Impurity detection method in the extraction process of sweet orange flower essential oil based on spectral characteristics

Abstract

The present application relates to the field of spectral analysis technology, and specifically to a method for detecting impurities in the extraction process of sweet orange flower essential oil based on spectral characteristics, the method comprising: collecting Raman spectral data of sweet orange flower essential oil samples; calculating the affine deviation coefficient and the left and right half-peak trend transformation ratio of each Raman peak in the Raman spectral data, and determining the leading-edge tailing affine deviation coefficient of each Raman peak based on this; analyzing the sample correlation deviation coefficient between any two sweet orange flower essential oil samples to calculate the main component impurity impact assessment coefficient of the sweet orange flower essential oil; using the PCA principal component analysis algorithm to perform dimensionality reduction processing on the Raman spectra of all sweet orange flower essential oil samples, and then determining the impurity content index of the sweet orange flower essential oil, and judging the impurities of the sweet orange flower essential oil. The present application adaptively adjusts the number of principal components according to the analysis of the impurity content in the sweet orange flower essential oil, reduces the interference of redundant information, and improves the detection accuracy.

Description

Technical Field

[0001] The present application relates to the technical field of spectral analysis, and in particular to a method for detecting impurities in a sweet orange flower essential oil extraction process based on spectral characteristics. Background Art

[0002] When extracting sweet orange flower essential oil, there will be interference from components such as flower stalks, branches and leaves, and young sweet orange fruits, causing the sweet orange flower essential oil to contain other non-volatile impurities. In addition, the extraction process is easily affected by microbial contamination and oxidation products, resulting in interference from impurity components in the sweet orange flower essential oil. These impurities may cause the Raman peaks in the sweet orange flower essential oil to overlap with the characteristic peaks, causing the peak intensity to vary.

[0003] Considering that the amount of information contained in the Raman spectrum data is relatively large, the variation information caused by impurities is overlapped with the spectral information, which makes it possible to detect impurities directly by using the spectral information of the collected samples and standard samples with large errors. The principal component analysis (PCA) algorithm is usually used to analyze the sample data containing impurities to improve the accuracy of impurity detection. However, a fixed number of principal components is usually selected in the PCA algorithm. Selecting too few principal components may result in the loss of some valid information, while selecting too many may result in too much interference information. Summary of the invention

[0004] In order to solve the above technical problems, the present application provides a method for detecting impurities in the sweet orange flower essential oil extraction process based on spectral characteristics to solve the existing problems.

[0005] The impurity detection method in the sweet orange flower essential oil extraction process based on spectral characteristics of the present application adopts the following technical solution:

[0006] One embodiment of the present application provides a method for detecting impurities in the extraction process of sweet orange flower essential oil based on spectral characteristics, the method comprising the following steps:

[0007] Collecting Raman spectral data of each sweet orange flower essential oil sample;

[0008] Determine the affine deviation coefficient of each Raman peak in each Raman spectral data based on the wave number and Raman scattering intensity corresponding to each Raman peak in each Raman spectral data; determine the left and right half-peak trend transformation ratio of each Raman peak based on the change of Raman scattering intensity on the left and right sides of each Raman peak; determine the leading edge tailing affine deviation coefficient of each Raman peak based on the affine deviation coefficient and the left and right half-peak trend transformation ratio;

[0009] Based on the leading-edge tailing affine deviation coefficient of each Raman peak of Raman spectrum data of any two sweet orange flower essential oil samples and the difference in Raman scattering intensity, the sample correlation deviation coefficient between the any two sweet orange flower essential oil samples is determined; the principal component impurity impact assessment coefficient of the sweet orange flower essential oil is determined in combination with the sample correlation deviation coefficient between all the any two sweet orange flower essential oil samples; based on the principal component impurity impact assessment coefficient, the PCA principal component analysis algorithm is used to perform dimensionality reduction processing on the Raman spectra of all sweet orange flower essential oil samples to obtain a Raman dimensionality reduction matrix; the Raman spectrum of the standard sweet orange flower essential oil is extracted, and accordingly, the standard Raman dimensionality reduction matrix is ​​obtained;

[0010] Based on the similarity between the Raman dimension reduction matrix and the standard Raman dimension reduction matrix, the impurity content index of the sweet orange flower essential oil is determined, and the impurities of the sweet orange flower essential oil are judged.

