Coal origin tracing method based on Fourier transform infrared spectroscopy combined with metabolomics analysis

Abstract

The invention discloses a coal origin tracing method combining Fourier infrared spectroscopy with metabolomics analysis. The method comprises the following steps: collecting coal to be analyzed from n different locations and storing the coal in sealed plastic barrels, and then preparing coal samples with a particle size less than 0.2 mm respectively; using a metabolomics analysis strategy to process the infrared spectrum, and then determining which wavenumbers have significantly different absorbances, and successfully screening out wavenumber markers for each coal group. Finally, by searching for wavenumber markers, a novel analysis method for exploring infrared spectral information based on a metabolomics analysis strategy is provided, so as to achieve the purpose of completely distinguishing coal from different geographical origins.

Description

Technical Field

[0001] The invention relates to the technical field of coal origin tracing, in particular to a coal origin tracing method combining Fourier transform infrared spectroscopy with metabolomics analysis. Background Art

[0002] Coal is known as "black gold" and "industrial food". The chemical composition of coal and the geographical origin of the formation environment are important indicators for measuring the quality grade of coal, and origin tracing is conducive to coal management. So far, some studies have been carried out on the origin tracing methods of coal. Traditional methods require a series of experiments and analyses, and have disadvantages such as high cost, low degree of automation and strong sample destructiveness. Although other methods can obtain results quickly, they cannot fully reflect the differences in coal origins. For example, microwaves and gamma rays are only used to determine single parameters of coal, such as moisture and ash content. Infrared spectroscopy has gradually become an effective tool for achieving coal origin tracing due to its simple, fast and non-destructive testing process. Since coal from different origins may be formed in different environments (such as temperature and humidity), there are differences in coal structure and content, and this difference can be well reflected by infrared spectroscopy. Based on this, predecessors have established a coal traceability model by combining infrared spectroscopy with chemometrics or machine learning algorithms, and achieved good prediction results, laying a solid foundation for the widespread application of infrared spectroscopy in coal origin tracing.

[0003] Infrared spectroscopy is characterized by high dimensionality, multi-redundancy and collinearity, which is very similar to liquid (or gas) chromatography-tandem mass spectrometry, which has been successfully used to trace the origin of many products after being combined with metabolomics analysis. So, is it possible to combine infrared spectroscopy with metabolomics analysis to achieve the origin traceability of coal?

[0004] Therefore, it is necessary to establish a reliable traceability method combining infrared spectroscopy with metabolomics analysis that can be used to identify the origin of coal from different origins. Summary of the invention

[0005] The purpose of the present invention is to solve the deficiencies in the prior art and to provide a method for tracing the origin of coal by combining Fourier transform infrared spectroscopy with metabolomics analysis.

[0006] To achieve the above object, the present invention is implemented according to the following technical solutions:

[0007] A method for tracing the origin of coal by combining Fourier transform infrared spectroscopy with metabolomics analysis comprises the following steps:

[0008] Step 1: Collect coal to be analyzed from n different locations and store them in sealed plastic barrels, and then prepare coal samples with a particle size less than 0.2 mm respectively;

[0009] Step 2, select 9 parallel samples from each coal sample, name them as sample 1-1 to sample 1-9, sample 2-1 to sample 2-9, ... and sample n-1 to sample n-9, corresponding to No.1, No.2, ... and No.n coal groups; in an air drying room where the temperature and humidity are set to 20°C and 10%, respectively, use a Fourier transform infrared spectrometer to record the infrared spectra of the 9 parallel samples of each coal sample, mix equal amounts of the 9 parallel samples of each coal sample and prepare them into quality control QC samples, use the Fourier transform infrared spectrometer to repeatedly scan the quality control QC samples 3 times to obtain 3 infrared spectra of the quality control QC samples, and scan the remaining coal samples once using the Fourier transform infrared spectrometer to obtain the original spectra of the coal samples;

