Method for classifying and identifying three types of kaolin based on terahertz time-domain spectroscopy

Abstract

The invention discloses a method for classifying and identifying three types of kaolin based on a terahertz time-domain spectroscopy technology, which comprises the following steps of: collecting time-domain spectrums of three types of kaolin samples, calculating absorption coefficients of different samples by utilizing corresponding formulas such as fast Fourier, and combining clustering analysis and principal component analysis methods to identify the three types of kaolin. The extraction of the Euclidean distance of spectral data and the dimension reduction processing of effective data are realized, and the Euclidean distance and the first principal component of a corresponding sample can be obtained. The result shows that the cumulative contribution rate of the first three principal components of the sample absorption coefficient screened by the PCA result can reach 97.26%, the samples are clustered into three classes according to the difference of the Euclidean distance by the CA result, and the cluster in the CA is consistent with the principal component result in the PCA. According to the method, three types of kaolin can be accurately identified and classified according to similarity and difference among different samples in combination with a CA-PCA model, and the identification rate can reach 99.99%. The method has the advantages of no damage to the detected kaolin sample, simple experiment operation, efficient spectrum acquisition and short detection period.

Description

Technical Field

[0001] This invention belongs to the field of terahertz technology application, and in particular relates to a method for classifying and identifying three types of kaolin based on terahertz time-domain spectroscopy. Background Technology

[0002] Terahertz waves are electromagnetic waves with frequencies ranging from 0.1 THz to 10 THz and wavelengths between 0.03 and 3 mm. Their band lies between microwaves and infrared radiation, belonging to the far-infrared band. Terahertz waves possess numerous characteristics such as fingerprint-like properties, strong penetrating power, and non-contact operation, containing rich chemical and physical information. Further spectral information of substances can be obtained through terahertz spectroscopy. Terahertz time-domain spectroscopy, as a non-destructive detection technique, utilizes the unique frequency response of many macromolecular vibrations and intermolecular forces, enabling qualitative and quantitative studies of substances to obtain their corresponding fingerprint spectra for efficient identification.

[0003] As is well known, kaolin has wide applications in industry and various sectors. With the expansion of its application areas, different types of kaolin possess their own characteristics and development and utilization methods. However, traditional classification and application methods often result in resource misallocation and waste. Traditional methods mainly include X-ray fluorescence spectroscopy, classifying kaolin based on the proportion of measured components; physical property determination methods, classifying based on differences in physical indicators; and microscopic observation and thermal analysis. However, traditional methods are destructive to samples, complex to operate, time-consuming and labor-intensive, and have high testing costs, with limited capacity for batch processing. Therefore, to achieve efficient utilization through resource identification, classification, and matching, and to accurately identify and classify kaolin to improve the utilization efficiency of kaolin resources, establishing three identification methods for kaolin is particularly important. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to overcome the defects of the prior art and provide a three-method classification and identification method for kaolin based on terahertz time-domain spectroscopy, so as to solve the limitations of the existing kaolin identification technology. The method has the characteristics of non-destructive testing, speed and efficiency.

[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution: This invention discloses a method for classifying and identifying three types of kaolin based on terahertz time-domain spectroscopy. The method mainly includes the following steps: (1) Measure the terahertz time-domain spectra of three standard kaolins: soft, sandy, and hard, and convert the terahertz time-domain spectra into absorption coefficient spectra to establish a database; (2) Establish a classification and identification model for the three types of kaolin based on their terahertz time-domain spectra; (3) Collect the terahertz time-domain spectra of the three types of kaolin to be identified, and determine the types of kaolin based on the terahertz spectra of the three types of kaolin and the established classification and identification model.

[0006] As a preferred technical solution of the present invention, the terahertz time-domain spectrum is converted into an absorption coefficient spectrum by the following method: the terahertz time-domain spectrum of three standard kaolin samples is measured by a terahertz time-domain spectroscopy measuring device, the terahertz time-domain spectrum is converted into the corresponding terahertz frequency-domain spectrum by fast Fourier transform, the absorption coefficient is calculated according to formula (1)-(2), and the absorption coefficient spectrum is established according to the frequency of the corresponding terahertz wave. (1) (2) Formula (1)-Formula (2), The phase difference between the sample and the reference signal; The thickness of the sample; The speed of light; ω is the angular frequency; where ω⁄ 2π is the frequency.

[0007] As a preferred embodiment of the present invention, step (2) mainly involves the following steps: (2.1) Establishment of classification model: 50% of the samples from different types of kaolin were selected to form the training set for model establishment. First, the terahertz spectrum of kaolin samples in the 0.5-2.75THz band was selected as data. Principal component analysis was used to analyze the corresponding data and extract the corresponding data features of the first principal component PC1 score map. Finally, Euclidean distance in the cluster analysis method was used to discriminate clusters. Kaolin identification models were established according to different types of kaolin. The modeling method should be completely consistent with the actual classification method of spectral processing. Finally, identification models of different types of kaolin were established according to different types of kaolin samples. (2.2) Validation of the classification model: The remaining 50% of the samples are used to form a validation set. First, the terahertz spectra in the 0.5-2.75THz band of the validation set are obtained as spectral data. Principal component analysis is used to analyze the data and obtain its features. Finally, the established kaolin identification model is called to obtain the sample categories of the validation set. The method of spectral processing of the samples in the validation set must be consistent with that in the training set.

