Poria cocos producing area identification method based on Poria cocos Raman spectrum principal component analysis

Abstract

The invention provides a poria cocos producing area identification method based on poria cocos Raman spectrum principal component analysis. The method comprises the following steps: firstly, acquiring Raman spectrums of poria cocos from different producing areas by using a Raman spectrometer, carrying out pretreatment such as averaging, base line removal and feature screening on the original spectrums, and constructing a spectrum data set and a test set of the poria cocos from different producing areas; then, principal component analysis is carried out on the spectral data sets of the poria cocos from different producing areas, and principal components with the spectral cumulative contribution rate larger than 90% are reserved to be used for constructing a feature vector matrix; and finally, respectively performing projection reconstruction on a test spectrum on the constructed feature vector matrix to obtain recovery spectrums of the test spectrum in different producing areas, comparing Euclidean distances between the recovery spectrums in different producing areas and the test spectrum, and determining that the test spectrum belongs to the producing area corresponding to the recovery spectrum with the minimum Euclidean distance. The method is rapid, simple, convenient and high in accuracy, does not need a large number of samples for training, and has great application potential in the aspect of identifying the origin of poria cocos.

Description

Technical Field

[0001] The invention relates to a method for identifying the origin of Poria cocos based on principal component analysis of Poria cocos Raman spectrum, which can be used to identify the origin of Poria cocos from different origins and belongs to the technical field of Raman spectrum analysis and detection. Background Art

[0002] Raman spectroscopy is a spectral analysis technology based on light scattering to analyze the molecular structure and composition of substances. It can achieve non-destructive in-situ detection with high detection efficiency and low cost. It is widely used in polymer materials, biological molecular structures, medicines, foods, etc. The application of Raman spectroscopy in the identification of the origin of Poria cocos makes up for the shortcomings of terahertz, hyperspectral, high performance liquid chromatography, electronic nose technology, etc. in the identification of the origin of Poria cocos, such as high instrument cost, complex data processing, professional operation, complex sample pre-treatment and destructiveness to samples.

[0003] Principal component analysis (PCA) is a simple and powerful unsupervised machine learning technique that can effectively process Raman spectral matrix data, while reducing the dimensionality of the data set and retaining the most critical features of the data. Using principal component analysis, the main structural information of the data can be mined to a certain extent when the sample size is not particularly sufficient. In contrast, some complex machine learning algorithms, such as deep neural networks, usually require a large number of samples to effectively learn the patterns of the data. If the sample size is small, these algorithms may not converge well or may overfit the training data, and principal component analysis just makes up for this shortcoming. Therefore, when the sample size is small, the data restoration characteristics of principal component dimensionality reduction can be used to distinguish small differences in complex data, thereby achieving data classification.

[0004] Based on the above background technology, the present invention proposes a method for identifying the origin of Poria cocos based on principal component analysis of Poria cocos Raman spectra. After preprocessing and feature screening, a data set and a test set of Poria cocos Raman spectra from different origins are established. Useful spectral signals are screened based on the size of the cumulative contribution rate, and the principal component matrices of different origins are constructed. The origin of Poria cocos can be identified by comparing the Euclidean distance between the test spectrum and the restored spectrum after dimensionality reduction and reconstruction on different principal component matrices. Summary of the invention

[0005] The purpose of the present invention is to provide a method for identifying the origin of Poria cocos based on principal component analysis of Poria cocos Raman spectrum.

[0006] To achieve the above purpose, the technical solutions adopted by the present invention are as follows:

[0007] Data preprocessing stage: The experimentally measured Raman spectra usually contain a high fluorescence background. Before performing principal component analysis, the original Raman spectra are averaged three times, baseline removed, and characteristic bands screened. This can effectively reduce the impact of random noise, improve data consistency, reduce data dimension and complexity, and focus on key information, thereby constructing an effective Poria cocos Raman spectrum data set and test set.

