CN112304918B

CN112304918B - Method and device for identifying mixture based on Raman spectrum and Raman spectrum detection equipment

Info

Publication number: CN112304918B
Application number: CN201910698180.8A
Authority: CN
Inventors: 王健年; 左佳倩
Original assignee: Beijing Jianzhi Technology Co ltd; Nuctech Co Ltd
Current assignee: Beijing Jianzhi Technology Co ltd; Nuctech Co Ltd
Priority date: 2019-07-30
Filing date: 2019-07-30
Publication date: 2022-04-01
Anticipated expiration: 2039-07-30
Also published as: CN112304918A

Abstract

The application discloses a method and a device for identifying a mixture based on a Raman spectrum and Raman spectrum detection equipment. The method comprises the following steps: a) finding out a linear combination with the minimum linear deviation of the Raman spectrum of the tested sample from all linear combinations of a plurality of reference components in a reference Raman spectrum library, and determining the spectrum characterized by the linear combination as a first target reference Raman spectrum, wherein the first target reference Raman spectrum is formed by linear combinations of the Raman spectra of the first group of reference components; b) removing spectral peaks in the Raman spectrum of the measured sample and the first target reference Raman spectrum which meet the requirement of similarity degree to obtain a first nonlinear deviation between the Raman spectrum of the measured sample and the first target reference Raman spectrum; and c) individually examining each component of the first set of reference components by means of the first non-linear deviation to determine a component of the sample under test.

Description

Method and device for identifying mixture based on Raman spectrum and Raman spectrum detection equipment

Technical Field

The embodiment of the invention relates to the technical field of spectrum detection, in particular to a method for identifying a mixture based on a Raman spectrum, a device for identifying the mixture based on the Raman spectrum and Raman spectrum detection equipment.

Background

Raman spectroscopy is a molecular spectrum based on the raman effect. The raman effect is: when a substance is illuminated with monochromatic excitation light, a series of raman scattered light is produced at wavelengths greater and less than the wavelength of the incident light. The shift (wavenumber) of these raman scattered light with respect to the excitation light corresponds to the type of molecular group of the substance, and the intensity reflects the number of molecular groups of the substance to be measured. The Raman spectrum can be used for qualitative and quantitative analysis of the substance molecules. The excitation light irradiates the tested sample to generate Raman scattering light. The raman scattered light can be collected and processed to form a raman spectrum of the sample under test.

In the identification method of the Raman spectrum, the comparison between the Raman spectrum of the tested sample and the sample in the standard spectrum library is the most efficient and feasible method, and the operators are not required to have deeper theoretical basic knowledge. For a mixture spectrum composed of a plurality of components, a sample combination most similar to the measured spectral line needs to be found from a spectrogram library. Methods such as multiple linear regression and principal component regression are commonly used, wherein principal component regression is an effective data dimension reduction method.

Disclosure of Invention

The embodiment of the invention provides a method for identifying a mixture based on Raman spectrum, which comprises the following steps:

a) finding out a linear combination with the minimum linear deviation of the Raman spectrum of the tested sample from all linear combinations of a plurality of reference components in a reference Raman spectrum library, and determining the spectrum characterized by the linear combination as a first target reference Raman spectrum, wherein the first target reference Raman spectrum is formed by linear combinations of the Raman spectra of the first group of reference components;

b) removing spectral peaks in the Raman spectrum of the measured sample and the first target reference Raman spectrum which meet the requirement of similarity degree to obtain a first nonlinear deviation between the Raman spectrum of the measured sample and the first target reference Raman spectrum; and

c) each component in the first set of reference components is individually interrogated with the first non-linear offset to determine a component of the sample under test.

In some embodiments, said step c) comprises:

c1) removing one component of the first set of reference components as a component to be verified from the first set of reference components to form a second set of reference components;

c2) comparing the Raman spectrum of the measured sample with various linear combinations of the Raman spectra of the second group of reference components to find out a second target reference Raman spectrum, wherein the second target reference Raman spectrum is the linear combination with the minimum linear deviation from the Raman spectrum of the measured sample in various linear combinations of the Raman spectra of the second group of reference components;

c3) removing spectral peaks in the Raman spectrum of the measured sample and the second target reference Raman spectrum that meet the requirement of similarity degree to obtain a second nonlinear deviation between the Raman spectrum of the measured sample and the second target reference Raman spectrum; and

c4) when the absolute value of the difference between the first nonlinear deviation and the second nonlinear deviation is smaller than a first threshold value, excluding the sample from containing the component to be verified; and when the absolute value of the difference between the first nonlinear deviation and the second nonlinear deviation is larger than or equal to a first threshold value, determining that the sample contains the component to be verified.

In some embodiments, said step c) further comprises:

c5) when the absolute value of the difference between the first and second non-linear deviations is greater than or equal to a first threshold, then the first set of reference components remains unchanged, selecting another component from the first set of reference components as the component to be verified and removing the other component to form a new second set of reference components, and re-performing steps c2) through c5) based on the new second set of reference components until the absolute value of the difference between the first and second non-linear deviations is less than the first threshold or determining whether each component of the first set of reference components is included in the sample; and when the absolute value of the difference between the first nonlinear deviation and the second nonlinear deviation is smaller than a first threshold value, replacing the first group of reference components with the second group of reference components to form a new first group of reference components, selecting another component from the second group of reference components as a component to be verified, and removing the other component from the second group of reference components to form a new second group of reference components, and executing the following steps:

c6) comparing the raman spectrum of the measured sample with all linear combinations of the raman spectra of the new first set of reference components to find a new first target reference raman spectrum, wherein the new first target reference raman spectrum is the linear combination with the minimum linear deviation from the raman spectrum of the measured sample in all linear combinations of the raman spectra of the new first set of reference components;

c7) removing spectral peaks in the raman spectrum of the measured sample and the new first target reference raman spectrum that meet the similarity requirement to obtain a new first nonlinear deviation between the raman spectrum of the sample and the new first target reference raman spectrum; and

c8) re-performing steps c2) to c7) based on the new first and second sets of reference components formed in step c5) and the new first non-linear deviation formed in step c7) until it is determined whether each component of the first set of reference components is contained in the sample.

In some embodiments, the step b) comprises:

b1) respectively extracting spectral peaks in the Raman spectrum of the sample and the first target reference Raman spectrum;

b2) comparing the Raman spectrum of the sample with the spectral peak corresponding to the position in the first target reference Raman spectrum to find out the spectral peak which meets the requirement of the similarity degree in the Raman spectrum of the sample and the first target reference Raman spectrum; and

b3) spectral peaks satisfying a similarity measure requirement are removed from the raman spectrum of the sample and the first target reference raman spectrum and the first nonlinear deviation is calculated.