[0010] Step 3, each derived infrared spectrum is expressed by a series of points on the coordinate axis, wherein the horizontal and vertical coordinates of the coordinate axis are wave number and absorbance, respectively, and the entire infrared spectrum is translated along the Y axis until the lowest point of the spectrum reaches the X axis, at which point the absorbance is zero; the wave number with a relative standard deviation of absorbance greater than 30% in the quality control QC sample or other coal groups is deleted to obtain a data matrix with the horizontal and vertical coordinates being the wave number and the sample name, respectively;

[0011] Step 4: After importing the data matrix, multivariate analysis was performed using principal component analysis (PCA), cluster analysis, and orthogonal partial least squares discriminant analysis (OPLS-DA) in turn to screen wavenumbers with significantly different absorbance as wavenumber markers to distinguish specific coal groups from other coal groups.

[0012] Furthermore, in the step 1, according to the requirements of the national standard "Method for preparing coal samples" (GB 474-2008), a KERP-180150B sealed hammer crusher, a ST-F200 automatic sampler and a 0.2 mm standard sieve are used to prepare the coal sample.

[0013] Furthermore, in step 2, the scanning mode, scanning range and spectral resolution of the Fourier transform infrared spectrometer are set to diffuse reflection, 400-4000cm -1 and 4cm -1 .

[0014] Compared with the prior art, the present invention adopts a metabolomics analysis strategy to process infrared spectra, and then determines which wavenumbers have significantly different absorbances, and successfully screens out the wavenumber markers of each coal group. Finally, by finding wavenumber markers, a novel analysis method based on a metabolomics analysis strategy to explore infrared spectral information is provided, thereby achieving the purpose of completely distinguishing coal from different geographical origins. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 Infrared spectra of 8 coal groups.

[0016] Figure 2 This is the PCA score plot for the eight coal groups.

[0017] Figure 3 This is a cluster analysis diagram of 8 coal groups.

[0018] Figure 4 OPLS-DA score chart for 8 coal groups.

[0019] Figure 5 S diagrams for eight coal groups.

[0020] Figure 6 This is the displacement diagram for the eight coal groups. DETAILED DESCRIPTION

[0021] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with embodiments. The specific embodiments described herein are only used to explain the present invention and are not used to limit the invention.

[0022] In this example, coal from 8 different locations in 3 countries, including Indonesia, Russia and the Philippines, was selected as the research object, as shown in Table 1. The detailed information of these coals is shown in Table 1. The coal was collected and stored in sealed plastic barrels to prevent moisture changes. According to the requirements of the national standard "Preparation Method of Coal Samples" (GB 474-2008), KERP-180150B sealed hammer crusher, ST-F200 automatic sampler and 0.2 mm standard sieve were used to prepare coal samples with a particle size of less than 0.2 mm.

[0023] Table 1

[0024]

[0025] Then, the coal sample spectra were recorded using a Shimadzu Fourier transform infrared (FTIR) spectrometer (IRTracer-100) in an air drying chamber with the temperature and humidity set to 20 °C and 10%, respectively. The scanning mode, scanning range, and spectral resolution were set to diffuse reflectance, 400–4000 cm -1 and 4cm -1According to the requirements of metabolomics analysis, each coal sample was prepared in parallel for 9 times, named as sample 1-1 to sample 1-9, sample 2-1 to sample 2-9, ... and sample 8-1 to sample 8-9, corresponding to No. 1, No. 2, ... and No. 8 coal groups. Equal amounts of all the above coal samples were mixed to prepare quality control QC samples for investigating the stability of the infrared spectrometer. Each group of quality control QC samples was scanned 3 times, and the remaining coal samples were scanned only once to obtain the original spectrum of the coal samples.

[0026] Each derived infrared spectrum can be expressed by a series of points on the coordinate axis, where the horizontal and vertical axes are wave number and absorbance, respectively. To facilitate spectral analysis and comparison, the entire infrared spectrum is translated along the Y axis until the lowest point of the spectrum reaches the X axis, at which point the absorbance is zero, such as Figure 1 shown.

