improves the stability of spectral data by optimizing the illumination and light collection. Taking the relative transmission spectrum as the independent variable, the spectral shift phenomenon caused by changes in sample density and light source intensity is effectively eliminated, and the waveform abnormality problem caused by the uniformity difference of the charge coupled device (CCD) in the spectrometer is solved. In addition, the adaptability of the model is significantly enhanced by combining with the spectral correction algorithm.

[0005] According to one aspect of the present invention, a particle sample transmission near-infrared analysis method with enhanced stability is provided. The following steps are involved: S1: The incident light is focused and collimated, the spot size is reduced, the intensity is increased, and it passes through the anti-reflection film glass baffle and the sample and hits the optical fiber carrying flat baffle. The N optical fibers above receive the transmitted energy, where N is 12-18. S2: The servo motor controls the lateral movement of the glass baffle to control the distance H between the glass baffle and the optical fiber carrying plane baffle. Each sample will collect a spectrum at distance h1 and distance h2, where the difference between h1 and h2 is 0.5 cm, and h1 and h2 are both selected in the range of 2-3 cm with a gradient of 0.5 cm; S3: The receiving optical fiber is N optical fibers evenly arranged on two circles with radius r1 and r2 from the detection center, with N / 2 receiving optical fibers in the inner and outer circles respectively. The energy intensities of the receiving optical fibers in the inner and outer circles are i_n and i_w respectively, where the difference between r1 and r2 is 0.5cm, and r1 and r2 are both selected within the range of 20mm; S4: The spectrometer adopts variable integration time sampling. The bottom rotating wheel controls the material feeding. After pausing for 3 to 5 seconds after each rotation, the spectrum is collected using 50ms and 200ms integration times respectively.

[0006] The above-mentioned stability-enhanced transmission near-infrared analysis method for particle samples also includes: wherein N is 16.

[0007] The above-mentioned stability-enhanced transmission near-infrared analysis method for particle samples also includes: wherein h1 is 1.5 cm and h2 is 2 cm.

[0008] The above-mentioned stability-enhanced transmission near-infrared analysis method for particle samples also includes: wherein r1 is 5 mm and r2 is 12 mm.

[0009] The above-mentioned enhanced stability particle sample transmission near infrared analysis method also includes: Define the number of modeling sample sets as m, the number of spectral bands as b, the independent variable spectral matrix as X0 (m, 2b), and the dependent variable as Y0 (m, 1), then X0 is The size of the matrix, Y0 is sized matrix and includes the following acquisition steps for each sample: S5: Adjust the baffle distance H to 1.5 cm; S6: Pour 300ml sample, the bottom wheel controls the material feeding, and after pausing for 3 seconds after each rotation, the spectrum is collected with 50ms and 200ms integration time respectively. Taking the first time as an example, the received energies of the inner and outer ring receiving fibers under 50ms are I50_n and I50_w respectively, and the received energies of the inner and outer ring receiving fibers under 200ms are I200_n and I200_w respectively; S7: Spectral band stitching, taking I50_n and I200_n as examples, calculate the ratio average ka of the spectrum of the unsaturated interval of I200_n and the corresponding spectrum interval of I50_n. For the saturated interval of I200_n, multiply the spectrum of I50_n in the corresponding interval by ka, so as to raise the spectrum of 50 integration time to replace the saturated part under 200 integration time, and obtain I_n. The same method is used to obtain I_w; S8: Calculate k=I_n / I_w to get the relative transmission spectrum k of this time. In the same way, continue to rotate the bottom wheel and repeat the above steps 20 times. The average value of the 20 relative transmission spectra k is saved as Ki1, which is the spectrum of the current sample at H of 1.5cm, and i is the serial number of the current sample. S9: Adjust the baffle distance H to 2 cm, repeat steps S6-8, and obtain the spectrum Ki2 of the current sample at H of 2 cm; S10: Merge the K1K2 spectral matrices to obtain the final spectrum Ai=[Ki1,Ki2] of the current sample.