In some embodiments, said step b2) comprises:

b21) calculating the height difference, the width difference and the central coordinate difference between the Raman spectrum of the sample and the spectrum peak corresponding to each group of positions in the first target reference Raman spectrum; and

b22) determining a spectral peak corresponding to a position in the Raman spectrum of the sample and the first target reference Raman spectrum, wherein the position meets the following conditions, as a spectral peak meeting the requirement of similarity degree: the height difference is less than a second threshold, the width difference is less than a third threshold, and the center coordinate difference is less than a fourth threshold.

In some embodiments, said step b2) comprises:

b23) searching a peak group in a Raman spectrum of a sample and the first target reference Raman spectrum, wherein the peak group at least comprises two adjacent spectral peaks, and the central coordinate difference of the adjacent spectral peaks is smaller than the minimum value of the peak width in the adjacent spectral peaks; and

b24) determining peak groups corresponding to positions in the Raman spectrum of the sample and the first target reference Raman spectrum, which meet the following conditions, as spectral peaks meeting the requirement of similarity degree: and the linear correlation degree of the peak clusters corresponding to the positions is greater than a fifth threshold value.

In some embodiments, step a) is implemented using a least squares method.

In some embodiments, step c2) is implemented using a least squares method.

Embodiments of the present application also provide a raman spectroscopy apparatus comprising:

a light source configured to emit excitation light to irradiate a sample;

an optical device configured to collect raman scattered light signals from the sample;

a spectrometer configured to generate a raman spectrum of the sample from the raman scattered light signal; and

a mixture identifier, the mixture identifier comprising:

the comparison module is configured to find out a linear combination with the minimum Raman spectrum linear deviation of the detected sample from all linear combinations of multiple reference components in the reference Raman spectrum library, and determine a spectrum characterized by the linear combination as a first target reference Raman spectrum, wherein the first target reference Raman spectrum is formed by linear combinations of Raman spectra of a first group of reference components;

a deviation calculation module configured to remove spectral peaks in the raman spectrum of the measured sample and the first target reference raman spectrum that meet a similarity requirement to obtain a first nonlinear deviation between the raman spectrum of the measured sample and the first target reference raman spectrum; and

a component-finding module configured to determine components of the sample under test by finding each component in the first set of reference components one-by-one.

Embodiments of the present application further provide an apparatus for identifying a mixture based on raman spectroscopy, including:

means for finding a linear combination with the minimum linear deviation from the raman spectrum of the sample to be measured among all linear combinations of the plurality of reference components included in the reference raman spectrum library, and determining a spectrum characterized by the linear combination as a first target reference raman spectrum consisting of linear combinations of raman spectra of the first group of reference components;

means for removing spectral peaks in the raman spectrum of the measured sample and the first target reference raman spectrum that meet a requirement for a degree of similarity to obtain a first non-linear deviation between the raman spectrum of the measured sample and the first target reference raman spectrum; and

means for determining the components of the sample being measured by examining each component of the first set of reference components one by one.

The method and the device for identifying the mixture based on the Raman spectrum can extract the nonlinear deviation between the measured Raman spectrum of the sample and the target reference Raman spectrum formed by linear combination of the Raman spectra of different components, and check whether each component in a group of reference components obtained by linear comparison is contained in the sample one by one based on the nonlinear deviation. This method can improve the accuracy of identification of the components of the mixture.

Drawings

For a better understanding of the invention, embodiments thereof will be described with reference to the following drawings:

FIG. 1A shows a flow diagram of a method for identifying a mixture based on Raman spectroscopy, in accordance with an embodiment of the present invention;

FIG. 1B illustrates an exemplary detailed flowchart of step S30 in FIG. 1A;

FIG. 2 shows an exemplary detailed flowchart of step S20 in FIG. 1A;

FIG. 3 illustrates an exemplary detailed flowchart of step S22 in FIG. 2;

FIG. 4 schematically shows a specific example of a Raman spectrum of a sample and a first target reference Raman spectrum, according to an embodiment of the present invention;

fig. 5 schematically shows a specific example of a raman spectrum of a sample after performing spectral peak extraction and a first target reference raman spectrum according to an embodiment of the present invention;

fig. 6A, 6B, and 6C schematically illustrate specific examples of nonlinear changes that may occur in the raman spectrum of the sample after performing the spectral peak extraction and the first target reference raman spectrum, respectively;

fig. 7 schematically shows a specific example of the raman spectrum of the sample after removing the spectral peaks satisfying the requirement of the degree of similarity and the first target reference raman spectrum according to an embodiment of the present invention;

FIG. 8 schematically illustrates a specific example of a Raman spectrum of a sample and a second target reference Raman spectrum, according to an embodiment of the present invention;

fig. 9 schematically shows a specific example of a raman spectrum of a sample after performing spectral peak extraction and a second target reference raman spectrum according to an embodiment of the present invention;

fig. 10 schematically shows a specific example of the raman spectrum of the sample after removing the spectral peaks satisfying the requirement of the degree of similarity and the second target reference raman spectrum according to an embodiment of the present invention;

FIG. 11 schematically shows exemplary Raman spectra of acetone, ethanol, and acetone-ethanol mixtures; and

fig. 12 shows a schematic block diagram of a raman spectroscopy detection apparatus according to an embodiment of the present invention.

Detailed Description

The technical scheme of the invention is further specifically described by the following embodiments and the accompanying drawings. In the specification, the same or similar reference numerals denote the same or similar components. The following description of the embodiments of the present invention with reference to the accompanying drawings is intended to explain the general inventive concept of the present invention and should not be construed as limiting the invention.

The use of raman spectroscopy to identify the composition of a mixture is a more complex problem than identifying the composition of a pure product. The mixture is composed of a plurality of different components. The common difference spectrum method, the mixture identification method of principal component regression analysis, etc. are all based on the principle that different substance spectra are linearly superposed. For example, for principal component regression analysis, considering each sample spectrum of the spectrum library as a vector, principal component regression can derive as few principal components as possible from a matrix composed of all sample spectra, so that they retain as much information as possible of the original spectrum and are not correlated with each other. These principal components are considered to be the linear combination of the spectra of the components in the mixture sample that identifies the components of the mixture.