[0008] As a preferred technical solution of the present invention, step (3) is as follows: the terahertz time-domain spectrum of the kaolin sample to be tested is measured by a terahertz time-domain spectroscopy measuring device, the time-domain spectrum of the kaolin sample to be tested in the 0.5-2.75THz terahertz band is selected, the data features are extracted by principal component analysis, and a classification and identification model is established by cluster analysis for identification.

[0009] As a preferred technical solution of the present invention, in obtaining the time-domain spectrum of the kaolin sample, the kaolin sample powder and the transparent auxiliary material polyethylene are mixed in a ratio of 1:5.

[0010] As a preferred technical solution of the present invention, the terahertz time-domain spectroscopy system should be operated at a temperature of 25°C, and the optical path should be vented with dry gas to reduce the relative humidity to below 5% RH.

[0011] As a preferred technical solution of the present invention, the terahertz frequency of the terahertz time-domain spectrum is between 0.5-2.75THz. Principal component analysis is used to extract data features, and cluster analysis is used to establish a classification and identification model.

[0012] Compared with the prior art, the beneficial effects of the present invention are as follows: The three kaolin classification and identification methods proposed in this invention are spectral identification methods that do not require complex sample pretreatment processes, do not consume chemical reagents, do not damage the kaolin samples being tested, can efficiently and accurately extract the fine structure and compositional differences of kaolin, improve the identification accuracy, are simple to operate, have efficient spectral acquisition, support batch sample analysis, and have a short detection cycle. Attached Figure Description

[0013] 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: Figure 1 The terahertz absorption coefficient spectra of three kaolin samples in the 0-8Thz band; Figure 2 The terahertz absorption coefficient spectra of three kaolin samples in the 0.5-2.75Thz band are shown. Figure 3 The terahertz time-domain spectra of three kaolin samples in the 0.5-2.75 THz band are shown. Figure 4 Dendrogram of cluster analysis for three kaolin samples; Figure 5 The histograms of scores for the three kaolin samples in the first principal component PC1 are shown. Detailed Implementation

[0014] The preferred embodiments of the present invention will be 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 present invention.

[0015] Example 1 The three kaolin classification and identification methods based on terahertz time-domain spectroscopy proposed in this invention include the following steps: (1) Measure the terahertz time-domain spectra of three standard kaolins: soft, sandy, and hard. Convert the terahertz time-domain spectra into absorption coefficient spectra and establish a database.

[0016] (2) A classification and identification model for the three types of kaolin was established based on the terahertz time-domain spectra of the three types of kaolin.

[0017] (3) Collect the terahertz time-domain spectra of the three types of kaolin to be identified, and determine the types of kaolin based on the terahertz spectra of the three types of kaolin and the established classification and identification model.

[0018] Based on the significant differences in the absorption of terahertz waves by different materials, when electromagnetic waves in the terahertz frequency band irradiate a sample within the terahertz frequency range, the molecules, atoms, or lattices in the sample will absorb the energy of the terahertz electromagnetic waves. The transmitted terahertz waves carry the information of the sample and generate a terahertz spectrum. Different types of kaolinite also have different responses in the terahertz frequency band, thus different types of materials have unique fingerprint characteristics.

[0019] Soft kaolin, sandy kaolin, and hard kaolin were selected as experimental samples for the study. Nine samples of each type were selected for the experiment. The implementation process of the three kaolin classification and identification methods based on terahertz time-domain spectroscopy is described in detail below: (1) Sample preparation and material preparation First, kaolin powder was mixed with polyethylene at a ratio of 1:5 and placed in a tableting mold. The mixture was then pressed at 20 MPa for 3 minutes to produce a 1.2 mm thick sample. To prevent the tablets from becoming brittle and cracking, and to prevent agglomeration, the experiment should be conducted at 25°C. Dry gas should be introduced through the optical path to reduce the relative humidity to below 5% RH, ensuring a dry environment. The entire process requires air-drying, pulverizing, grinding, and sieving the sample using a 200-mesh sieve, and the sample must be consistent with other reference samples.

[0020] The terahertz time-domain spectrum after passing through the sample was measured using a transmission terahertz spectroscopy system, and the frequency domain spectrum of the sample was obtained by fast Fourier transform. The absorption coefficient of the sample was calculated according to the formulas (1)-(2) described in the instruction manual, and the absorption coefficient spectrum was plotted. It can be seen that there are obvious differences in the terahertz time-domain spectra of different types of kaolin, and they also show different delay phenomena.