[0011] Preferably, the method for determining the affine deviation coefficient is:

[0012] Each Raman peak of each sweet orange flower essential oil Raman spectrum data is extracted by using a peak automatic detection algorithm, and the Raman scattering intensity of each Raman peak at each corresponding wave number is combined into a Raman peak sequence of each Raman peak;

[0013] determining a reference symmetry axis for each Raman peak based on the wave number of each Raman peak at the peak;

[0014] The affine deviation coefficient of each Raman peak is determined based on the reference symmetry axis of each Raman peak and the Raman scattering intensity of the corresponding Raman peak sequence at each wave number.

[0015] Preferably, the reference symmetry axis is the wave number corresponding to the peak value of each Raman peak.

[0016] Preferably, the affine deviation coefficient corresponds to the calculation formula:

[0017]

[0018] In the formula, represents the affine deviation coefficient of the kth Raman peak in the Raman spectrum data of the mth sweet orange flower essential oil sample, represents the number of wavenumbers in the kth Raman peak sequence in the Raman spectrum data of the mth sweet orange flower essential oil sample, represents the exponential function with natural number e as base, represents the Raman scattering intensity of the kth Raman peak sequence at the vth wave number in the Raman spectrum data of the mth sweet orange flower essential oil sample, Indicates the wavenumber corresponding to the kth Raman peak in the Raman spectrum data of the mth sweet orange flower essential oil sample.

[0019] Preferably, the method for determining the left and right half-peak trend conversion ratio is:

[0020] Each Raman peak sequence is divided into two sequences, left and right, based on the reference symmetry axis, and recorded as the left and right half-peak scattering sequences, respectively;

[0021] The STL time series decomposition algorithm is used to decompose the left and right half-peak scattering series respectively, and the trend term strength formula is used to calculate the left and right trend term strengths.

[0022] The absolute value of the intensity ratio of the left and right trend items is taken as the true number of a logarithmic function with the number 2 as the base, and the absolute value of the calculation result of the logarithmic function is taken as the left and right half-peak trend transformation ratio of each Raman peak.

[0023] Preferably, the method for determining the leading edge trailing affine deviation coefficient is:

[0024] Calculate the sum of the Pearson correlation coefficient of the left half peak scattering sequence and the right half peak scattering sequence of each Raman peak and the preset parameter adjustment factor, record it as the denominator, calculate the sum of the affine deviation coefficient of each Raman peak and the trend transformation ratio of the left and right half peaks, record it as the numerator;

[0025] The result of dividing the numerator by the denominator is taken as the leading-edge tailing affine deviation coefficient of each Raman peak.

[0026] Preferably, the method for determining the sample correlation deviation coefficient is:

[0027] Calculate the absolute value of the difference between the maximum values ​​of the Raman peak sequence of the Raman peaks corresponding to the Raman peaks of the Raman spectrum data of any two portions of sweet orange flower essential oils, and record it as the maximum absolute value of the difference; calculate the sum of the leading-edge tailing affine deviation coefficients of the corresponding Raman peaks of the any two portions of sweet orange flower essential oils;

[0028] Calculate the product of the sum value and the absolute value of the maximum difference; and take the sum of the products calculated from all corresponding Raman peaks of any two portions of sweet orange flower essential oil as the sample correlation deviation coefficient between the any two portions of sweet orange flower essential oil samples.

[0029] Preferably, the method for determining the main component impurity impact assessment coefficient of the sweet orange flower essential oil is:

[0030] The sample association deviation coefficients between all arbitrary two sweet orange flower essential oil samples are combined into a sample association deviation matrix, the F norm of the sample association deviation matrix is ​​calculated, and the result of dividing the F norm by the number of sweet orange flower essential oil samples is used as the main component impurity impact assessment coefficient of sweet orange flower essential oil.