[0010] The above-mentioned enhanced stability transmission near-infrared analysis method for particle samples further includes the following modeling steps: S11: Collect 5 standard sample spectra x_t (5, 2b) with known contents, and the corresponding contents are y_t (5, 1). Calculate the standard spectrum x_t_b (5, 2b) of x_t according to the distance weight of each value in y_t to the modeling sample y0. The conversion matrix is ​​R (5, m). Define the number of modeling samples as m and the number of independent variables as 2b. Then we get , each row of m elements in R is the inverse distance weight of the m samples in x0 to the corresponding samples in the current row. The farther the distance, the smaller the weight. The weight function is simulated according to the distribution of the m sample dependent variables. For example, if it conforms to the normal distribution, the Gaussian function can be used as the weight function. The variance of Y0 of the m modeling sample set is defined as σ, then .

[0011] S12: Perform principal component analysis on the residuals of x_t_b and x_t, with the projection axis as W, extract the principal component T of the first 85% explained variance, and the corresponding load is P .

[0012] S13: Use the p matrix to correct the modeling sample x0 to obtain the corrected modeling sample spectrum x0_c, and use x0_c and y0 as independent variables and dependent variables for modeling. .

[0013] According to another aspect of the present invention, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed, the analysis method described in the embodiment of the present invention is implemented.

[0014] According to another aspect of the present invention, there is provided a computer device, including a processor, a memory and a computer program stored in the memory, wherein the processor implements the analysis method described in the embodiment of the present invention when executing the computer program.

[0015] The present invention provides a transmission near-infrared analysis method for particle samples with enhanced stability. Compared with the traditional method, the stability of spectral data is improved by optimizing the illumination and light collection. Taking the relative transmission spectrum as the independent variable, the spectral shift phenomenon caused by the change of sample density and light source intensity is effectively eliminated, and the waveform abnormality problem caused by the uniformity difference of the charge coupled device (CCD) in the spectrometer is solved. In addition, combined with the spectral correction algorithm, the adaptability of the model is significantly enhanced. Through the above beneficial effects, the analysis method of the present invention performs well in practical applications, especially in the rapid detection and analysis of particle samples. It not only improves the accuracy of measurement, but also greatly shortens the analysis time due to its simple operation, making the method have high practical value in industrial production. In addition, the method also has good repeatability and reliability, and can meet the strict requirements of different industries for particle sample analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The accompanying drawings, which are incorporated into the specification and constitute a part of the specification, illustrate embodiments of the present invention and are used to explain the principles of the present invention together with the relevant text description. In these drawings, similar reference numerals are used to represent similar elements. The drawings described below are some embodiments of the present invention, but not all embodiments. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without paying creative work.

[0017] Figure 1 A hardware schematic diagram of a particle sample transmission near-infrared analysis method with enhanced stability provided by an embodiment of the present invention is shown.

[0018] Figure 2 A schematic diagram of the distribution of receiving optical fibers of an optical fiber-mounted flat baffle in a transmission-type near-infrared analysis method for particle samples with enhanced stability provided by an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0019] In order to make the purpose, technical scheme and advantages of the embodiments of the present invention clearer, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. It should be noted that the embodiments in this application and the features in the embodiments can be combined with each other arbitrarily without conflict.

[0020] When dealing with the analysis of powder and solution samples, it is relatively easy to maintain the consistency of the modeling environment. However, for particle samples, this task becomes complicated because particle samples usually need to adopt a transmission measurement method. There are significant individual differences between particle samples. In order to increase the amount of information, a single sample must be sampled multiple times. In the conventional transmission near-infrared analysis method, the density of the sample is difficult to keep constant during the feeding process, which leads to poor repeatability of the spectral data. In addition, the energy intensity of the near-infrared spectrum in the short-wave near-infrared region varies greatly, especially the band energy before 800nm ​​is strong, and the signal instability after 800nm ​​will have a negative impact on the stability of the model. In actual application, the measurement conditions will inevitably be disturbed by various factors such as light source fluctuations, temperature changes and instrument system differences, all of which may lead to reduced model stability. The present embodiment aims to solve the above problems through a new type of particle sample transmission near-infrared analysis method with enhanced stability, and improve the accuracy of particle sample analysis and the stability of the model.

