Tobacco leaf raw material quality characteristic analysis method based on data fusion

Abstract

The invention relates to the technical field of tobacco leaf raw material quality characteristic analysis, in particular to a tobacco leaf raw material quality characteristic analysis method based on data fusion, which comprises the following steps: obtaining related indication data of tobacco leaves from different types of tobacco leaf samples, and converting the related indication data of the tobacco leaves into a related indication data matrix of the tobacco leaves; measuring infrared spectrum data of different types of tobacco leaf samples, preprocessing the measured infrared spectrum data of the tobacco leaf samples, performing principal component analysis on the preprocessed infrared spectrum data of the tobacco leaf samples, determining a principal component number k, and obtaining score P data and load V data; combining the score P data matrix with each related indication data matrix of the tobacco leaves to obtain a combined tobacco leaf quality data matrix D, and normalizing the tobacco leaf quality data matrix D to obtain a combined tobacco leaf quality data matrix X; and carrying out clustering analysis processing on the fused tobacco leaf quality data matrix X, and analyzing quality characteristics of different types of samples.

Description

Technical Field

[0001] The present invention relates to the technical field of tobacco leaf raw material quality characteristic analysis, and in particular to a tobacco leaf raw material quality characteristic analysis method based on data fusion. Background Art

[0002] In the field of industrial production, the stability of production materials, products and production processes plays a very important role. Among them, the similarity between production materials is equally important and has great significance for maintaining the stability of products. In the production of tobacco products, the quality or grade of tobacco raw materials is usually assessed by indicator data such as appearance, sensory, conventional components and trace chemical components. However, due to the one-sidedness or subjectivity of these data, it is often difficult to obtain satisfactory judgment results.

[0003] Spectroscopic technology plays an increasingly important role in the quality monitoring of products or raw materials. In particular, in recent years, a lot of research has been carried out on the use of near-infrared spectroscopy technology to evaluate the quality of tobacco raw materials. Specifically, the content of the main chemical components is first detected through the quantitative model of the spectrum, and then the quality of the tobacco raw materials is evaluated through the content of various components, or the chemometric method is directly used to model the spectrum of the tobacco raw materials to quantitatively describe the quality characteristics of the tobacco raw materials, so as to evaluate the quality of the tobacco raw materials based on the quality characteristics of the tobacco raw materials.

[0004] Although the above-mentioned near-infrared spectroscopy technology can be used to evaluate the quality of tobacco raw materials, this method has the following defects: due to the similarities in the spectra of different categories of tobacco raw materials, it is difficult to obtain satisfactory results using spectral analysis.

[0005] Therefore, it is necessary to improve the method of evaluating the quality of tobacco raw materials in order to solve the problem that different types of tobacco raw materials have similarities in spectra and it is difficult to obtain satisfactory results using spectral analysis when evaluating the quality of tobacco raw materials using infrared spectroscopy technology. Summary of the invention

[0006] The purpose of the present invention is to propose a tobacco raw material quality feature analysis method based on data fusion, utilize data from different sources, adopt data fusion technology to fuse different data, and then perform cluster analysis on the tobacco raw materials, make full use of the data describing the different characteristics of the tobacco raw materials, improve the accuracy and objectivity of the tobacco raw material feature analysis, so as to solve the problem that in the quality evaluation of tobacco raw materials using infrared spectroscopy technology, different categories of tobacco raw materials have similarities in spectra, and it is difficult to obtain satisfactory results using spectral analysis.

[0007] In order to solve the above technical problems, the present invention adopts the following technical solutions:

[0008] The tobacco leaf raw material quality characteristic analysis method based on data fusion includes the following steps:

[0009] Step 1: Obtain tobacco leaf related indicator data from tobacco leaf samples of different categories, and convert the tobacco leaf related indicator data into tobacco leaf related indicator data matrix;

[0010] Step 2: Measure the infrared spectrum data of tobacco samples of different categories, pre-process the measured infrared spectrum data of tobacco samples, and then perform principal component analysis on the pre-processed infrared spectrum data of tobacco samples to determine the number of principal components k and obtain the score P data and the load V data;

[0011] Step 3: Merge the score P data matrix in Step 2 with the tobacco leaf related indicator data matrix in Step 1 to obtain a merged tobacco leaf quality data matrix D, and normalize the tobacco leaf quality data matrix D to obtain a fused tobacco leaf quality data matrix X;

[0012] Step 4: Perform cluster analysis on the fused tobacco leaf quality data matrix X, and analyze the quality characteristics of samples of different categories from the tobacco leaf quality data matrix X in the cluster analysis process.