[0031] Preferably, the dimensionality reduction processing method is:

[0032] The Raman spectrum data of all sweet orange flower essential oil samples are organized into a Raman detection matrix, the eigenvalues ​​of the Raman detection matrix are extracted, and the adaptive principal component screening threshold of the PCA principal component analysis algorithm is set to the product of the principal component impurity impact assessment coefficient and the maximum value of the eigenvalue;

[0033] The eigenvalues ​​greater than or equal to the product are retained as principal components, and the Raman detection matrix is ​​subjected to dimensionality reduction processing in combination with the PCA principal component analysis algorithm.

[0034] Preferably, the method for determining the impurity content index of the sweet orange flower essential oil is:

[0035] Calculating the absolute value of the Hadamard product of the Raman dimension reduction matrix and the standard Raman dimension reduction matrix, recorded as the first absolute value, and counting the maximum dimensions of the Raman dimension reduction matrix and the standard Raman dimension reduction matrix;

[0036] The ratio of the first absolute value to the maximum dimension is used as the impurity content index of sweet orange flower essential oil.

[0037] This application has at least the following beneficial effects:

[0038] This application analyzes the affine transformation of the Raman peak and the left and right half-peak trend intensity in the Raman spectrum data of sweet orange flower essential oil after noise reduction, obtains the front-tailing affine deviation coefficient according to the difference information, and constructs the sample correlation deviation matrix according to the peak intensity difference between the two combinations in all samples, and further obtains the principal component impurity impact assessment coefficient, thereby obtaining the adaptive principal component screening threshold. Finally, the Raman data of the batch of samples after dimensionality reduction and the database standard sweet orange flower essential oil Raman data are analyzed to obtain a more accurate sweet orange flower essential oil impurity content index. It solves the problem of missing information or retaining too much interference information when screening a fixed number of principal components in the traditional PCA algorithm, and can adjust the number of principal components according to the content of impurities in the monitored sweet orange flower essential oil, reduce the interference of redundant information, and improve the detection accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present application or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0040] Figure 1 A flowchart of the steps of the method for detecting impurities in the extraction process of sweet orange flower essential oil based on spectral characteristics provided in this application;

[0041] Figure 2Schematic diagram of the extraction process of the impurity content index of sweet orange flower essential oil. DETAILED DESCRIPTION

[0042] In order to further explain the technical means and effects adopted by the present application to achieve the predetermined invention purpose, the following is a detailed description of the impurity detection method in the sweet orange flower essential oil extraction process based on spectral characteristics proposed in the present application, its specific implementation method, structure, characteristics and effects as follows in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0043] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

[0044] The specific scheme of the impurity detection method in the sweet orange flower essential oil extraction process based on spectral characteristics provided by the present application is described in detail below in conjunction with the accompanying drawings.

[0045] An embodiment of the present application provides a method for detecting impurities in the process of extracting sweet orange flower essential oil based on spectral characteristics. Specifically, the following method for detecting impurities in the process of extracting sweet orange flower essential oil based on spectral characteristics is provided. Figure 1 , the method comprises the following steps:

[0046] Example:

[0047] Step S001, collecting spectral data of sweet orange flower essential oil and performing data preprocessing.

[0048] This embodiment aims to detect impurities in the extraction process of sweet orange flower essential oil by spectral analysis of sweet orange flower essential oil. In order to improve the impurity detection accuracy of sweet orange flower essential oil, this embodiment uses LabRAMSoleil™ high-resolution Raman spectrometer produced by HORIBA to collect spectral data of sweet orange flower essential oil. The implementer can also use other spectrometers to collect spectral data of sweet orange flower essential oil. In the acquisition process of this embodiment, the parameters of the Raman spectrometer are respectively set as follows: the output laser wavelength of the spectrometer is 760nm, the spectral scanning range is 100~3000cm-1, and the laser power is set to 70% of the rated power. The implementer can also set it by himself, based on which the spectral data of sweet orange flower essential oil is collected.