The inventors have realised that for identifying components of a mixture using raman spectroscopy, errors can arise from using only a linear combination of the component spectra described above for identification. This is because the molecular structural characteristics of the different components in the mixture and their concentrations appear as characteristic peaks in the raman spectrum of the mixture that are all different in position, height, shape. However, due to the interaction between molecules in the mixture, various nonlinear changes of the raman spectrum of a single component, such as spectrum peak broadening, spectrum peak shifting, spectrum peak coincidence and the like, tend to occur, so that the relation between the spectrum peak intensity and the concentration deviates from linearity.

To this end, embodiments of the present application provide a method 100 of identifying a mixture based on raman spectroscopy. In this method, some non-linear factors in the raman spectrum of the mixture are taken into account. As shown in fig. 1, the method may include:

step S10: finding out a linear combination with the minimum linear deviation of the Raman spectrum of the tested sample from all linear combinations of a plurality of reference components in a reference Raman spectrum library, and determining the spectrum represented by the linear combination as a first target reference Raman spectrum, wherein the first target reference Raman spectrum is formed by linear combinations of the Raman spectra of the first group of reference components;

step S20: removing spectral peaks satisfying a similarity requirement in the raman spectrum of the measured sample and the first target reference raman spectrum to obtain a first nonlinear deviation between the raman spectrum of the measured sample and the first target reference raman spectrum; and

step S30: each component in the first set of reference components is individually interrogated with the first non-linear offset to determine a component of the sample under test.

It is emphasized that the above-mentioned "reference raman spectrum library" comprises raman spectra of various reference components. And the first target reference Raman spectrum is selected from one of all possible linear combinations of the Raman spectra of various reference components in the reference Raman spectrum library, namely the Raman spectra of various reference components are multiplied by respective coefficients respectively and then summed, and the coefficient combination with the minimum difference with the Raman spectrum of the measured sample, which is the finally determined linear combination, is obtained. At this time, if the coefficient preceding a certain reference component is equal to 0, it is verified that the first target reference raman spectrum does not include the reference component, and all reference components having coefficients not equal to zero become the first set of reference components, which are actually included in the first target reference raman spectrum. When the first step of the method is performed for the first time, the reference raman spectrum library may be a library comprising reference raman spectra of all existing standard samples. While the method is performed in subsequent iterations, the categories of the first set of reference components (i.e. the categories of the standard sample) may be updated at any time, e.g. become less and less.

Unlike conventional raman spectrum identification methods, in the embodiments of the present application, not only a linear combination (first target reference raman spectrum) of a set of possible reference components of the mixture is considered, but also a first nonlinear deviation between the raman spectrum of the sample and the first target reference raman spectrum is obtained by removing spectral peaks of the raman spectrum of the sample and the first target reference raman spectrum that are similar to each other to a higher degree. This first non-linear deviation reflects the effect of non-linear factors in the raman spectrum of the mixture. After taking such non-linear factors into account, the components of the sample may be finally determined by performing a one-by-one investigation of the various components of a set of possible components obtained based on a comparison of the linear combination of reference components with the actual raman spectrum of the sample. In this way, the accuracy of identification of the mixture can be improved, and especially the advantage of the method is more obvious when the Raman spectrum of the tested mixture sample is processed, wherein the nonlinear phenomenon is more (for example, the spectral peaks of different components are more overlapped).

As an example, the above step S10 may be implemented by a least square method. For example, the measured raman spectrum of the sample is taken as vector y, and the matrix consisting of all reference raman spectrum vectors in the spectral library is set as X. A set of materials (i.e. the first set of reference components described above) is found by a least squares method, the linearly combined line X β of the raman spectra of the set of reference components most closely resembling the raman spectrum of the sample. The least squares method can be calculated based on the following formula 1:

β＝(X^TX)^-1X^Ty (formula 1)

Wherein beta is the coefficient of the Raman spectrum vector of each reference component in the spectrum library (the coefficient forms a one-dimensional vector), X is the matrix formed by the Raman spectrum vectors of each reference component, and y is the Raman spectrum vector of the tested sample. In the matrix X, each column is a vector, which is the raman spectrum vector of a certain substance in the library (because the abscissa of the raman spectrum of each substance is the same scale of all wave numbers, and the ordinate is different intensities at each wave number, the raman spectrum of a substance can be characterized by only requiring intensities at each wave number, and can be used as a one-dimensional signal, and thus can be represented by a vector). For example, for a matrix X of M X N, raman spectrum vectors of N substances are contained therein, and the length (number of sampling points) of each raman spectrum vector is M. By way of example, M and N may each be a large integer, such as hundreds, thousands, etc., to improve the accuracy of the calculations.

It should be noted that the method of obtaining the first target reference raman spectrum and the first set of reference components by using the least squares method is merely exemplary, and the embodiments of the present application are not limited thereto, and other optimization methods for determining coefficients of linear combinations known in the art may be adopted.

In the above step S20, it is necessary to remove a spectral peak satisfying the requirement of the similarity degree in the raman spectrum of the sample and the first target reference raman spectrum. For "meeting the similarity requirement", it can be considered mainly from the non-linear factor of the raman spectrum of the mixture. Three typical non-linear factors are shown in fig. 6A, 6B and 6C, respectively. In fig. 6A, 6B, and 6C, the solid line represents a peak in the measured raman spectrum y of the sample, and the broken line represents a peak of the first target reference raman spectrum X β, which are positionally corresponding.

The situation of spectral peak broadening is shown in fig. 6A, and it is clear from fig. 6A that the spectral peak in y has a significantly increased width relative to the spectral peak of X β corresponding to its position. The situation of spectral peak shift is shown in fig. 6B, and it is clear from fig. 6B that the center of the spectral peak in y is significantly shifted from the center of the spectral peak with respect to X β corresponding to its position. Fig. 6C shows the overlapping of the spectral peaks, and it is clear from fig. 6C that two adjacent spectral peaks in y are close together to form a peak cluster, and two adjacent spectral peaks corresponding to X β are overlapped to form a peak cluster (it should be noted that when two adjacent spectral peaks are very close, one peak cluster may look like a wider spectral peak due to the overlapping effect). The term "peak cluster" means that the difference in the center coordinates of adjacent spectral peaks is smaller than the peak width of the spectral peak having the smallest peak width among the adjacent spectral peaks.

These non-linear factors, when of large amplitude, will interfere with the determination of the mixture composition, and therefore spectral peaks that are significantly affected by the non-linear factors need to be selected for special handling. For this purpose, in step S20, peaks satisfying the requirement of similarity (or having higher similarity) in the raman spectrum of the sample and the first target reference raman spectrum are removed. I.e. spectral peaks that are not or less affected by non-linear factors are removed, thus leaving spectral peaks that are more affected by non-linear factors.