[0008] Principal component analysis stage: Principal component analysis was performed on the Raman spectrum data sets of Poria cocos from different origins, and the eigenvector matrices of different origins were established. Each eigenvector matrix represents the main characteristics and change laws of Poria cocos from its corresponding origin, and the eigenvector matrices are independent of each other.

[0009] Principal component screening stage: The threshold of the cumulative contribution rate is reasonably set according to different needs, and the principal component information greater than the threshold is used as the effective spectral information to construct the characteristic vector matrix of each origin.

[0010] Origin identification stage: a test spectrum is projected and reconstructed on the eigenvector matrices of different origins to obtain the restored spectra of the test spectrum in different origins. The Euclidean distances between the restored spectra of different origins and the test spectrum are compared. It can be considered that the origin of the test spectrum belongs to the origin corresponding to the restored spectrum with the smallest Euclidean distance.

[0011] It has been verified that when the threshold of the cumulative contribution rate of the principal component is set to 90%, the best classification effect is obtained. The method of combining Raman spectroscopy with principal component analysis to identify the origin of Poria cocos is fast and simple, with high accuracy, and does not require a large number of samples for training. It has great application potential in identifying the origin of Poria cocos. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 The present invention is a flowchart of a method for identifying the origin of Poria cocos based on principal component analysis of Poria cocos Raman spectrum according to an embodiment of the present invention.

[0013] Figure 2 The preprocessed Raman spectral dataset of Poria cocos from different origins was obtained for the experiment.

[0014] Figure 3 Comparison diagram of Yunnan test sample 1 and its restored spectrum reconstructed on different eigenvector matrices.

[0015] Figure 4 This is a comparison chart of the Euclidean distance between Yunnan test sample 1 and its different restored spectra. DETAILED DESCRIPTION

[0016] The technical solution of the present invention is further described below in conjunction with specific embodiments:

[0017] This embodiment:

[0018] The Poria cocos blocks used in the experiment were purchased from Chuxiong, Yunnan, China, Shangyu, Shaanxi, Yuexi, Anhui, and Shennongjia, Hubei. Considering the uneven size of the purchased Poria cocos blocks, the oxidation of the surface, and the contamination of other substances during the production, preparation and transportation process, in order to reduce the experimental error, the surface of the Poria cocos to be tested was cut off each time the data was collected, and the Raman spectrum data inside the Poria cocos block was collected. The wavelength of the excitation light of the Raman spectroscopy measurement equipment is 785nm, the laser power is about 800mW, and the integration time of each spectrum is 5s.

[0019] Step S1, scrape off the surface to be tested of the Poria cocos blocks from different origins, place them in a glass dish, and use a Raman spectrometer to collect Raman spectrum information inside the Poria cocos blocks from different origins.

[0020] Step S2, after preprocessing the collected original Raman spectra by averaging three times, removing the baseline and screening the characteristic bands, the Raman spectrum dataset and test set of Poria cocos from different origins are constructed. The constructed Raman spectrum datasets of Poria cocos from four origins are as follows: Figure 2 As shown, each origin includes 25 Raman spectra.

[0021] Step S3, principal component analysis is performed on the Poria cocos Raman spectrum data sets from different origins, and the eigenvalues ​​of each principal component are sorted, the contribution rate and cumulative contribution rate are calculated, and the appropriate number of principal components is selected to construct the eigenvector matrices of different origins.

[0022] Step S4, project and reconstruct a test spectrum, namely, Yunnan test sample 1, on the eigenvector matrix of different origins to obtain the restored spectrum of the test spectrum in different origins, such as Figure 3 As shown in the figure, the Euclidean distance between the recovered spectra from different origins and the test spectrum is compared, as shown in the figure. Figure 4 As shown, from Figure 4 It can be seen that the Euclidean distance between Yunnan test sample 1 and its restored spectrum in Yunnan is the smallest, and it can be determined that the origin of the test spectrum is Yunnan.