[0021] One embodiment of the present invention provides a method for analyzing particle samples by transmission near infrared with enhanced stability, the operation and configuration details of which can be found in the attached Figure 1 The hardware diagram in the figure provides a more intuitive understanding, including: S1: The incident light is focused and collimated, the spot size is reduced, the intensity is increased, and it passes through the anti-reflection film glass baffle ① and the sample and hits the optical fiber carrying flat baffle ②. The 12-18 optical fibers above receive the transmitted energy. The preferred light is 16 (see attached Figure 2 As shown in the figure); by focusing and collimating the incident light, the spot size is reduced and the light intensity is increased, which helps to improve the accuracy and sensitivity of spectral acquisition. After passing through the anti-reflection film glass baffle ① and the sample, the light hits the fiber-mounted flat baffle ②, and the transmitted energy is received by 16 optical fibers. This design enhances the light receiving efficiency and signal stability.

[0022] S2: The servo motor controls the lateral movement of the glass baffle ① to control the distance H between the glass baffle ① and the optical fiber carrying plane baffle ②. Each sample will perform a spectrum acquisition at a distance h1 and a distance h2, wherein the difference between h1 and h2 is 0.5 cm, and h1 and h2 are both selected within the range of 2-3 cm with a gradient of 0.5 cm. Preferably, each sample will perform a spectrum acquisition at a distance H of 1.5 cm and 2 cm. The servo motor controls the lateral movement of the glass baffle ① to accurately control the distance H between the glass baffle ① and the optical fiber carrying plane baffle ② to ensure that the spectrum is collected at a distance of 1.5 cm and 2 cm. This precise control helps to reduce errors caused by changes in sample position and improve measurement repeatability and accuracy.

[0023] S3: The receiving optical fiber is 16 evenly arranged optical fibers on two circles with a radius of r1 and r2 from the detection center. Preferably, r1 and r2 are 5 mm and 12 mm respectively. There are 8 outer circle receiving optical fibers ③ and 8 inner circle receiving optical fibers ④. The energy intensities of the inner and outer circle receiving optical fibers are i_n and i_w respectively. The receiving optical fiber is designed to be 16 evenly arranged optical fibers on two circles with a radius of r1 and r2 from the detection center. The difference between r1 and r2 is 0.5 cm, and r1 and r2 are both selected within the range of 20 mm. Preferably, r1 is 5 mm and r2 is 12 mm. There are 8 receiving optical fibers in the inner and outer circles, which ensures that the number of inner and outer circles is equal and evenly arranged, and the inner and outer circle optical fibers and the center of the circle are on the same straight line. This layout helps to evenly collect the spectral information of the sample from different angles, and enhances the comprehensiveness and representativeness of the spectral data.

[0024] S4: The spectrometer adopts variable integration time sampling, and the bottom rotating wheel ⑤ controls the material feeding. After pausing for 3 seconds after each rotation, the spectrum is collected with 50ms and 200ms integration times respectively. The spectrometer adopts variable integration time sampling, and the bottom rotating wheel ⑤ controls the material feeding. After pausing for 3 seconds after each rotation, the spectrum is collected with 50ms and 200ms integration times respectively. This sampling strategy helps to capture spectral changes at different time scales and improve the dynamic range and information content of spectral data. In specific implementation, the rotating wheel controls the material feeding amount by controlling the angle of each rotation (45 degrees per grid), and the material feeding amount per grid is maintained at about 20ml. The rotating wheel speed is kept constant, and the pause time is adjusted to ensure uniform and stable material feeding.

[0025] The enhanced stability transmission near-infrared analysis method for particle samples according to the embodiment of the present invention can provide the following advantages: By precisely controlling the distance between the sample and the optical fiber, and using variable integral time sampling technology, the system can reduce spectral shifts caused by changes in sample density and light source intensity, as well as waveform anomalies caused by differences in the uniformity of the spectrometer CCD. Multiple sampling of a single sample, combined with energy intensity measurements of the inner and outer ring light-receiving fibers, can provide more comprehensive spectral data and enhance the amount of information analyzed. Specifically, this optimized lighting method, combined with the use of anti-reflection film glass baffles, ensures that light effectively passes through the sample and is received by the 16 inner and outer ring light-receiving fibers on the optical fiber-mounted flat baffle. This design not only improves the quality of spectral data collection, but also enhances the light reception efficiency and signal stability through the uniform arrangement of the inner and outer ring fibers, thereby providing a solid data foundation for subsequent spectral analysis.