[0021] (2) Principal component analysis of the sample After measuring and calculating the time-domain spectrum and absorption spectrum of the samples, principal component analysis was used to analyze the absorption coefficient of the samples in the frequency range of 0.5–2.75 THz. The characteristic values ​​of the sample data, as well as the contribution rate and cumulative contribution rate of the principal components, were obtained. The data characteristics of the first three principal components of the terahertz absorption coefficient were retained. The contribution rate of the first principal component reached 78.48%, while the contribution rate of the first three principal components to the original data reached 97.27%. This indicates that the analysis results of the three principal components basically fully reflect the information of the sample. Therefore, the three principal components can be used to analyze the terahertz absorption spectral characteristics of the three types of kaolin.

[0022] (3) Cluster analysis of samples Simultaneously, cluster analysis was used to classify the three kaolin samples. The absorption coefficient data of the samples in the frequency range of 0.5 to 2.75 THz were used as the data input set. The Euclidean distance between samples in each category was calculated according to formula (3). Based on the differences in Euclidean distance, the samples were clustered into each category. Figure 4 As shown in the cluster analysis dendrogram, the Euclidean distance is calculated using formula (3): (3) in, and These represent the spectral data feature vectors of the m-th and n-th kaolin samples, respectively.

[0023] This patented method uses terahertz time-domain spectroscopy (THz-TDS) combined with machine learning to qualitatively analyze three types of kaolinite: soft, hard, and sandy. Dendrograms and PC1 score maps obtained through cluster analysis and principal component analysis comprehensively reflect the differences and similarities among the three types of kaolinite. Principal component analysis visually presents the sample differences, while cluster analysis, based on Euclidean distance, achieves precise category classification. The combination of these two methods objectively and comprehensively reveals the category differences among the three types of kaolinite. In practical applications, a corresponding classification model can be established according to the method described in this patent to classify the kaolinite species to be identified.

[0024] Table 1: Eigenvalues ​​of Principal Component Data for Absorption Coefficient .

[0025] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present 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 method for classifying and identifying three types of kaolin based on terahertz time-domain spectroscopy, characterized in that, This method mainly includes the following steps: (1) Measure the terahertz time-domain spectra of three standard kaolins: soft, sandy, and hard, and convert the terahertz time-domain spectra into absorption coefficient spectra to establish a database; (2) Establish a classification and identification model for the three types of kaolin based on their terahertz time-domain spectra; (3) Collect the terahertz time-domain spectra of the three types of kaolin to be identified, and determine the types of kaolin based on the terahertz spectra of the three types of kaolin and the established classification and identification model.

2. The method according to claim 1, characterized in that, The following method is used to convert the terahertz time-domain spectrum into the absorption coefficient spectrum: the terahertz time-domain spectrum of three standard kaolin samples is measured by a terahertz time-domain spectroscopy measuring device, the terahertz time-domain spectrum is obtained by fast Fourier transform to obtain the corresponding terahertz frequency-domain spectrum, the absorption coefficient is calculated according to formula (1)-(2), and the absorption coefficient spectrum is established according to the frequency of the corresponding terahertz wave. (1) (2) Formula (1)-Formula (2), The phase difference between the sample and the reference signal; The thickness of the sample; The speed of light; ω is the angular frequency; where ω⁄ 2π is the frequency.

3. The method according to claim 1, characterized in that, The main steps of step (2) are as follows: (2.1) Establishment of classification model: 50% of the samples from different types of kaolin were selected to form the training set for model establishment. First, the terahertz spectrum of kaolin samples in the 0.5-2.75THz band was selected as data. Principal component analysis was used to analyze the corresponding data and extract the corresponding data features of the first principal component PC1 score map. Finally, Euclidean distance in the cluster analysis method was used to discriminate clusters. Kaolin identification models were established according to different types of kaolin. The modeling method should be completely consistent with the actual classification method of spectral processing. Finally, identification models of different types of kaolin were established according to different types of kaolin samples. (2.2) Validation of the classification model: The remaining 50% of the samples were used to form a validation set. First, the terahertz spectra in the 0.5-2.75 THz band of the validation set were obtained as spectral data. Principal component analysis was used to analyze the data and obtain its features. Finally, the established kaolin identification model is invoked to obtain the sample categories in the validation set, and the method and process of sample spectral processing in the validation set must be consistent with those in the training set.

4. The method according to claim 1, characterized in that, The step (3) is as follows: the terahertz time-domain spectrum of the kaolin sample to be tested is measured by the terahertz time-domain spectroscopy measuring device, the time-domain spectrum of the kaolin sample to be tested in the 0.5-2.75THz terahertz band is selected, the data features are extracted by principal component analysis, and a classification and identification model is established by cluster analysis for identification.

5. The method according to claim 1, characterized in that, In obtaining the time-domain spectrum of the kaolin sample, the kaolin sample powder was mixed with the transparent auxiliary material polyethylene at a ratio of 1:

5.

6. The method according to claim 1, characterized in that, Terahertz time-domain spectroscopy should be performed at a temperature of 25°C, and the optical path should be circulated with dry gas to reduce the relative humidity to below 5% RH.

7. The method according to claim 1, characterized in that, The terahertz time-domain spectra have terahertz frequencies ranging from 0.5 to 2.75 THz. Principal component analysis was used to extract data features, and cluster analysis was employed to establish a classification and identification model.

Images