[0023] Step S5, filter the effective spectral signals according to the set threshold value. The threshold value can be changed according to different needs to achieve the best classification effect. The experiment retains the main components with a spectral cumulative contribution rate greater than 90% as the effective spectral information of Poria cocos from each origin to construct the characteristic vector matrix of each origin. At this time, the classification effect is optimal, with an accuracy rate of up to 97.5%.

[0024] The above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention is described in detail with reference to specific embodiments, a person skilled in the art should understand that the technical solutions of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solutions of the present invention, which should be included in the scope of the claims of the present invention. The technologies, shapes, and structural parts not described in detail in the present invention are all well-known technologies.

Claims

1. A method for identifying the origin of Poria cocos based on principal component analysis of Poria cocos Raman spectrum, characterized in that: The steps include: Step S1, scraping off the surface of the Poria cocos blocks from different origins, placing them in a glass dish, and collecting Raman spectrum information of the inside of the Poria cocos blocks from different origins using a Raman spectrometer; Step S2, constructing a Raman spectrum data set and a test set of Poria cocos from different origins after preprocessing the collected Raman spectra by averaging three times, removing the baseline, and screening characteristic bands; Step S3, performing principal component analysis on the Raman spectrum data sets of Poria cocos from different origins, and selecting the appropriate number of principal components according to the spectral contribution rate to construct a feature vector matrix; Step S4, projecting and reconstructing a test spectrum on the eigenvector matrices of different origins to obtain restored spectra of the test spectrum at different origins, and comparing the Euclidean distances between the restored spectra at different origins and the test spectrum, it can be considered that the origin of the test spectrum belongs to the origin corresponding to the restored spectrum with the smallest Euclidean distance; Step S5, retaining the principal components with a spectral cumulative contribution rate greater than 90% as the effective spectral information of Poria cocos from each origin, reconstructing the spectrum of the test spectrum and predicting its origin.

2. The method for identifying the origin of Poria cocos based on the principal component analysis of Poria cocos Raman spectrum according to claim 1, characterized in that: In step S1, considering the uneven size of the purchased Poria cocos blocks, surface oxidation, and contamination by other substances during the production, preparation and transportation process, in order to reduce experimental errors, each time data is collected, the surface of the Poria cocos to be tested is cut off and the Raman spectrum data inside the Poria cocos block is collected.

3. The method for identifying the origin of Poria cocos based on principal component analysis of Poria cocos Raman spectrum according to claim 1, characterized in that: In step S2, a series of preprocessing such as averaging three times, removing the baseline, and screening characteristic bands are performed on the original Raman spectrum, which can effectively reduce the impact of random noise, improve data consistency, reduce data dimension and complexity, and focus on key information, thereby constructing an effective Poria cocos Raman spectrum data set and test set.

4. The method for identifying the origin of Poria cocos based on principal component analysis of Poria cocos Raman spectrum according to claim 1, characterized in that: In step S3, starting from the differences of Poria cocos caused by different origins, characteristic vector matrices of different origins are established. Each characteristic vector matrix represents the main characteristics and change rules of Poria cocos in its corresponding origin, and the characteristic vector matrices are independent of each other.

5. The method for identifying the origin of Poria cocos based on principal component analysis of Poria cocos Raman spectrum according to claim 1, characterized in that: In step S4, the eigenvector matrix defines the characteristic subspace of each origin spectrum. The projection reconstruction of the spectrum in the subspace can be regarded as the explanatory power of the spectrum on the characteristic distribution of the current origin. If the test spectrum belongs to a certain origin, then the test spectrum is most similar to the characteristic distribution of a certain origin, and its reconstructed spectrum and its test spectrum should have the minimum error in the characteristic subspace of the origin, that is, the Euclidean distance is the minimum.

6. The method for identifying the origin of Poria cocos based on principal component analysis of Poria cocos Raman spectrum according to claim 1, characterized in that: In step S5, effective principal component information is analyzed and appropriately selected. Experimental verification shows that the best classification effect is obtained when the principal component cumulative contribution rate threshold is set to 90%. This value can be modified according to different experimental requirements.

Images