[0026] Another embodiment of the present invention provides a particle sample transmission near-infrared analysis method with enhanced stability, further comprising defining the number of modeling sample sets as m, the number of spectral bands as b, the independent variable spectral matrix as X0 (m, 2b), and the dependent variable as Y0 (m, 1), then X0 is The size of the matrix, Y0 is sized matrix and perform the following acquisition steps for each sample: S5: Adjust the baffle distance H to 1.5 cm; S6: Pour 300ml of sample, and the bottom wheel controls the feeding. After pausing for 3 seconds after each rotation, the spectrum is collected with 50ms and 200ms integration time respectively. Taking the first time as an example, the energy received by the inner and outer ring light receiving fibers under 50ms is I50_n and I50_w respectively, and the energy received by the inner and outer ring light receiving fibers under 200ms is I200_n and I200_w respectively; by controlling the sample feeding and collection time, this step ensures the uniform distribution of the sample and the consistency of the spectral data. The design of pausing for 3 seconds allows the system to stabilize and prepare for the next collection, while the setting of different integration times is used to capture the spectral characteristics of the sample under different exposure conditions, providing more comprehensive data for subsequent data analysis.

[0027] S7: Spectral band stitching, taking I50_n and I200_n as examples, calculate the ratio average ka of the spectrum of the unsaturated interval of I200_n and the corresponding spectral interval of I50_n. For the saturated interval of I200_n, multiply the spectrum of I50_n in the corresponding interval by ka, so as to raise the spectrum of 50 integration time to replace the saturated part under 200 integration time, and obtain I_n. The same method is used to obtain I_w. Through spectral band stitching and correction processing, the spectral saturation problem caused by different integration times is solved. By calculating the ratio average ka and applying it to the saturated interval, the accuracy and integrity of the spectral data are ensured, providing a reliable basis for the subsequent relative transmission spectrum calculation.

[0028] S8: Calculate k=I_n / I_w to obtain the relative transmission spectrum k of this time. In the same way, continue to rotate the bottom wheel, repeat the steps for more than 20 times, and save the average value of the relative transmission spectrum k of 20 times as Ki1 is the spectrum of the current sample at H of 1.5cm, and i is the current sample number; by repeating the measurement and calculating the average value, this step significantly improves the stability and reliability of the data, thereby reducing the influence of random errors, and provides a more accurate description of the sample transmission characteristics, providing a solid data basis for subsequent sample analysis. Moreover, the present invention uses the relative transmission spectrum k (k=I_n / I_w) as an independent variable, where I_n and I_w are the energy intensities received by the inner and outer ring optical fibers, respectively. This innovative method effectively eliminates the spectral shift caused by changes in sample density and light source intensity, as well as the waveform anomaly caused by differences in the uniformity of the spectrometer CCD, by comparing the energy received by the inner and outer ring optical fibers. The layout of the optical fiber and the configuration of the spectrometer in the figure intuitively demonstrate this design intent, ensuring the accuracy and comprehensiveness of the spectral data.

[0029] S9: Adjust the baffle distance H to 2 cm, repeat steps S6-8, and obtain the spectrum Ki2 of the current sample at H of 2 cm; this step is designed to capture the transmission characteristics of the sample at different optical path lengths. This design allows the system to evaluate the impact of changes in sample transmission characteristics with distance, providing important data for establishing a more comprehensive sample model.

[0030] S10: Merge the K1K2 spectral matrices to obtain the final spectrum Ai=[Ki1,Ki2] of the current sample; by integrating the spectral information at two distances, the final spectrum Ai provides a more complete and accurate sample description, which is crucial for subsequent sample identification and quantitative analysis.

[0031] In this embodiment, the number of modeling sample sets is first defined as m, the number of spectral bands is b, and the independent variable spectral matrix X0 (m, 2b) and the dependent variable Y0 (m, 1) are constructed, where X0 is a matrix of size m2b and Y0 is a matrix of size m1. Then, a series of detailed acquisition steps are performed, including adjusting the baffle distance H, controlling sample feeding, collecting spectra using different integration times, and performing band splicing and relative transmission spectrum calculation on the collected spectral data. By repeating these steps and calculating the average value, the spectra Ki1 and Ki2 of the sample at different H are obtained, and they are merged to obtain the final spectrum Ai. This method embodiment significantly improves the stability and accuracy of particle sample analysis through precise spectral acquisition and advanced data processing technology.