[0013] Preferably, the Step 1: obtaining tobacco leaf related indicator data from tobacco leaf samples of different categories, and converting the tobacco leaf related indicator data into a tobacco leaf related indicator data matrix, comprises the following steps:

[0014] Step 1.1: Collect different types of tobacco raw material samples from various tobacco raw material producing areas, and obtain tobacco related indicator data such as appearance data, sensory data, conventional component data and aroma component data from the collected different types of tobacco raw material samples;

[0015] Step 1.2: Convert the acquired appearance data, sensory data, conventional component data, aroma component data and other tobacco leaf related indicator data into appearance data matrix D 1 , sensory data matrix D 2 , conventional component data matrix D 3 And the aroma component data matrix D 4 .

[0016] Preferably, the Step 1.1: the tobacco leaf related indicator data are any one or more of appearance data, sensory data, conventional component data and aroma component data, or are other data of tobacco leaf related indicators.

[0017] Preferably, the Step 1.1: the number of tobacco leaf raw material samples of different categories collected is infinite, and the proportion of each type of tobacco leaf raw material samples collected is balanced to ensure the representativeness of the samples.

[0018] Preferably, the Step 2: measuring infrared spectrum data of tobacco leaf samples of different categories, first preprocessing the measured infrared spectrum data of tobacco leaf samples, and then performing principal component analysis on the preprocessed infrared spectrum data of tobacco leaf samples to determine the number of principal components k, and obtain score P data and load V data, includes the following steps:

[0019] Step 2.1: Using near infrared spectroscopy technology, different types of tobacco leaf samples are processed to obtain infrared spectral data of different types of tobacco leaf samples;

[0020] Step 2.2: Using wavelet transform technology or other spectral signal processing technology, the infrared spectrum data of different categories of tobacco leaf samples are processed by wavelet transform to obtain the infrared spectrum data of different categories of tobacco leaf samples after preprocessing;

[0021] Step 2.3: Perform principal component analysis on the infrared spectrum data of the pretreated tobacco leaf samples, determine the number of principal components k, and obtain the score P data and the load V data;

[0022] Preferably, the Step 2.1: using near-infrared spectroscopy technology to process different categories of tobacco leaf samples to obtain infrared spectral data of different categories of tobacco leaf samples, or using Raman spectroscopy technology to process different categories of tobacco leaf samples to obtain Raman spectral data of different categories of tobacco leaf samples.

[0023] Preferably, the Step 3: merging the score P data matrix in Step 2 with the tobacco leaf related indicator data matrices in Step 1 to obtain a merged tobacco leaf quality data matrix D, and normalizing the tobacco leaf quality data matrix D to obtain a fused tobacco leaf quality data matrix X, comprises the following steps:

[0024] Step 3.1: Compare the score data matrix P in Step 2 with the appearance data matrix D in Step 1 1 , sensory data matrix D 2 , conventional component data matrix D 3 And the aroma component data matrix D 4 Fusion, get the combined tobacco leaf quality data matrix D, its expression is:

[0025] D=[D 1 D 2 D 3 D 4 P]

[0026] Among them, D 1 is the appearance data matrix; D 2 is the sensory data matrix; D 3 is the regular component matrix; D4 Aroma component data matrix; P is the score P data matrix.

[0027] Step 3.2: Normalize the merged tobacco leaf quality data matrix D to obtain the fused tobacco leaf quality data matrix X.

[0028] Preferably, the Step 3.2: normalizing the merged tobacco leaf quality data matrix D to obtain the fused tobacco leaf quality data matrix X, or performing any one of centering, standardization, vector normalization, area normalization, maximum normalization and average normalization on the merged tobacco leaf quality data matrix D to obtain the fused tobacco leaf quality data matrix X.

[0029] Preferably, the Step 4: cluster analysis is performed on the fused tobacco leaf quality data matrix X to analyze the quality characteristics of samples of different categories, as follows:

[0030] Cluster analysis technology is used to perform cluster analysis on the fused tobacco leaf quality data matrix X, and the quality characteristics of samples of different categories are analyzed from the tobacco leaf quality data matrix X in the cluster analysis process, and the analyzed quality characteristics of samples of different categories are extracted.