The "similarity requirement" may be defined separately for specific non-linear factors. For example, for the case of spectral peak broadening and spectral peak shifting, the following can be considered:

let the width of a certain peak in the measured raman spectrum y of the sample be d1 and the width of the peak corresponding thereto in position in the first target reference raman spectrum X β be d2, then if the absolute value | d1-d2| of the difference between d1 and d2 is less than the width change threshold Dth, both can be considered to satisfy a similar change rule of width. Dth can be adjusted in the range of 0.5 × d1 to 0.1 × d1 depending on the performance of the instrument.

It is also possible to let the height of a certain peak in the raman spectrum y of the measured sample be h1 and the height of the peak corresponding thereto in position in the first target reference raman spectrum X β be h2, then if the absolute value | h1-h2| of the difference between h1 and h2 is smaller than the height variation threshold Hth, both can be considered to satisfy the rule of similar variation in height. Hth can be adjusted in the range of 0.5 h1 to 0.1 h1 depending on the performance of the instrument.

It is also possible to let the center coordinate of a certain peak in the raman spectrum y of the measured sample be X1 and the center coordinate of the peak corresponding thereto in the first target reference raman spectrum X β be X2, then if the absolute value | X1-X2| of the difference between X1 and X2 is smaller than the drift change threshold Xth, both can be considered to satisfy a similar change rule of the drift. Xth can be adjusted in the range of 3 to 10 (the abscissa of the raman spectrum is in wavenumber) depending on the instrument performance.

When a certain peak in the raman spectrum y of the measured sample and a peak corresponding thereto in position in the first target reference raman spectrum X β satisfy the above-described similarity variation rule of height, similarity variation rule of width, and similarity variation rule of drift at the same time, the two peaks are considered to be "peaks satisfying the similarity degree requirement" (or "peaks whose nonlinear variation is smaller than a certain threshold").

As can be seen from the above, when a certain peak in the raman spectrum y of the measured sample and a peak corresponding thereto in position in the first target reference raman spectrum X β simultaneously satisfy the above-described similarity variation rule of height, similarity variation rule of width, and similarity variation rule of drift, it means that the two peaks do not differ greatly in height, width, and lateral center coordinates, and thus are referred to as "peaks satisfying the similarity degree requirement". Conversely, as long as the two spectral peaks have a large difference in any of the height, width, and lateral center coordinates, the two spectral peaks will not be considered as spectral peaks that satisfy the requirement of the degree of similarity.

For the case of overlapping spectral peaks, the following can be considered:

taking two spectral peaks a and b in a raman spectrum as an example, when the absolute value | x1 '-x 2' | of the difference between the central coordinate x1 'of the spectral peak a and the central coordinate x 2' of the spectral peak b is smaller than the peak width of the spectral peak with the smaller peak width, the spectral peak a and the spectral peak b are considered to form a peak group.

Definition of start and end coordinates of the peak clusters:

the starting coordinate xs of the peak cluster is the center coordinate of the first spectral peak minus the peak width of the first spectral peak, the ending coordinate xe of the peak cluster is the center coordinate of the last spectral peak plus the peak width of the last spectral peak, and the starting coordinate and the ending coordinate of a single spectral peak are defined in the same way.

Let the starting coordinate of a certain peak cluster of the measured raman spectrum y of the sample be x1s and the ending coordinate be x1 e; the first target reference raman spectrum X β has a certain peak cluster with a starting coordinate X2s and an ending coordinate X2 e.

Looking for clusters on the raman spectrum y of the sample and the first target reference raman spectrum X β, when the absolute value of the difference between the starting coordinates | X1s-X2s | and the absolute value of the difference between the ending coordinates | X1e-X2e | of a cluster in y and a cluster in X β are both less than the threshold Dth2, then the two are considered to be location-corresponding. The Dth2 can be adjusted in the range of 0 to 5 wave numbers depending on the performance of the instrument.

When the linear correlation degree of a certain peak cluster in y and the peak cluster corresponding to the position of the certain peak cluster in X beta is greater than a peak cluster correlation degree threshold Cth, the nonlinear difference between the two is considered to be smaller than a certain threshold, or to be a spectrum peak meeting the requirement of the similarity degree. Cth can be adjusted, for example, in the range from 0.6 to 0.8. The linear correlation of two peak clusters can be defined in various ways, e.g.

Wherein Corr is the linear correlation, y_iIntensity of the ith sample point of the cluster in y, x_iThe intensity of the ith sample point for the cluster corresponding to the location of the cluster in X beta in y,

is the intensity y of n sample points₁，y₂,., average value of yn,

is the intensity x of n sample points₁，x₂，...，x_nAverage value of (a). For the same i, x_iAnd y_iCorresponding to the same wavenumber.

However, the calculation of the linear correlation of the peak clusters in the embodiment of the present application is not limited to the above formula, and other algorithms of linear correlation known in the art may be used.

The greater the linear correlation of the two clusters, the greater their similarity.

In summary, it can be seen that in step S20, spectral peaks with larger similarity in the measured raman spectrum y of the sample and the first target reference raman spectrum X β are removed, while spectral peaks with larger difference in the measured raman spectrum y of the sample and the first target reference raman spectrum X β are retained. While the remaining more distinct spectral peaks can be used as a measure of the difference between the raman spectrum of the sample and the first target reference raman spectrum, referred to herein as a first non-linear deviation. For the convenience of the quantization process, the first non-linear deviation may be defined as:

where B1 is a first nonlinear deviation between the raman spectrum of the sample and the first target reference raman spectrum, y is the raman spectrum of the measured sample, and y ═ y is defined in the above formula for convenience of description_i}(1≤i≤n)，y_cIs as followsThe spectrum obtained after removing a peak satisfying the requirement of the degree of similarity from the measured Raman spectrum y of the sample is defined in the above formula for the convenience of description_c＝{y_ci}(1≤i≤n)，Xβ_cIs a spectrum obtained after a first target reference raman spectrum X β is removed from a spectrum peak satisfying a requirement of a degree of similarity, and X β is defined in the above formula for convenience of description_c＝{z_i}(1≤i≤n)。

In equation 3, the first nonlinear deviation B1 is y_cAnd X beta_cDivided by the sum of the intensities of the measured raman spectra y of the sample. This normalization is defined in a manner that facilitates examination of the proportion of the portion of the raman spectrum of the sample and the first target reference raman spectrum that is strongly affected by the non-linear factor with respect to the sum of the intensities of the measured raman spectrum y of the sample. However, this definition is merely for the sake of computational convenience and is not the only way to define the first non-linear deviation B1. The definition of the first nonlinear deviation B1 in the embodiment of the present application is not limited thereto as long as it reflects the spectral peak intensities that are largely different in the measured raman spectrum y of the sample and the first target reference raman spectrum X β, and for example, it may be normalized by another basis, or may not even be normalized.