[0049] Further, after the sweet orange flower essential oil extracted from the same batch is stirred evenly, this embodiment randomly selects Raman spectroscopy is performed on the The Raman spectrum data of this batch of sweet orange flower essential oil was obtained. Represents the Raman scattering intensity of the Raman spectrum data of the mth portion of sweet orange flower essential oil sample at the vth wave number. The number of wave numbers is determined by the resolution and scanning range of the Raman spectrometer. There are baseline drifts and intensity fluctuations caused by environmental changes in the collected Raman spectrum. The collected Raman spectrum data is preprocessed using baseline subtraction and normalization methods to improve the accuracy of subsequent analysis data. The baseline subtraction and normalization methods are well-known techniques and will not be described in detail in this embodiment.

[0050] Step S002, based on the affine transformation of the Raman peak of a single sample after noise reduction and the difference in the trend intensity of the left and right half peaks, the leading-edge trailing affine deviation coefficient is obtained, all samples are combined in pairs, and the sample correlation deviation matrix is ​​obtained in combination with the peak intensity difference, and the principal component impurity impact assessment coefficient is further analyzed. Based on this, the adaptive principal component screening threshold in the PCA algorithm is adjusted, and the sweet orange flower essential oil impurity content index is obtained by comparing with the standard sweet orange flower essential oil data in the database.

[0051] Normally, sweet orange flower essential oil is extracted from degenerated male flowers or petals on sweet orange trees. However, impurities such as pedicels, branches and leaves, and young sweet orange fruits may be collected during the picking of petals. Therefore, there may be impurities in the extracted sweet orange flower essential oil. The presence of impurities affects the medicinal value and nursing effect of sweet orange flower essential oil, resulting in a certain degree of variation in the Raman spectrum of sweet orange flower essential oil. Raman spectroscopy provides more data information of sweet orange flower essential oil, which contains a lot of redundant information. At the same time, the variation information caused by impurities is cross-overlapped with the normal spectral information, so it is necessary to extract the main information of sweet orange flower essential oil. However, a fixed number of principal components are usually selected in the PCA algorithm. The selection of the number of principal components directly determines the effect of dimensionality reduction of the extracted Raman spectral data. Therefore, in order to reduce the information loss and overfitting caused by the fixed number of principal components, adaptive adjustments are made according to the analysis results of the Raman spectral data of sweet orange flower essential oil, as follows:

[0052] When there are impurities in the sweet orange flower essential oil, the Raman spectrum of the sweet orange flower essential oil will vary to a certain extent. Usually, the Raman peak tailing, front, peak overlap and peak intensity change in the Raman spectrum, and even the variation of new Raman peaks. The impurity content is relatively small compared to the content of the components of the sweet orange flower essential oil, so there may also be differences in impurities between samples. Based on the above characteristics, the overall situation of impurities in this batch of sweet orange flower essential oil is analyzed.

[0053] In the Raman spectrum of sweet orange flower essential oil, Raman peaks represent characteristic peaks in the Raman spectrum corresponding to specific vibration modes of sample molecules. There are multiple Raman peaks in the spectrum, and the position and intensity of each peak correspond to different molecular structures and chemical composition information in the sweet orange flower essential oil. Therefore, this embodiment is based on the collected Raman spectrum data of a single sample as the input of the peak automatic detection AMPD algorithm. The algorithm outputs all local peaks in the Raman spectrum of a single sample, and combines the baseline information during baseline correction to obtain all Raman peaks in the Raman spectrum through a Gaussian fitting algorithm. Represents the Raman scattering intensity of the kth Raman peak at the vth wavenumber position of the Raman spectrum of the mth portion of sweet orange flower essential oil. The Raman scattering intensity of each Raman peak at the corresponding wavenumber constitutes the Raman peak sequence of each Raman peak. The peak automatic detection AMPD algorithm and the Gaussian fitting algorithm are well-known technologies and will not be described in detail in this embodiment. For each Raman peak sequence, a denoising algorithm is used to denoise the sequence to avoid the influence of noise data. It should be noted that there are many methods for sequence denoising, such as non-local mean algorithm, wavelet denoising, smoothing filtering and other algorithms. In this embodiment, a wavelet denoising algorithm is used to denoise each sequence, and this embodiment does not impose any special restrictions on this.