[0031] Preferably, the cluster analysis technology is used to perform cluster analysis on the fused tobacco leaf quality data matrix X, and the quality characteristics of samples of different categories are analyzed from the tobacco leaf quality data matrix X in the cluster analysis process, and the analyzed quality characteristics of samples of different categories are extracted. Alternatively, any one of system clustering, k-NN, C-means and fuzzy clustering is used to perform cluster analysis on the fused tobacco leaf quality data matrix X, and the quality characteristics of samples of different categories are analyzed from the tobacco leaf quality data matrix X in the cluster analysis process, and the analyzed quality characteristics of samples of different categories are extracted.

[0032] Compared with the prior art, the present invention has at least one of the following beneficial effects:

[0033] 1. The present invention makes full use of data from different sources, then uses data fusion technology to fuse different data, and then performs cluster analysis on tobacco raw materials. It not only makes full use of the data describing different characteristics of tobacco raw materials, but also improves the accuracy and objectivity of characteristic analysis of tobacco raw materials, so as to solve the problem that different categories of tobacco raw materials have similarities in spectra and it is difficult to obtain satisfactory results when using spectral analysis in quality evaluation of tobacco raw materials using infrared spectroscopy technology.

[0034] 2. The calculation method in the present invention only involves parameters related to principal component analysis and clustering method, and the parameter values ​​have little effect on the calculation results, making the analysis results more accurate.

[0035] 3. The tobacco raw material quality characteristic analysis method based on data fusion in the present invention has diverse data sources, reliable classification results, simple principles, easy to understand, simple algorithms, easy programming and calculation, simple parameter settings and strong practicality. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 The figure is a flow chart of the tobacco leaf raw material quality characteristic analysis method based on data fusion in the present invention.

[0037] Figure 2 It is the near infrared spectra of different types of tobacco leaf raw material samples in the present invention.

[0038] Figure 3 It is the near infrared spectra of different types of tobacco raw material samples after wavelet transform processing in the present invention.

[0039] Figure 4 It is the intensity distribution diagram of 125 parameters after the tobacco quality data matrix D is normalized in the present invention.

[0040] Figure 5 This is the result of the C-means cluster analysis of the fused tobacco leaf quality data matrix in the present invention.

[0041] Table 1 is a table of quality characteristics analysis of different types of tobacco samples in the present invention DETAILED DESCRIPTION

[0042] like Figure 1-5 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 the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0043] Example

[0044] Spectroscopic technology plays an increasingly important role in the quality monitoring of products or raw materials. In particular, in recent years, a lot of research has been carried out on the use of near-infrared spectroscopy technology to evaluate the quality of tobacco raw materials. Specifically, the content of the main chemical components is first detected through the quantitative model of the spectrum, and then the quality of the tobacco raw materials is evaluated through the content of various components, or the chemometric method is directly used to model the spectrum of the tobacco raw materials to quantitatively describe the quality characteristics of the tobacco raw materials, so as to evaluate the quality of the tobacco raw materials based on the quality characteristics of the tobacco raw materials.

[0045] Although the above-mentioned near-infrared spectroscopy technology can be used to evaluate the quality of tobacco raw materials, this method has the following defects: due to the similarities in the spectra of different categories of tobacco raw materials, it is difficult to obtain satisfactory results using spectral analysis.

[0046] Therefore, this application proposes a tobacco raw material quality feature analysis method based on data fusion to solve the problem that different types of tobacco raw materials have spectral similarities and it is difficult to obtain satisfactory results when using spectral analysis when evaluating the quality of tobacco raw materials using infrared spectroscopy technology.

[0047] For details, please refer to Figure 1 , a tobacco leaf raw material quality characteristic analysis method based on data fusion includes the following steps:

[0048] Step 1: Obtain tobacco leaf related indicator data from tobacco leaf samples of different categories, and convert the tobacco leaf related indicator data into tobacco leaf related indicator data matrix. The specific steps are as follows:

[0049] 1) Collect at least 93 tobacco raw material samples of different categories from each tobacco raw material producing area, among which the tobacco raw material samples of different categories may be 32 upper (including top leaves), 21 lower (including foot leaves), and 37 middle tobacco samples, and then obtain the appearance data, sensory data, conventional component data, aroma component data and other tobacco related indicator data from 93 different categories of tobacco raw material samples, including 32 upper (including top leaves), 21 lower (including foot leaves) and 37 middle tobacco samples;

[0050] 2) Convert the acquired appearance data, sensory data, conventional component data, aroma component data and other tobacco leaf related indicator data into appearance data matrix D 1 , sensory data matrix D 2 , conventional component data matrix D 3 And the aroma component data matrix D 4 , where the appearance data matrix D 1 For 7, sensory data matrix D 2 The regular component data matrix D is 20 3 is the data matrix D of the six aroma components 4 There are 72 of them.