As for step S20, as shown in fig. 2, it may include the following specific steps, as an example:

step S21: respectively extracting spectral peaks in the Raman spectrum of the sample and the first target reference Raman spectrum;

step S22: comparing the Raman spectrum of the sample with the spectral peak corresponding to the position in the first target reference Raman spectrum to find out the spectral peak which meets the requirement of the similarity degree in the Raman spectrum of the sample and the first target reference Raman spectrum; and

step S23: spectral peaks satisfying a similarity measure requirement are removed from the raman spectrum of the sample and the first target reference raman spectrum and the first nonlinear deviation is calculated.

Step S21 is the basis for finding the spectral peaks in the raman spectrum of the sample and the first target reference raman spectrum that meet the requirement of similarity. As an example, spectral peaks may be extracted from the raman spectrum of the sample based on changes in curvature of the raman spectrum and the first target reference raman spectrum (e.g., by calculating a derivative of a spectral curve). However, the specific way of extracting the spectral peak is not limited to this, and the spectral peak may also be extracted by other peak search ways, for example, two high and low thresholds may be estimated by using the statistical distribution of peak signals in the raman spectral curve, in the search process, when the signal amplitude is greater than the high threshold, the peak is considered to be found, and the point smaller than the low threshold is considered to be the start point and the end point of the peak forward and backward, and then the parameters such as the peak height, the peak width, and the like are calculated.

Specific examples of step S22 and step S23 have already been described in the foregoing sections, and are not described herein again.

As described above, in step S22, when the cases of the spectral peak broadening and the spectral peak shifting are considered, it is necessary to determine whether or not a certain spectral peak in the raman spectrum y of the measured sample is a spectral peak satisfying the requirement of the degree of similarity by considering the difference in height, width, and lateral center coordinates of the spectral peak corresponding to the position thereof in the first target reference raman spectrum X β. Thus, in one example, step S22 may include:

step S221: calculating the height difference, the width difference and the central coordinate difference between the Raman spectrum of the sample and the spectrum peak corresponding to each group of positions in the first target reference Raman spectrum; and

step S222: determining a spectral peak corresponding to a position in the Raman spectrum of the sample and the first target reference Raman spectrum, wherein the position meets the following conditions, as a spectral peak meeting the requirement of similarity degree: the height difference is less than a second threshold (height change threshold), the width difference is less than a third threshold (width change threshold), and the center coordinate difference is less than a fourth threshold (drift change threshold).

When the situation of overlapping of the spectral peaks is considered, the linear correlation degree of the peak groups corresponding to the positions in the Raman spectrum of the sample and the first target reference Raman spectrum needs to be considered. Thus, as an example, step S22 may include:

step S223: searching a peak group in a Raman spectrum of a sample and the first target reference Raman spectrum, wherein the peak group at least comprises two adjacent spectral peaks, and the central coordinate difference of the adjacent spectral peaks is smaller than the minimum value of the peak width in the adjacent spectral peaks; and

step S224: determining peak groups corresponding to positions in the Raman spectrum of the sample and the first target reference Raman spectrum, which meet the following conditions, as spectral peaks meeting the requirement of similarity degree: the linear correlation degree of the peak clusters corresponding to the positions is larger than a fifth threshold value (peak cluster correlation degree threshold value).

As an example, as shown in fig. 3, step S22 may also include the above steps S221, S222, S223, and S224 to consider the above situations of spectrum peak broadening, spectrum peak shifting, and spectrum peak overlapping. Steps S221 and S222 and steps S223 and S224 may be processed in parallel or in series. This may more fully balance the effect of non-linear factors in the raman spectra of the mixture.

After finding a peak in the raman spectrum of the sample and the first target reference raman spectrum, which is subject to a nonlinear factor, through step S20, each component in the first set of reference components is individually examined to determine the components of the sample in step S30. The idea of investigating the components may be: forming a second set of reference components after removing a certain component J from the first set of reference components and replacing the first set of reference components with the second set of reference components to perform the aforementioned alignment with the measured raman spectrum of the sample and to calculate a deviation (similar to a first non-linear deviation) between the raman spectrum of the sample and a target reference raman spectrum obtained based on the second set of reference components. The second set of reference components differs from the first set of reference components only in that the components that are removed are not included. By comparing the deviation between the raman spectrum of the sample and the target reference raman spectrum obtained based on the second set of reference components with the aforementioned first non-linear deviation, it can be determined whether the component taken out of the first set of reference components is contained in the sample under test.

As an example, as shown in fig. 1B, the step S30 may include:

step S31: removing one component of a first set of reference components as component J to be verified from the first set of reference components to form a second set of reference components;

step S32: comparing the Raman spectrum of the measured sample with all linear combinations of the Raman spectra of the second group of reference components to find out a second target reference Raman spectrum, wherein the second target reference Raman spectrum is the linear combination with the minimum linear deviation from the Raman spectrum of the measured sample in all linear combinations of the Raman spectra of the second group of reference components;

step S33: removing spectral peaks in the Raman spectrum of the measured sample and the second target reference Raman spectrum that meet the requirement of similarity degree to obtain a second nonlinear deviation between the Raman spectrum of the measured sample and the second target reference Raman spectrum; and

step S34: when the absolute value of the difference between the first nonlinear deviation and the second nonlinear deviation is smaller than a first threshold value, excluding the sample from containing the component to be verified; and when the absolute value of the difference between the first nonlinear deviation and the second nonlinear deviation is larger than or equal to a first threshold value, determining that the sample contains the component to be verified.