[0054] For the denoised Raman spectrum signal, the wave number corresponding to the peak value of each Raman peak is used as the reference symmetry axis of each Raman peak. For example, use Represents the wave number corresponding to the kth Raman peak in the Raman spectrum data of the mth sweet orange flower essential oil sample, which is used as the reference symmetry axis of the kth Raman peak. According to the reference symmetry axis of the Raman peak, the denoised Raman peak sequence is divided into a left half-peak scattering sequence and a right half-peak scattering sequence, which are used as the input of the STL decomposition algorithm respectively. The algorithm outputs the random item sequence, trend item sequence and seasonal item sequence of the left and right half-peaks. The trend item strength formula is used to calculate the left and right trend item strengths for the left and right half-peak trend item sequences, which are respectively recorded as and , where the STL decomposition algorithm and the trend term strength calculation formula are well-known technologies and will not be described in detail in this embodiment. Thus, the leading-edge trailing affine deviation coefficient can be obtained:

[0055]

[0056]

[0057]

[0058] In the formula, represents the affine deviation coefficient of the kth Raman peak in the Raman spectrum data of the mth sweet orange flower essential oil sample, represents the number of wavenumbers in the kth Raman peak sequence in the Raman spectrum data of the mth sweet orange flower essential oil sample, represents the exponential function with natural number e as base, represents the Raman scattering intensity of the kth Raman peak sequence at the vth wave number in the Raman spectrum data of the mth sweet orange flower essential oil sample, represents the wave number corresponding to the kth Raman peak in the Raman spectrum data of the mth sweet orange flower essential oil sample, that is, the reference symmetry axis of the Raman peak, It means to calculate the wave value corresponding to the vth wave number about the reference symmetry axis. It represents the Raman scattering intensity of the kth Raman peak sequence and the vth wave number symmetric position about the reference symmetry axis in the mth sample spectrum data after noise reduction;

[0059] It represents the left and right half peak trend transformation ratio of the kth Raman peak in the mth sample spectral data. represents the logarithmic function with base 2, and Respectively represent the trend strength of the left and right half peak trend sequence of the Raman peak;

[0060] represents the leading-edge tailing affine deviation coefficient of the kth Raman peak in the mth sample spectral data, and They respectively represent that the kth Raman peak sequence in the mth sample spectral data is divided into a left half peak scattering sequence and a right half peak scattering sequence by the reference symmetry axis, represents the calculation of the Pearson correlation coefficient of the left half peak scattering sequence and the right half peak scattering sequence of the kth Raman peak in the mth sample spectral data, obtained by the known technology, is the denominator adjustment factor. In this embodiment, .

[0061] The front and tailing phenomena are the most common influencing features of impurities on Raman peaks. If the sample is seriously affected by impurities, the front or tailing phenomenon may occur, which will cause the affine of the left and right half peaks of the Raman peak to be inconsistent. That is, by referring to the symmetry axis, the affine of the left and right half peaks is greatly different, and the affine deviation coefficient is obtained. At the same time, the trend intensity deviation of the left and right half-peak Raman scattering sequences is large, and the ratio is far from 1, which makes the left and right half-peak trend transformation ratio In addition, due to the influence of the front or tail, the correlation between the left and right half-peak Raman scattering sequences is low, and finally the front and tail affine deviation coefficient is obtained. On the contrary, if the Raman peak is not affected by impurities, it tends to change smoothly, and the affine correspondence between the left and right half peaks is relatively consistent, making The value of decreases.