[0051] In this example, the obtained tobacco leaf related indicator data may be any one or more of appearance data, sensory data, conventional component data and aroma component data, or other data of tobacco leaf related indicators.

[0052] Step 2: Measure the infrared spectrum data of tobacco leaf samples of different categories, pre-process the measured infrared spectrum data of tobacco leaf samples, and then perform principal component analysis on the pre-processed infrared spectrum data of tobacco leaf samples to determine the number of principal components k and obtain the score P data and load V data. The specific steps are as follows:

[0053] 1) Using near infrared spectroscopy technology, 93 tobacco leaf samples of different categories were processed to obtain infrared spectral data of tobacco leaf samples of different categories, such as Figure 2 As shown, the wavelength range of infrared spectrum data is 12000–4000cm -1 , the number of data points is 2074;

[0054] 2) Using wavelet transform technology or other spectral signal processing technology, the infrared spectrum data of 93 different categories of tobacco leaf samples are processed by wavelet transform to obtain the infrared spectrum data of different categories of tobacco leaf samples after preprocessing, such as Figure 3 As shown;

[0055] 3) Perform principal component analysis on the infrared spectrum data of the pretreated tobacco leaf samples, determine the number of principal components k=20, and obtain the score P data and load V data of the 20 principal components.

[0056] In this embodiment, in addition to using near-infrared spectroscopy technology to process different categories of tobacco leaf samples and obtain infrared spectral data of different categories of tobacco leaf samples, Raman spectroscopy technology can also be used to process different categories of tobacco leaf samples and obtain Raman spectral data of different categories of tobacco leaf samples.

[0057] Step 3: Merge the score P data matrix in Step 2 with the tobacco leaf related indicator data matrix in Step 1 to obtain the merged tobacco leaf quality data matrix D. Normalize the tobacco leaf quality data matrix D to obtain the fused tobacco leaf quality data matrix X. The specific steps are as follows:

[0058] 1) Compare the score P data matrix in Step 2 with the appearance data matrix D in Step 1 1 , sensory data matrix D 2 , conventional component data matrix D 3 And the aroma component data matrix D 4 Fusion, get the combined tobacco leaf quality data matrix D, its expression is:

[0059] D=[D 1 D 2 D 3 D 4 P]

[0060] Among them, D 1 is the appearance data matrix; D 2is the sensory data matrix; D 3 is the regular component matrix; D 4 Aroma component data matrix; P is the score P data matrix.

[0061] 2) The combined tobacco leaf quality data matrix D is normalized to obtain the fused tobacco leaf quality data matrix X, where the fused tobacco leaf quality data parameter distribution is as follows: Figure 4 shown.

[0062] In an embodiment, the merged tobacco leaf quality data matrix D is normalized to obtain a fused tobacco leaf quality data matrix X. Alternatively, the merged tobacco leaf quality data matrix D is subjected to any one of centralization, standardization, vector normalization, area normalization, maximum normalization and average normalization to obtain a fused tobacco leaf quality data matrix X.

[0063] Step 4: Perform cluster analysis on the fused tobacco leaf quality data matrix X to analyze the quality characteristics of samples of different categories, as follows:

[0064] Cluster analysis technology is used to perform cluster analysis on the fused tobacco leaf quality data matrix X. The cluster analysis sets the number of categories to 5, and the following is obtained: Figure 4 The results shown are then used to analyze the quality features of samples of different categories from the tobacco leaf quality data matrix X in the cluster analysis process, and extract the analyzed quality features of samples of different categories.

[0065] Specifically, according to Figure 4 The results shown are saved as a table file as shown in Table 1, and feature analysis is performed in combination with the appearance, sensory, conventional ingredients, aroma ingredients and spectral score data of the samples; among them, Table 1 lists the clustering results of 82 samples of known parts.