The above steps S32 and S33 are similar to the above steps S10 and S20, with the main difference being that the first set of reference constituents is replaced by the second set of reference constituents. The first target reference raman spectrum X β is replaced by the second target reference raman spectrum X β' accordingly. Accordingly, to facilitate calculation with the first non-linear deviation B1, if the first non-linear deviation B1 is defined according to equation 3 above, the second non-linear deviation may be defined as:

where B2 is a second nonlinear deviation between the raman spectrum of the sample and a second target reference raman spectrum, y is the measured raman spectrum of the sample, and y ═ y is defined in the above formula for descriptive convenience_i}(1≤i≤n)，y’_cIt is necessary that the Raman spectrum y of the measured sample be removed to satisfy the degree of similarityThe spectrum obtained after the spectral peak is taken (possibly compared to the second target reference raman spectrum and therefore likely to be y_cDifferent), y 'is defined in the above formula for convenience of description'_c＝{y’_ci}(1≤i≤n)，Xβ’_cIs a spectrum obtained after the second target reference raman spectrum X β ' has removed a peak satisfying the requirement of the degree of similarity, and for the convenience of description, X β ' is defined as { z '_i}(1≤i≤n)。

In equation 4, a second nonlinear deviation B2 is defined as y'_cAnd X beta'_cDivided by the sum of the intensities of the measured raman spectra y of the sample. This normalization is defined in a manner consistent with the definition of the first non-linear deviation B1 in equation 3. Likewise, the definition of the second nonlinear deviation B2 in the embodiment of the present application is not limited thereto, as long as it reflects the less similar spectral peak intensities in the measured raman spectrum y of the sample and the second target reference raman spectrum X β', and for example, it may be normalized by other criteria, or even may not be normalized.

In step S34, if the absolute value of the difference between the first nonlinear deviation B1 and the second nonlinear deviation B2 is smaller than the first threshold, it means that the variation of the deviation between the raman spectrum of the sample and the target reference raman spectrum with respect to the nonlinear factor is not large after the component J to be verified is taken out, and it can be determined that the component J to be verified is not contained in the sample to be tested; on the contrary, if the absolute value of the difference between the first nonlinear deviation B1 and the second nonlinear deviation B2 is greater than or equal to the first threshold, it means that the deviation with respect to the nonlinear factor between the raman spectrum of the sample and the target reference raman spectrum after taking out the component J to be verified is greatly changed, and it can be determined that the component J to be verified is contained in the sample to be tested. The first threshold value may be selected according to actual requirements.

In order to finally determine the components of the sample, all the components in the first set of reference components obtained in step S10 may be individually examined as components to be verified. To this end, as shown in fig. 1B, as an example, step S30 may further include:

step S35: when the absolute value of the difference between the first nonlinear deviation and the second nonlinear deviation is greater than or equal to a first threshold value, then the first group of reference components is kept unchanged, another component K is selected from the first group of reference components as a component to be verified and taken out to form a new second group of reference components (the component K to be verified can also be removed from the old second group of reference components and added to the already confirmed old component J to form a new second group of reference components), and the above steps S32 to S35 are re-executed based on the new second group of reference components until the absolute value of the difference between the first nonlinear deviation and the second nonlinear deviation is less than the first threshold value or whether each component in the first group of reference components is included in the sample is determined; and when the absolute value of the difference between the first nonlinear deviation and the second nonlinear deviation is smaller than a first threshold value, removing the old component J to be verified from the first group of reference components to form a new first group of reference components (the second group of reference components can also replace the first group of reference components to form a new first group of reference components), selecting another component K from the second group of reference components as the component to be verified, removing the other component K from the second group of reference components to form a new second group of reference components, and executing the following steps:

step S36: comparing the raman spectrum of the measured sample with all linear combinations of the raman spectra of the new first set of reference components to find a new first target reference raman spectrum, wherein the new first target reference raman spectrum is the linear combination with the minimum linear deviation from the raman spectrum of the measured sample in all linear combinations of the raman spectra of the new first set of reference components;

step S37: removing spectral peaks in the raman spectrum of the measured sample and the new first target reference raman spectrum that meet the similarity requirement to obtain a new first nonlinear deviation between the raman spectrum of the sample and the new first target reference raman spectrum; and

step S38: based on the new first and second sets of reference constituents formed in step S35 and the new first non-linear deviation formed in step S37, steps S32 through S37 are re-executed until it is determined whether each of the first set of reference constituents is included in the sample.

It should be noted that when the absolute value of the difference between the first non-linear deviation and the second non-linear deviation is smaller than the first threshold, the second set of reference components is required to replace the first set of reference components to form a new first set of reference components, and since the first set of reference components is updated, the first non-linear deviation also needs to be recalculated based on the new first set of reference components. Thus, in practice, in the above steps, the first set of reference components, the second set of reference components, the first target reference raman spectrum, the second target reference raman spectrum, the first non-linear deviation and the second non-linear deviation may be iteratively updated depending on the calculation result related to the change of the component to be verified.

As described above, both the first target reference raman spectrum and the second target reference raman spectrum can be found by the least squares method similar to that shown in formula 1. However, embodiments of the present application are not limited thereto.

The method for identifying a mixture based on raman spectroscopy according to an embodiment of the present application is further described below with reference to an example.

The solid line of fig. 4 shows the measured raman spectrum of one of the measured samples (hereinafter referred to as measured line y), and the dashed line is the first target reference raman spectrum found by the least-squares method, which is a linear combination of the raman spectra of acetonitrile, ethanol and cyclohexanone (hereinafter referred to as combined line X β), i.e. the first set of reference components is acetonitrile, ethanol and cyclohexanone. Although it is seen from fig. 4 that the deviation of the combined line X β of acetonitrile, ethanol and cyclohexanone from the measured line y is small at this time, due to the influence of the nonlinear factors of the raman spectrum of the mixture, cyclohexanone, acetonitrile and ethanol are not really correct answers.

And extracting the peaks of the measured spectral line y and the combined spectral line X beta according to the curvature change. As shown in fig. 5, the solid line represents the peaks of the extracted measured line y, and the dotted line represents the peaks of the combined line X β of cyclohexanone, acetonitrile, and ethanol. According to the rules for measuring nonlinear change such as spectrum peak broadening, spectrum peak drifting and spectrum peak coincidence, the two spectrum peaks meeting the requirement of similarity degree are deleted. The remaining peaks of the measured line y and the combined line X β after eliminating the peaks both satisfying the similarity requirement are shown in fig. 7. And summing the two residual spectral peaks, and dividing the sum by the sum of the original spectral peak intensities of the measured spectral line y to obtain a first nonlinear deviation B1 (0.0507) of the combined spectral line X beta of cyclohexanone, acetonitrile and ethanol and the measured spectral line y.