[0032] Another embodiment of the present invention provides a particle sample transmission near-infrared analysis method with enhanced stability, further comprising the following modeling steps: S11: Collect 5 standard sample spectra x_t(5,2b) with known contents (these 5 samples are collected in the current usage environment, suitable for the current environment and the current machine, and are different from the collection environment of the previous m modeling sample sets, and the machine may also be different). The corresponding content is y_t(5,1). The standard spectrum x_t_b(5,2b) of x_t is calculated according to the distance weight of each value in y_t to the modeling sample y0. The conversion matrix is ​​R(5,m). The number of modeling samples is defined as m and the number of independent variables is 2b. Then we get , each row of m elements in R is the inverse distance weight of the m samples in x0 to the corresponding samples in the current row. The farther the distance, the smaller the weight. The weight function is simulated according to the distribution of the m sample dependent variables. For example, if it conforms to the normal distribution, the Gaussian function can be used as the weight function. The variance of Y0 of the m modeling sample set is defined as σ, then .

[0033] In the modeling process, the present invention collects spectra x_t of 5 standard samples with known contents, and calculates the standard spectrum x_t_b of x_t according to the distance weight of each value in y_t to the modeling sample y0. This step embodies the process of spectrally correcting the original modeling sample set using standard samples to adapt the model to the current environment and equipment, which is crucial to improving the adaptability and accuracy of the model. Through principal component analysis and correction of modeling samples, the present invention significantly enhances the adaptability of the model to different environments and equipment, making it more widely applicable and more practical in practical applications.

[0034] S12: Perform principal component analysis on the residuals of x_t_b and x_t, with the projection axis as W, extract the principal component T of the first 85% explained variance, and the corresponding load is P .

[0035] S13: Use the p matrix to correct the modeling sample x0 to obtain the corrected modeling sample spectrum x0_c, and use x0_c and y0 as independent variables and dependent variables for modeling.

[0036] This involves collecting the spectrum of a standard sample x_t with a known content, and calculating the standard spectrum x_t_b of x_t based on the distance weight from each value in y_t to the modeling sample y0. The key components are extracted through principal component analysis, and these components are used to calibrate the modeling sample, and finally a prediction model that can adapt to the current environment and equipment is established. This method not only improves the stability of the analysis results, but also enhances the adaptability of the model to different environments and equipment, making it more widely applicable and more practical in practical applications.

[0037] In summary, the embodiments of the present invention effectively solve the key challenges in particle sample analysis through a series of technical improvements, including sample density changes, light source intensity fluctuations, spectrometer CCD uniformity differences, etc., thereby significantly improving the robustness and applicability of the model.

[0038] First, by precisely controlling the focusing and collimation of the incident light and the layout of the fiber receiving system, the present invention ensures high-quality acquisition of spectral data and enhances the strength and stability of the signal. Second, by adjusting the baffle distance and adopting variable integration time sampling technology, the system can adapt to the characteristics of different samples and reduce the spectral shift caused by changes in sample density and light source intensity. In addition, by measuring the energy intensity of the inner and outer ring light receiving fibers and splicing the spectral bands, the system further improves the accuracy and comprehensiveness of the spectral data.

[0039] In terms of data processing, the present invention effectively reduces random errors and improves the repeatability and reliability of data by calculating the relative transmission spectrum and the average value of multiple acquisitions. Through these technical means, the present invention not only improves the accuracy of single sample analysis, but also enhances the adaptability of the model to different environments and equipment through principal component analysis and correction of modeling samples, making it more widely applicable and more practical in practical applications.

[0040] In general, the embodiments of the present invention provide a particle sample transmission near-infrared analysis solution with enhanced stability through technological innovation, which can meet the needs of fast, efficient and accurate analysis in industrial production and has important practical application value and broad market prospects.

[0041] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A particle sample transmission near-infrared analysis method with enhanced stability, characterized in that: The following steps are involved: S1: The incident light is focused and collimated, the spot size is reduced, the intensity is increased, and it passes through the anti-reflection film glass baffle and the sample and hits the optical fiber carrying flat baffle. The N optical fibers above receive the transmitted energy, where N is 12-18. S2: The servo motor controls the lateral movement of the glass baffle to control the distance H between the glass baffle and the optical fiber carrying plane baffle. Each sample will collect a spectrum at distance h1 and distance h2, where the difference between h1 and h2 is 0.5 cm, and h1 and h2 are both selected in the range of 2-3 cm with a gradient of 0.5 cm; S3: The receiving optical fiber is N optical fibers evenly arranged on two circles with radius r1 and r2 from the detection center, with N / 2 receiving optical fibers in the inner and outer circles respectively. The energy intensities of the receiving optical fibers in the inner and outer circles are i_n and i_w respectively, where the difference between r1 and r2 is 0.5cm, and r1 and r2 are both selected within the range of 20mm; S4: The spectrometer adopts variable integration time sampling. The bottom rotating wheel controls the material feeding. After pausing for 3 to 5 seconds after each rotation, the spectrum is collected using 50ms and 200ms integration times respectively.