[0062] In addition, the influence of impurities may cause the Raman peak to have a front or tail, and the noise Raman peak intensity may change, and even the influence of the impurity molecular structure may cause new Raman peaks. Since the impurity content is generally small, there are differences in the impurity content in each sample. Therefore, the difference in Raman peaks between samples can be used to measure the overall impact of impurities on the batch of sweet orange flower essential oil, and the sample correlation deviation matrix and the main component impurity impact evaluation coefficient can be obtained:

[0063]

[0064]

[0065] In the formula, represents the sample correlation deviation coefficient between the i-th and j-th sweet orange flower essential oil samples, Indicates the maximum number of Raman peaks between the two. If the Raman peak at the corresponding position of a single sample does not exist, it will be filled with the baseline. and They represent the kth Raman peak sequence of the i-th and j-th sweet orange flower essential oil samples after denoising, It means to select the maximum value in the sequence. and They represent the affine deviation coefficients of the leading edge of the kth Raman peak of the i-th and j-th sweet orange flower essential oil samples, respectively;

[0066] Indicates the main component impurity impact assessment coefficient of this batch of sweet orange flower essential oil, The sample correlation deviation matrix of this batch of sweet orange flower essential oil is composed of the sample correlation deviation coefficients between any two sweet orange flower essential oil samples. is the sample correlation deviation matrix The element in the i-th row and j-th column in the , can be obtained by traversing all samples of the batch of sweet orange flower essential oil , Indicates the calculation of the F norm of the matrix, Indicates the total number of samples extracted from this batch of sweet orange flower essential oil.

[0067] If the batch of sweet orange flower essential oil contains more impurities, it will affect the deformation of the Raman peak of a single sample and cause large differences between the samples. As a result, there will be certain differences in the Raman spectrum data of a single Raman peak in different samples, and even when a new Raman peak appears, In addition, the Raman peak has a front tailing phenomenon due to the influence of impurities, which makes The value of is large, which eventually makes the sample correlation deviation coefficient The value of is large. Thus, all samples in this batch are traversed to obtain the sample correlation deviation matrix The value of the internal element is large, and after calculating the F norm, the main component impurity affects the evaluation coefficient. On the contrary, if the batch of sweet orange flower essential oil contains less impurities, the variation of a single Raman peak is small or even non-existent, and the Raman peaks of each sample are basically consistent, which ultimately makes The value of decreases.

[0068] In order to reduce the impact of redundant information in Raman spectra on impurity detection, the Raman spectrum data of all samples are organized into a Raman detection matrix, specifically: ,in represents the Raman detection matrix of all samples in this batch, represents the denoised Raman spectral data sequence of the first sweet orange flower essential oil sample, The Raman detection matrix is ​​used as the input of the PCA principal component analysis algorithm, where the eigenvalues ​​of the Raman detection matrix are obtained through the SVD singular value decomposition algorithm, and the principal component impurity impact evaluation coefficient is Normalization is performed, and the adaptive principal component screening threshold is specifically calculated as follows: , To obtain the eigenvalue set of the Raman detection matrix through the SVD singular value decomposition algorithm, which is greater than or equal to The eigenvalues ​​of are retained as the main components, and the remaining eigenvalues ​​are discarded. The output of the PCA algorithm is the Raman dimension reduction matrix after dimensionality reduction. The PCA principal component analysis algorithm, the SVD singular value decomposition algorithm and the normalization technique are well-known technologies and will not be described in detail in this embodiment.

[0069] In order to obtain the overall impurity situation of this batch of sweet orange flower essential oil, the Raman spectrum sequence of the standard sweet orange flower essential oil is obtained through data, and it is filled into the Raman detection matrix row by row. The same dimension as the above PCA algorithm is used to reduce the dimension of the standard Raman spectrum matrix, and the standard Raman dimension reduction matrix is ​​obtained by using the same adaptive principal component screening threshold as the above PCA algorithm. The impurity content index of sweet orange flower essential oil is obtained as follows:

[0070]

[0071] In the formula, Indicates the impurity content index of this batch of sweet orange flower essential oil. and represents the sample Raman dimension reduction matrix and the database standard Raman dimension reduction matrix, It represents the calculation of the Hadamard product of the matrices, that is, the multiplication of the corresponding elements of the matrices. If the dimensions of the two matrices are inconsistent, they are filled to the same dimension through bilinear interpolation. The bilinear interpolation method is a well-known technology and will not be described in detail in this embodiment. represents the dimension of the calculation matrix, Indicates the maximum dimension of the Raman dimension reduction matrix and the standard Raman dimension reduction matrix. The extraction process diagram of the impurity content index of sweet orange flower essential oil is as follows: Figure 2 shown.