Cyclohexanone is removed from the combination of cyclohexanone, acetonitrile and ethanol, and the raman spectrum of the standard sample of acetonitrile and ethanol is subjected to a least square method to obtain a new combination spectral line X β' as shown in fig. 8. The peaks of the measured line y and the new combined line X β' are likewise extracted. As shown in fig. 9, the solid line indicates the peaks of the extracted measured line y, and the dotted line indicates the peaks of the combined line X β' of acetonitrile and ethanol. According to the rules for measuring nonlinear change such as spectrum peak broadening, spectrum peak drifting and spectrum peak coincidence, the two spectrum peaks meeting the requirement of similarity degree are deleted. The remaining peaks of the measured line y and the new combined line X β' after eliminating the peaks both satisfying the similarity requirement are shown in fig. 10. And summing the two rest peaks, and dividing the sum by the sum of the original peak intensities of the measured spectral line y to obtain a second nonlinear deviation B2 of the new combined spectral line X beta' and the measured spectral line y which is 0.0537.

The first threshold TH is set to 0.05, which is smaller than the first non-linear deviation | B2-B1| -0.0507-0.0537 | -0.003. Compared with the combined spectral line X beta of cyclohexanone, acetonitrile and ethanol, the deviation caused by nonlinear factors has small change, so that the cyclohexanone can be removed. If not cyclohexanone but ethanol is removed, i.e. the above-mentioned deviation B2 'due to non-linear factors is calculated to be 0.3485 based on the combined spectrum of acetonitrile, cyclohexanone and measured spectrum y, in which case | B2' -B1| -0.2978 will be greater than the first threshold value TH, so cyclohexanone is taken as the correct choice at this step.

If ethanol or acetonitrile is continuously taken out of the combination of acetonitrile and ethanol, the new second nonlinear deviation B2' is calculated to reach 0.5439 or 0.5444, and the absolute value of the difference between the new second nonlinear deviation B2 (new first nonlinear deviation) of the previous step is larger than the first threshold value, so that the acetonitrile and ethanol are the final components of the tested sample determined by the method. This corresponds exactly to the actual composition of the sample being measured.

Comparing figure 8 with figure 5, it can be seen that the linear deviation of the combined line X β' from the measured line y is greater than the linear deviation of the combined line X β from the measured line y over the values of the lines, but in fact acetonitrile ethanol is the correct answer to the mixture composition. This is due to the fact that the interaction between the different substances of the mixture causes a non-linear change in the spectral peaks of the final mixture.

According to the method for identifying the mixture, compared with the traditional linear identification method, various nonlinear changes such as spectral peak overlapping, spectral peak broadening, spectral peak height change and drifting and the like in the Raman spectrum of the mixture can be effectively overcome, the measurement precision is effectively improved, and the false alarm rate is reduced.

The existing linear analysis method usually assumes that the spectral lines of a mixture are linear superposition of the spectral lines of various component substances, so that the accuracy of the algorithm is reduced when the conditions with more nonlinear changes, such as more spectral peaks, are overlapped. For example, the acetone-ethanol mixture shown in fig. 11 has a relatively large nonlinear change, which can be seen by comparing the raman spectrum of acetone, the raman spectrum of ethanol and the raman spectrum of the acetone-ethanol mixture shown in fig. 11. The method provided by the embodiment of the application is particularly advantageous when dealing with the situation that the raman spectrum of the mixture has more nonlinear changes.

As an example, the above method may further comprise the step of measuring a raman spectrum of the sample.

The present application also provides a raman spectroscopy detection device 200. As shown in fig. 12, the raman spectrum detection apparatus 200 includes: a light source 10, optics 20, a spectrometer 30, and a mixture identifier 40. A light source 10, such as a laser, may be configured to emit excitation light 11 to illuminate the sample 50. An optical device 20, such as a projection system or a lens, may be configured to collect the raman scattered light signal from the sample 50. The spectrometer 30 may be configured to generate a raman spectrum of the sample 50 from the raman scattered light signal. The mixture identifier 40 may include: a comparison module 41, a deviation calculation module 42 and a component checking module 43.

The comparison module 41 may be configured to find a linear combination with the smallest linear deviation from the raman spectrum of the measured sample from all linear combinations of the plurality of reference components included in the reference raman spectrum library, and determine a spectrum characterized by the linear combination as a first target reference raman spectrum, where the first target reference raman spectrum is formed by linear combinations of raman spectra of the first group of reference components. The bias calculation module 42 may be configured to remove spectral peaks in the raman spectrum of the measured sample and the first target reference raman spectrum that meet the similarity requirement to obtain a first non-linear bias between the raman spectrum of the measured sample and the first target reference raman spectrum. The component-investigating module 43 may be configured to determine the components of the measured sample by investigating each component of the first set of reference components one by one. As an example, the comparison module 41 may be configured to perform various exemplary steps S10, the deviation calculation module 42 may be configured to perform various exemplary steps S20, and the component checking module 43 may be configured to perform various exemplary steps S30.

In some embodiments, the mixture identifier 40 may be implemented by a processor operable to perform any of the exemplary steps of the methods described above. As an example, the different steps may be implemented by different processes. In embodiments of the present application, mixture identifier 40 may be provided separately from spectrometer 30 or may be integrated with spectrometer 30.

Embodiments of the present application further provide an apparatus for identifying a mixture based on raman spectroscopy, including: means for finding a linear combination with the minimum linear deviation from the raman spectrum of the sample to be measured among all linear combinations of the plurality of reference components included in the reference raman spectrum library, and determining a spectrum characterized by the linear combination as a first target reference raman spectrum consisting of linear combinations of raman spectra of the first group of reference components; means for removing spectral peaks in the raman spectrum of the measured sample and the first target reference raman spectrum that meet a similarity requirement to obtain a first non-linear deviation between the raman spectrum of the measured sample and the first target reference raman spectrum; and means for determining the components of the sample being measured by examining each component of the first set of reference components one by one.