[0066] Table 1 Analysis of quality characteristics of different types of tobacco leaf samples

[0067] category Upper part (parietal lobe) Central Lower part (leaf of foot) feature 1 3 3 0 ripe 2 1 9 15 Orange, lemon yellow, mature 3 16 3 0 Orange, ripe, moderately ripe 4 6 8 1 Variegated, slightly green, immature 5 1 13 3 Orange, lemon yellow, mature

[0068] From Table 1, we can clearly see that:

[0069] 1) The category is highly correlated with the "part". The third category is mainly composed of upper tobacco leaves, the fifth category is mainly composed of middle tobacco leaves, and the second category is mainly composed of lower tobacco leaves, but a part of it is middle tobacco leaves.

[0070] 2) The category is highly correlated with "maturity". "Well-ripened", "moderately ripe" and "unripe" correspond to categories 1, 3 and 4 respectively. "Mature" tobacco leaves are divided into two categories, namely category 5 which is mainly composed of middle tobacco leaves and category 2 which is mainly composed of middle and lower tobacco leaves.

[0071] In this embodiment, in addition to using cluster analysis technology to perform cluster analysis on the fused tobacco leaf quality data matrix X, analyzing the quality characteristics of samples of different categories from the tobacco leaf quality data matrix X in the cluster analysis process, and extracting the analyzed quality characteristics of samples of different categories, any one of system clustering, k-NN, C-means and fuzzy clustering can also be used to perform cluster analysis on the fused tobacco leaf quality data matrix X, analyze the quality characteristics of samples of different categories from the tobacco leaf quality data matrix X in the cluster analysis process, and extract the analyzed quality characteristics of samples of different categories.

[0072] This application first uses data from different sources, then uses data fusion technology to fuse different data, and then performs cluster analysis on tobacco leaf raw materials, making full use of data describing different characteristics of tobacco leaf raw materials, and improving the accuracy and objectivity of tobacco leaf raw material characteristic analysis, so as to solve the problem of using infrared spectroscopy technology to evaluate the quality of tobacco leaf raw materials. Due to the similarities in the spectra of different categories of tobacco leaf raw materials, it is difficult to obtain satisfactory results using spectral analysis.

[0073] Although the present invention is described herein with reference to a number of illustrative embodiments of the present invention, it will be appreciated that those skilled in the art may devise many other modifications and implementations that fall within the scope and spirit of the principles disclosed herein. More specifically, within the scope of the present disclosure, drawings, and claims, a variety of variations and modifications may be made to the components and / or layout of the subject combination layout. In addition to the variations and modifications made to the components and / or layout, other uses will also be apparent to those skilled in the art.

Claims

1. A tobacco leaf raw material quality characteristic analysis method based on data fusion, characterized in that: The following steps are involved: Step 1: Obtain tobacco leaf related indicator data from tobacco leaf samples of different categories, and convert the tobacco leaf related indicator data into tobacco leaf related indicator data matrix; Step 2: Measure the infrared spectrum data of tobacco samples of different categories, pre-process the measured infrared spectrum data of tobacco samples, and then perform principal component analysis on the pre-processed infrared spectrum data of tobacco samples to determine the number of principal components k and obtain the score P data and the load V data; Step 3: Merge the score P data matrix in Step 2 with the tobacco leaf related indicator data matrix in Step 1 to obtain a merged tobacco leaf quality data matrix D, and normalize the tobacco leaf quality data matrix D to obtain a fused tobacco leaf quality data matrix X; Step 4: Perform cluster analysis on the fused tobacco leaf quality data matrix X to analyze the quality characteristics of samples of different categories.

2. The tobacco leaf raw material quality characteristic analysis method based on data fusion according to claim 1, characterized in that: The Step 1: obtaining tobacco leaf related indicator data from tobacco leaf samples of different categories, and converting the tobacco leaf related indicator data into a tobacco leaf related indicator data matrix, comprises the following steps: Step 1.1: Collect different types of tobacco raw material samples from various tobacco raw material producing areas, and obtain tobacco related indicator data such as appearance data, sensory data, conventional component data and aroma component data from the collected different types of tobacco raw material samples; Step 1.2: Convert the acquired appearance data, sensory data, conventional component data, aroma component data and other tobacco leaf related indicator data into appearance data matrix D1, sensory data matrix D2, conventional component data matrix D3 and aroma component data matrix D4 respectively.

3. The tobacco leaf raw material quality characteristic analysis method based on data fusion according to claim 2, characterized in that: The Step 1.1: the tobacco leaf related indicator data are any one or more of appearance data, sensory data, conventional component data and aroma component data, or other data of tobacco leaf related indicators.