[0072] Thus, the overall impurity content of the batch of sweet orange flower essential oil is obtained. If the impurity content is high, the adaptive principal component screening threshold is obtained. The value of is large, so the number of principal components selected is small, that is, only the most important data in the original Raman spectrum data is retained after dimensionality reduction, reducing the interference of other random components on impurity detection. Finally, the impurity content index of sweet orange essential oil obtained by comparison with the standard Raman dimensionality reduction matrix of the database is On the contrary, if the impurity content is small, the adaptive principal component screening threshold is obtained. The value of is smaller, thereby screening out more principal components, thereby retaining more detailed information in the original Raman spectrum and improving the detection accuracy.

[0073] At this point, according to the above process of this embodiment, the impurity content index of sweet orange flower essential oil can be obtained. The higher the impurity content index, the more impurities in the corresponding sweet orange flower essential oil, and the less the extracted sweet orange flower essential oil meets the quality requirements. Submit it to the technicians for reference, and provide data support for the technicians to further analyze the molecular structure of sweet orange flower essential oil. As an example, the sweet orange flower essential oil impurity content index is normalized, and the sweet orange flower essential oil impurity content index threshold is set. The value of the embodiment is 0.5. When the sweet orange flower essential oil impurity content index is higher than the impurity content index threshold, the sweet orange flower essential oil is judged to be unqualified and needs to be removed. Otherwise, the corresponding sweet orange flower essential oil is qualified.

[0074] It should be noted that the above sequence of the embodiments of the present application is for description only and does not represent the advantages and disadvantages of the embodiments. The above is a description of a specific embodiment of this specification. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0075] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