[0027] According to the requirements of metabolomics analysis, wavenumbers with a relative standard deviation of absorbance greater than 30% in the quality control QC sample group or each coal group are considered invalid and should be deleted. Afterwards, a data matrix with the horizontal and vertical coordinates of wavenumber and sample name (e.g., sample 1-1) was obtained. The purpose of this study was to screen wavenumbers with significantly different absorbance as wavenumber markers to distinguish specific coal groups from other coal groups. After importing the data matrix, principal component analysis (PCA), cluster analysis, and orthogonal partial least squares discriminant analysis (OPLS-DA) were used for multivariate analysis.

[0028] As an unsupervised analysis method, PCA aims to reduce dimension and classify. It is used in metabolomics to evaluate data quality and identify outliers without knowing the sample category. Cluster analysis is another unsupervised method. Through this method, samples are gradually clustered according to the similarity of quality features. The samples with the highest similarity are clustered first. According to the comprehensive property similarity, all samples are clustered until the entire cluster analysis is completed. Figure 2 As shown, the credibility of all coal groups is within the 95% confidence level, and no obvious extreme data and outliers are observed. The quality control QC sample groups are well clustered together, indicating that the data obtained in this study is of high quality and worthy of further analysis. For groups No. 1, 2, 4, 5 and 8 of coal, each group of samples is clustered together separately and well separated from other groups, indicating that the samples within the group have high similarity, but the differences between the samples between the groups are obvious, which lays the foundation for further seeking "markers" between specific coal groups and other coal groups. However, for groups No. 3, 6 and 7, they intersect with each other, indicating that there is a high similarity between groups, which needs to be further distinguished by supervised analysis methods.

[0029] Different from PCA and cluster analysis, OPLS-DA is a supervised analysis method that can design sample groupings and effectively separate sample groups that cannot be separated by PCA and cluster analysis, thus better obtaining information on inter-group differences. Figure 3 As shown in the figure, groups No.2, 5 and 8 are all 9 parallel samples clustered together, indicating that the samples in this group have high similarity in comprehensive properties, while group No.6 is divided into 3 subgroups, including (a) sample 6-4 and sample 6-9; (b) sample 6-1, sample 6-2 and sample 6-7; (c) sample 6-3, sample 6-5, sample 6-6 and sample 6-8, indicating that the similarity of this group of samples is relatively low. The samples of groups No.1 and No.4, as well as groups No.3 and No.7 are intertwined, which means that the two groups have high inter-group similarity. In general, the cluster analysis results are basically consistent with the PCA results, indicating that both methods are reliable and can be used for coal origin tracing.

[0030] Other parameters embedded in OPLS-DA, including permutation test, S-chart and variable projection importance (VIP), are used for wave number marker screening. Among them, permutation test is used to determine whether the OPLS-DA model is overfitting. The S-chart consists of a series of points representing the wave number, and its importance depends on the position of the point. The points at both ends of the S-chart play the most important role in distinguishing the two camps and are also the most likely to become "marker" candidates. VIP can evaluate the contribution of candidate wave number markers by numerical value. The greater the contribution, the higher the VIP value. VIP>1 is usually regarded as the threshold of candidate wave number markers. The paired t-test in univariate analysis was performed in SPSSStatistics V17.0 software to finally confirm the validity of the wave number marker.

[0031] like Figure 4 As shown in the OPLS-DA score diagram, all coal groups can be clearly divided into two camps on the first principal component axis. The ones representing specific coal groups are marked in green, and the other groups are marked in blue, indicating that there are indeed wave numbers with significant differences in absorbance in the two camps. It is worth mentioning that Figure 2 The No. 3, 6, and 7 coal groups that were not fully distinguished by PCA can also be well separated, indicating that OPLS-DA has a greater advantage in distinguishing samples between groups.