4. The method for analyzing tobacco leaf raw material quality characteristics based on data fusion according to claim 2, characterized in that: Step 1.1: The number of tobacco leaf raw material samples of different categories collected is infinite, and the proportion of each type of tobacco leaf raw material samples collected is balanced to ensure the representativeness of the samples.

5. The method for analyzing tobacco leaf raw material quality characteristics based on data fusion according to claim 1, characterized in that: The Step 2: measuring infrared spectrum data of tobacco leaf samples of different categories, first pre-processing the measured infrared spectrum data of tobacco leaf samples, and then performing principal component analysis on the pre-processed infrared spectrum data of tobacco leaf samples to determine the number of principal components k, and obtain score P data and load V data, includes the following steps: Step 2.1: Using near infrared spectroscopy technology, different types of tobacco leaf samples are processed to obtain infrared spectral data of different types of tobacco leaf samples; Step 2.2: Using wavelet transform technology or other spectral signal processing technology, the infrared spectrum data of different categories of tobacco leaf samples are processed by wavelet transform to obtain the infrared spectrum data of different categories of tobacco leaf samples after preprocessing; Step 2.3: Perform principal component analysis on the infrared spectrum data of the pretreated tobacco leaf samples, determine the number of principal components k, and obtain the score P data and load V data.

6. The method for analyzing tobacco leaf raw material quality characteristics based on data fusion according to claim 5, characterized in that: The Step 2.1: using near infrared spectroscopy technology to process different categories of tobacco leaf samples to obtain infrared spectral data of different categories of tobacco leaf samples, or using Raman spectroscopy technology to process different categories of tobacco leaf samples to obtain Raman spectral data of different categories of tobacco leaf samples.

7. The method for analyzing tobacco leaf raw material quality characteristics based on data fusion according to claim 1, characterized in that: The step 3: merging the score P data matrix in step 2 with the tobacco leaf related indicator data matrix in step 1 to obtain a merged tobacco leaf quality data matrix D, and normalizing the tobacco leaf quality data matrix D to obtain a fused tobacco leaf quality data matrix X, comprises the following steps: Step 3.1: Merge the score P data matrix in Step 2 with the appearance data matrix D1, sensory data matrix D2, conventional component data matrix D3 and aroma component data matrix D4 in Step 1 to obtain the combined tobacco leaf quality data matrix D, which is expressed as: D=[D1 D2 D3 D4 P] Among them, D1 is the appearance data matrix; D2 is the sensory data matrix; D3 is the conventional component number matrix; D4 is the aroma component data matrix; and P is the score P data matrix. Step 3.2: Normalize the merged tobacco leaf quality data matrix D to obtain the fused tobacco leaf quality data matrix X.

8. The method for analyzing tobacco leaf raw material quality characteristics based on data fusion according to claim 7, characterized in that: Said Step 3.2: Normalize the merged tobacco leaf quality data matrix D to obtain the fused tobacco leaf quality data matrix X, or perform any of the following processes on the merged tobacco leaf quality data matrix D: centering, standardization, vector normalization, area normalization, maximum normalization, and average normalization to obtain the fused tobacco leaf quality data matrix X.

9. The method for analyzing tobacco leaf raw material quality characteristics based on data fusion according to claim 1, characterized in that: Step 4: cluster analysis is performed on the fused tobacco leaf quality data matrix X to analyze the quality characteristics of samples of different categories, as follows: Cluster analysis technology is used to perform cluster analysis on the fused tobacco leaf quality data matrix X, and the quality characteristics of samples of different categories are analyzed from the tobacco leaf quality data matrix X in the cluster analysis process, and the analyzed quality characteristics of samples of different categories are extracted.

10. The tobacco leaf raw material quality characteristic analysis method based on data fusion according to claim 1, characterized in that: The cluster analysis technology is used to perform cluster analysis on the fused tobacco leaf quality data matrix X, analyze the quality characteristics of samples of different categories from the tobacco leaf quality data matrix X in the cluster analysis process, and extract the analyzed quality characteristics of samples of different categories. Alternatively, any one of system clustering, k-NN, C-means and fuzzy clustering is used to perform cluster analysis on the fused tobacco leaf quality data matrix X, analyze the quality characteristics of samples of different categories from the tobacco leaf quality data matrix X in the cluster analysis process, and extract the analyzed quality characteristics of samples of different categories.

Images