[0076] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Modifications to the technical solutions recorded in the aforementioned embodiments, or equivalent replacement of some of the technical features therein, do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A method for detecting impurities in the extraction process of sweet orange flower essential oil based on spectral characteristics, characterized in that: The method comprises the following steps: Collecting Raman spectral data of each sweet orange flower essential oil sample; Determine the affine deviation coefficient of each Raman peak in each Raman spectral data based on the wave number and Raman scattering intensity corresponding to each Raman peak in each Raman spectral data; determine the left and right half-peak trend transformation ratio of each Raman peak based on the change of Raman scattering intensity on the left and right sides of each Raman peak; determine the leading edge tailing affine deviation coefficient of each Raman peak based on the affine deviation coefficient and the left and right half-peak trend transformation ratio; Based on the leading-edge tailing affine deviation coefficient of each Raman peak of Raman spectrum data of any two sweet orange flower essential oil samples and the difference in Raman scattering intensity, the sample correlation deviation coefficient between the any two sweet orange flower essential oil samples is determined; the principal component impurity impact assessment coefficient of the sweet orange flower essential oil is determined in combination with the sample correlation deviation coefficient between all the any two sweet orange flower essential oil samples; based on the principal component impurity impact assessment coefficient, the PCA principal component analysis algorithm is used to perform dimensionality reduction processing on the Raman spectra of all sweet orange flower essential oil samples to obtain a Raman dimensionality reduction matrix; the Raman spectrum of the standard sweet orange flower essential oil is extracted, and accordingly, the standard Raman dimensionality reduction matrix is ​​obtained; Based on the similarity between the Raman dimension reduction matrix and the standard Raman dimension reduction matrix, the impurity content index of the sweet orange flower essential oil is determined, and the impurities of the sweet orange flower essential oil are judged; The method for determining the affine deviation coefficient is: Each Raman peak of each sweet orange flower essential oil Raman spectrum data is extracted by using a peak automatic detection algorithm, and the Raman scattering intensity of each Raman peak at each corresponding wave number is combined into a Raman peak sequence of each Raman peak; determining a reference symmetry axis for each Raman peak based on the wave number of each Raman peak at the peak; Determine the affine deviation coefficient of each Raman peak based on the reference symmetry axis of each Raman peak and the Raman scattering intensity of the corresponding Raman peak sequence at each wave number; The affine deviation coefficient corresponds to the calculation formula: In the formula, represents the affine deviation coefficient of the kth Raman peak in the Raman spectrum data of the mth sweet orange flower essential oil sample, represents the number of wavenumbers in the kth Raman peak sequence in the Raman spectrum data of the mth sweet orange flower essential oil sample, represents the exponential function with natural number e as base, represents the Raman scattering intensity of the kth Raman peak sequence at the vth wave number in the Raman spectrum data of the mth sweet orange flower essential oil sample, Indicates the wave number corresponding to the kth Raman peak in the Raman spectrum data of the mth sweet orange flower essential oil sample; The method for determining the left and right half peak trend conversion ratio is: Each Raman peak sequence is divided into two sequences, left and right, based on the reference symmetry axis, and recorded as the left and right half-peak scattering sequences, respectively; The STL time series decomposition algorithm is used to decompose the left and right half-peak scattering series respectively, and the trend term strength formula is used to calculate the left and right trend term strengths. The absolute value of the ratio of the intensity of the left and right trend items is taken as the true number of a logarithmic function with the number 2 as the base, and the absolute value of the calculation result of the logarithmic function is taken as the left and right half-peak trend transformation ratio of each Raman peak; The method for determining the leading edge trailing affine deviation coefficient is: Calculate the sum of the Pearson correlation coefficient of the left half peak scattering sequence and the right half peak scattering sequence of each Raman peak and the preset parameter adjustment factor, record it as the denominator, calculate the sum of the affine deviation coefficient of each Raman peak and the trend transformation ratio of the left and right half peaks, record it as the numerator; The result of dividing the numerator by the denominator is used as the leading-edge tailing affine deviation coefficient of each Raman peak; The method for determining the sample correlation deviation coefficient is: Calculate the absolute value of the difference between the maximum values ​​of the Raman peak sequence of the Raman peaks corresponding to the Raman peaks of the Raman spectrum data of any two portions of sweet orange flower essential oils, and record it as the maximum absolute value of the difference; calculate the sum of the leading-edge tailing affine deviation coefficients of the corresponding Raman peaks of the any two portions of sweet orange flower essential oils; Calculating the product of the sum value and the absolute value of the maximum difference; taking the sum of the products calculated from all corresponding Raman peaks of any two portions of sweet orange flower essential oil as the sample correlation deviation coefficient between the any two portions of sweet orange flower essential oil samples; The method for determining the main component impurity impact assessment coefficient of the sweet orange flower essential oil is: The sample association deviation coefficients between any two sweet orange flower essential oil samples are used to form a sample association deviation matrix, the F norm of the sample association deviation matrix is ​​calculated, and the result of dividing the F norm by the number of sweet orange flower essential oil samples is used as the main component impurity impact assessment coefficient of the sweet orange flower essential oil; The method of dimensionality reduction processing is: The Raman spectrum data of all sweet orange flower essential oil samples are organized into a Raman detection matrix, the eigenvalues ​​of the Raman detection matrix are extracted, and the adaptive principal component screening threshold of the PCA principal component analysis algorithm is set to the product of the principal component impurity impact assessment coefficient and the maximum value of the eigenvalue; The eigenvalues ​​greater than or equal to the product are retained as the main components, and the Raman detection matrix is ​​subjected to dimensionality reduction processing in combination with a PCA principal component analysis algorithm; The method for determining the impurity content index of the sweet orange flower essential oil is: Calculating the absolute value of the Hadamard product of the Raman dimension reduction matrix and the standard Raman dimension reduction matrix, recorded as the first absolute value, and counting the maximum dimensions of the Raman dimension reduction matrix and the standard Raman dimension reduction matrix; The ratio of the first absolute value to the maximum dimension is used as the impurity content index of sweet orange flower essential oil.

2. The impurity detection method in the sweet orange flower essential oil extraction process based on spectral characteristics as claimed in claim 1, characterized in that: The reference symmetry axis is the wave number corresponding to the peak value of each Raman peak.

Images