2. The method for transmission near-infrared analysis of particle samples with enhanced stability according to claim 1, characterized in that: in, N is 16.

3. The method for transmission near-infrared analysis of particle samples with enhanced stability according to claim 2, characterized in that: in, h1 is 1.5cm and h2 is 2cm.

4. The method for transmission near-infrared analysis of particle samples with enhanced stability according to claim 3, characterized in that: in, r1 is 5mm and r2 is 12mm.

5. The method for transmission near-infrared analysis of particle samples with enhanced stability according to claim 4, characterized in that: Also includes: Define the number of modeling sample sets as m, the number of spectral bands as b, the independent variable spectral matrix as X0 (m, 2b), and the dependent variable as Y0 (m, 1), then X0 is The size of the matrix, Y0 is sized matrix and includes the following acquisition steps for each sample: S5: Adjust the baffle distance H to 1.5 cm; S6: Pour 300 ml of sample, and the bottom rotating wheel controls the material feeding. After pausing for 3 seconds after each rotation, the spectrum is collected with 50 ms and 200 ms integration time respectively. The received energies of the inner and outer ring receiving optical fibers under 50 ms are I50_n and I50_w respectively, and the received energies of the inner and outer ring receiving optical fibers under 200 ms are I200_n and I200_w respectively; S7: Spectral band splicing, calculate the ratio average ka of the spectrum of the unsaturated interval of I200_n and the corresponding spectrum interval of I50_n. For the saturated interval of I200_n, multiply the spectrum of I50_n in the corresponding interval by ka, so as to raise the spectrum of 50 integration time to replace the saturated part under 200 integration time, and obtain I_n. The same method is used to obtain I_w; S8: Calculate k=I_n / I_w to get the relative transmission spectrum k of this time. In the same way, continue to rotate the bottom wheel and repeat the above steps 20 times. The average value of the 20 relative transmission spectra k is saved as Ki1, which is the spectrum of the current sample at H of 1.5cm, and i is the serial number of the current sample. S9: Adjust the baffle distance H to 2 cm, repeat steps S6-8, and obtain the spectrum Ki2 of the current sample when H is 2 cm; S10: Merge the K1K2 spectral matrices to obtain the final spectrum A of the current sample i =[Ki1,Ki2].

6. The method according to claim 5, characterized in that The following modeling steps are also included: S11: Collect 5 standard sample spectra x_t (5, 2b) with known contents, and the corresponding contents are y_t (5, 1). Calculate the standard spectrum x_t_b (5, 2b) of x_t according to the distance weight of each value in y_t to the modeling sample y0. The conversion matrix is ​​R (5, m). Define the number of modeling samples as m and the number of independent variables as 2b. Then we get , each row of m elements in R is the inverse distance weight of the m samples in x0 to the corresponding samples in the current row. The farther the distance, the smaller the weight. The weight function is simulated according to the distribution of the m sample dependent variables. For example, if it conforms to the normal distribution, the Gaussian function can be used as the weight function. The variance of Y0 of the m modeling sample set is defined as σ, then S12: Perform principal component analysis on the residuals of x_t_b and x_t, with the projection axis as W, extract the principal component T of the first 85% explained variance, and the corresponding load is P S13: Use the p matrix to correct the modeling sample x0 to obtain the corrected modeling sample spectrum x0_c, and use x0_c and y0 as independent variables and dependent variables for modeling. 。 7. A computer-readable storage medium, characterized in that: A computer program is stored, and when the computer program is executed, the method according to any one of claims 1 to 6 is implemented.

8. A computer device, characterized in that: The method comprises a processor, a memory and a computer program stored in the memory, wherein when the processor executes the computer program, the method according to any one of claims 1 to 6 is implemented.

Images