[0032] exist Figure 5 In the S-diagram, each point represents a wave number, and the X-axis and Y-axis represent the contribution and confidence of the wave number, respectively. Points along the X-axis and Y-axis that are far from the origin indicate that these wave numbers have a greater contribution and higher confidence in distinguishing the two camps. Therefore, the points at both ends of the S-diagram are considered to be the most different components and the most suitable candidate wave number markers. The OPLS-DA model was examined for overfitting by 200 iterative permutation tests, where R2 Y and Q 2 are commonly used parameters that describe the model’s explanatory level and prediction level along the Y axis. 2 Y and Q 2 The closer the value is to 1, the less overfitting the OPLS-DA model has. Figure 6 It can be seen that R 2 Y and Q 2 The values ​​were all greater than 0.85, indicating that all OPLS-DA models were not overfitted and had good reliability and predictability. The VIP>1 principle was used to continue screening wavenumber markers. As shown in Table 2, 58, 40, 51, 17, 17, 26, 41 and 23 wavenumber markers were identified in groups No. 1 to 8, respectively, including 3593.384-3703.326, 2167.989-2243.213, 989.482-1085.923, 1697.358-1735.934, 1083.994-1114.855, 530.424-993.340, 2902.868-2980.020 and 1521.835-1564.269 cm -1 They are the wave number marker bands corresponding to the coal groups.

[0033] Table 2

[0034]

[0035]

[0036] A paired t-test was performed on the absorbance of the wave number marker between a specific coal group and other groups, and a significant difference was found (P<0.05), which further confirmed the effectiveness of the "marker". Significantly high absorbance can be observed on the wave number marker band of the No. 2, 3, 4 and 8 coal groups, but significantly low absorbance is observed on the wave number marker band of the No. 1, 5, 6 and 7 coal groups. Through the verification of the above examples, the present invention can completely distinguish coals from different geographical sources.

[0037] The technical solution of the present invention is not limited to the above-mentioned specific embodiments. All technical variations made according to the technical solution of the present invention fall within the protection scope of the present invention.

Claims

1. A method for tracing the origin of coal by combining Fourier transform infrared spectroscopy with metabolomics analysis, characterized in that: The following steps are involved: Step 1: Collect coal to be analyzed from n different locations and store them in sealed plastic barrels, and then prepare coal samples with a particle size less than 0.2 mm respectively; Step 2, select 9 parallel samples from each coal sample, name them as sample 1-1 to sample 1-9, sample 2-1 to sample 2-9, ... and sample n-1 to sample n-9, corresponding to No. 1, No. 2, ... and No. n coal groups; in an air drying room where the temperature and humidity are set to 20°C and 10%, respectively, use a Fourier transform infrared spectrometer to record the infrared spectra of the 9 parallel samples of each coal sample, mix equal amounts of the 9 parallel samples of each coal sample and prepare quality control QC samples, use the Fourier transform infrared spectrometer to repeatedly scan the quality control QC samples 3 times to obtain 3 infrared spectra of the quality control QC samples, and scan the remaining coal samples once using the Fourier transform infrared spectrometer to obtain the original spectra of the coal samples; Step 3, each derived infrared spectrum is expressed by a series of points on the coordinate axis, wherein the horizontal and vertical coordinates of the coordinate axis are wave number and absorbance, respectively, and the entire infrared spectrum is translated along the Y axis until the lowest point of the spectrum reaches the X axis, at which time the absorbance of the lowest point is zero; the wave number with a relative standard deviation of absorbance greater than 30% in the quality control QC sample or other coal groups is deleted to obtain a data matrix with the horizontal and vertical coordinates being the wave number and the sample name, respectively; Step 4: After importing the data matrix, multivariate analysis was performed using principal component analysis (PCA), cluster analysis, and orthogonal partial least squares discriminant analysis (OPLS-DA) in order to screen wavenumbers with significantly different absorbance as wavenumber markers to distinguish specific coal groups from other coal groups; In step 2, the scanning mode, scanning range and spectral resolution of the Fourier transform infrared spectrometer are set to diffuse reflection, 400-4000cm -1 and 4cm -1 .

2. The method for tracing the origin of coal by combining Fourier transform infrared spectroscopy with metabolomics analysis according to claim 1, characterized in that: In the step 1, according to the requirements of the national standard "GB 474-2008 Preparation method of coal samples", a KERP-180150B sealed hammer crusher, a ST-F200 automatic sampler and a 0.2 mm standard sieve are used to prepare the coal sample.

Images