The foregoing detailed description has set forth numerous embodiments of the above-described test subject security detection apparatus and method via the use of schematics, flowcharts, and/or examples. Where such diagrams, flowcharts, and/or examples contain one or more functions and/or operations, it will be understood by those within the art that each function and/or operation within such diagrams, flowcharts, or examples can be implemented, individually and/or collectively, by a wide range of structures, hardware, software, firmware, or virtually any combination thereof. In one embodiment, portions of the subject matter described by embodiments of the invention may be implemented by Application Specific Integrated Circuits (ASICs), Field Programmable Gate Arrays (FPGAs), Digital Signal Processors (DSPs), or other integrated formats. However, those skilled in the art will recognize that some aspects of the embodiments disclosed herein, in whole or in part, can be equivalently implemented in integrated circuits, as one or more computer programs running on one or more computers (e.g., as one or more programs running on one or more computer systems), as one or more programs running on one or more processors (e.g., as one or more programs running on one or more microprocessors), as firmware, or as virtually any combination thereof, and that designing the circuitry and/or writing the code for the software and or firmware would be well within the skill of one of skill in the art in light of this disclosure. In addition, those skilled in the art will appreciate that the mechanisms of the subject matter described herein are capable of being distributed as a program product in a variety of forms, and that an illustrative embodiment of the subject matter described herein applies regardless of the particular type of signal bearing media used to actually carry out the distribution. Examples of signal bearing media include, but are not limited to: recordable type media such as floppy disks, hard disk drives, compact disks (CDs, DVDs), digital tape, computer memory, etc.; and a transmission type medium such as a digital and/or an analog communication medium (e.g., a fiber optic cable, a waveguide, a wired communications link, a wireless communication link, etc.).

Unless a technical obstacle or contradiction exists, the above-described various embodiments of the present invention may be freely combined to form further embodiments, which are within the scope of the present invention.

Although the present invention has been described in connection with the accompanying drawings, the embodiments disclosed in the drawings are intended to be illustrative of preferred embodiments of the present invention and should not be construed as limiting the invention.

Although a few embodiments of the present general inventive concept have been shown and described, it would be appreciated by those skilled in the art that changes may be made in these embodiments without departing from the principles and spirit of the general inventive concept, the scope of which is defined in the claims and their equivalents.

Claims

1. A method of identifying a mixture based on raman spectroscopy, comprising the steps of:

b) removing spectral peaks satisfying a similarity requirement in the raman spectrum of the measured sample and the first target reference raman spectrum to obtain a first nonlinear deviation between the raman spectrum of the measured sample and the first target reference raman spectrum; and

c) (ii) examining each component of the first set of reference components one by one with the aid of the first non-linear deviation to determine a component of the sample under test,

wherein the step c) comprises:

c4) when the absolute value of the difference between the first nonlinear deviation and the second nonlinear deviation is smaller than a first threshold value, excluding the sample from containing the component to be verified; when the absolute value of the difference between the first nonlinear deviation and the second nonlinear deviation is greater than or equal to a first threshold value, determining that the sample contains the component to be verified, and

wherein the similarity degree is defined according to the nonlinear factor of the Raman spectrum of the mixture.

2. The method for identifying a mixture based on raman spectroscopy of claim 1, wherein the step c) further comprises:

c5) when the absolute value of the difference between the first and second non-linear deviations is greater than or equal to a first threshold, then the first set of reference components remains unchanged, selecting another component from the first set of reference components as the component to be verified and removing the other component to form a new second set of reference components, and re-performing steps c2) through c5) based on the new second set of reference components until the absolute value of the difference between the first and second non-linear deviations is less than the first threshold or determining whether each component of the first set of reference components is included in the sample; and when the absolute value of the difference between the first nonlinear deviation and the second nonlinear deviation is smaller than a first threshold value, replacing the first group of reference components with a second group of reference components to form a new first group of reference components, selecting another component from the second group of reference components as a component to be verified, and removing the other component from the second group of reference components to form a new second group of reference components, and executing the following steps:

c8) based on the new first and second sets of reference components formed in step c5) and the new first non-linear deviation formed in step c7), re-performing steps c2) through c7) until it is determined whether each component of the first set of reference components is contained in the sample.

3. The method of identifying a mixture based on raman spectroscopy of claim 1, wherein the step b) comprises:

4. The method for identifying a mixture based on raman spectroscopy of claim 3, wherein the step b2) comprises:

5. The method for identifying a mixture based on raman spectroscopy of claim 3, wherein the step b2) comprises:

6. A method for identifying mixtures based on Raman spectroscopy according to any one of claims 1 to 5 wherein step a) is carried out using a least squares method.

7. Method for identifying mixtures based on raman spectroscopy according to any one of claims 1 to 5, wherein step c2) is carried out using a least squares method.

8. A raman spectroscopy detection apparatus comprising:

a light source configured to emit excitation light to irradiate a sample;

a mixture identifier, the mixture identifier comprising:

a component-finding module configured to determine components of a sample under test by finding each component in the first set of reference components one-by-one,

wherein determining the components of the test sample by examining each component of the first set of reference components one by one comprises:

removing one component of the first set of reference components as a component to be verified from the first set of reference components to form a second set of reference components;

comparing the Raman spectrum of the measured sample with various linear combinations of the Raman spectra of the second group of reference components to find out a second target reference Raman spectrum, wherein the second target reference Raman spectrum is the linear combination with the minimum linear deviation from the Raman spectrum of the measured sample in various linear combinations of the Raman spectra of the second group of reference components;

removing spectral peaks in the Raman spectrum of the measured sample and the second target reference Raman spectrum that meet the requirement of similarity degree to obtain a second nonlinear deviation between the Raman spectrum of the measured sample and the second target reference Raman spectrum; and

when the absolute value of the difference between the first nonlinear deviation and the second nonlinear deviation is smaller than a first threshold value, excluding the sample from containing the component to be verified; when the absolute value of the difference between the first nonlinear deviation and the second nonlinear deviation is greater than or equal to a first threshold value, determining that the sample contains the component to be verified, and

9. An apparatus for identifying a mixture based on raman spectroscopy, comprising:

means for determining the components of the sample under test by examining each component of the first set of reference components one by one,

wherein the means for determining the composition of the sample under test by individually interrogating each of the first set of reference components comprises:

means for removing one component of the first set of reference components from the first set of reference components as a component to be verified to form a second set of reference components;

means for comparing the raman spectrum of the measured sample with various linear combinations of raman spectra of the second set of reference components to find a second target reference raman spectrum, the second target reference raman spectrum being a linear combination of the various linear combinations of raman spectra of the second set of reference components that has the least linear deviation from the raman spectrum of the measured sample;

means for removing spectral peaks in the raman spectrum of the measured sample and the second target reference raman spectrum that meet a similarity requirement to obtain a second non-linear deviation between the raman spectrum of the measured sample and the second target reference raman spectrum; and

means for performing the following: when the absolute value of the difference between the first nonlinear deviation and the second nonlinear deviation is smaller than a first threshold value, excluding the sample from containing the component to be verified; when the absolute value of the difference between the first nonlinear deviation and the second nonlinear deviation is greater than or equal to a first threshold value, determining that the sample contains the component to be verified, and