CN109001181B - Method for rapidly identifying type of edible oil by combining Raman spectrum typical correlation analysis - Google Patents
Method for rapidly identifying type of edible oil by combining Raman spectrum typical correlation analysis Download PDFInfo
- Publication number
- CN109001181B CN109001181B CN201810979738.5A CN201810979738A CN109001181B CN 109001181 B CN109001181 B CN 109001181B CN 201810979738 A CN201810979738 A CN 201810979738A CN 109001181 B CN109001181 B CN 109001181B
- Authority
- CN
- China
- Prior art keywords
- raman
- edible oil
- fusion
- primary
- spectrum
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/62—Systems in which the material investigated is excited whereby it emits light or causes a change in wavelength of the incident light
- G01N21/63—Systems in which the material investigated is excited whereby it emits light or causes a change in wavelength of the incident light optically excited
- G01N21/65—Raman scattering
Landscapes
- Health & Medical Sciences (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- General Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Investigating, Analyzing Materials By Fluorescence Or Luminescence (AREA)
Abstract
The invention discloses a method for rapidly identifying the type of edible oil by Raman spectrum typical correlation analysis fusion, which comprises the following steps: selecting an unknown type of edible oil sample to be identified; collecting samples at 1050-1350 cm‑1A raman spectrum of the range of the spectrum,obtaining a primary Raman spectrogram; collecting samples at 1400-1700 cm‑1Obtaining a secondary Raman spectrogram through a Raman spectrogram of the range; respectively extracting the characteristics of the acquired primary Raman spectrogram and the acquired secondary Raman spectrogram to obtain a primary Raman characteristic variable and a secondary Raman characteristic variable; performing typical correlation analysis fusion on the primary Raman characteristic variable and the secondary Raman characteristic variable to obtain a fusion Raman spectrum variable; and identifying the type of the edible oil sample of unknown type by adopting an optimized qualitative model according to the fused Raman spectrum variable of the sample. The method for rapidly identifying the type of the edible oil by combining the typical correlation analysis of the Raman spectrum is safe and rapid, is convenient and fast to detect and has high identification accuracy.
Description
Technical Field
The invention relates to the technical field of edible oil rapid identification, in particular to a method for rapidly identifying the type of edible oil by Raman spectrum typical correlation analysis fusion.
Background
The edible oil contains a plurality of important nutrient components required by human body, which are indispensable in our daily diet life, and researches show that part of the edible oil is rich in a plurality of characteristic components, including palmitic acid, oleic acid, linoleic acid and iodine value, and different kinds of edible oil have different contents of a plurality of characteristic components; in addition, the content of various characteristic components mainly determines the nutritional value of the edible oil, so the content is generally used as an important component index for evaluating the quality of the edible oil and is also an important basis for determining the commercial value of the edible oil. With the continuous increase of the price of the edible oil, in order to gain violence, many illegal merchants launch low-value oil as high-value oil to market or blend the low-value oil into the high-value oil, so that the benefits of consumers and legal production and sales enterprises are seriously harmed. Therefore, there is a need to develop a method for rapidly identifying the type of edible oil, which is important for maintaining the benefits of consumers and legal operators and maintaining the normal order of the edible oil market.
Disclosure of Invention
Aiming at the defects in the technology, the invention provides a safe, reliable, convenient and efficient method for rapidly identifying the type of edible oil by Raman spectrum typical correlation analysis fusion.
The present invention solves the technical problemsThe technical scheme adopted by the subject is as follows: a method for rapidly identifying the type of edible oil by Raman spectrum canonical correlation analysis fusion comprises the following steps: step one, selecting a sample: selecting an unknown type of edible oil sample to be identified; step two, primary spectrum acquisition: collecting an unknown type of edible oil sample at 1050-1350 cm-1Obtaining a primary Raman spectrogram through a Raman spectrogram of the range; step three, secondary spectrum acquisition: collecting an unknown edible oil sample at 1400-1700 cm-1Obtaining a secondary Raman spectrogram through a Raman spectrogram of the range; step four, spectral feature extraction: respectively extracting the characteristics of the acquired primary Raman spectrogram and the acquired secondary Raman spectrogram to obtain a primary Raman characteristic variable and a secondary Raman characteristic variable; step five, typical correlation analysis fusion: performing typical correlation analysis fusion on the primary Raman characteristic variable and the secondary Raman characteristic variable to obtain a fusion Raman spectrum variable; step six, species identification: and D, identifying the type of the edible oil sample of unknown type by adopting an optimized qualitative model according to the fused Raman spectrum variable of the edible oil sample of unknown type obtained in the step five.
Preferably, the unknown edible oil sample selected in the step one is any one of soybean oil, peanut oil, rapeseed oil, tea seed oil, rice oil, corn oil, sunflower seed oil and olive oil.
Preferably, the conditions for the primary spectrum collection in the second step are as follows: placing the edible oil sample in a flow cell, setting the temperature of the edible oil sample to be 60 +/-5 ℃, the humidity to be 30 +/-5% RH and the flow rate of the edible oil sample to be 0.8-1 cm/s; setting the resolution of the Raman spectrometer to be 4cm, the laser power to be 220mW, the integration time to be 5S, and forming an incident angle of 45 degrees by the incident direction of the spectrum acquisition probe and the flow direction of the edible oil sample; the average of three measurements was taken for each edible oil sample as the final acquired primary raman spectrum.
Preferably, the conditions for the secondary spectrum collection in the third step are as follows: placing the edible oil sample in a flow cell, setting the temperature of the edible oil sample to be 60 +/-5 ℃, the humidity to be 30 +/-5% RH and the flow rate of the edible oil sample to be 0.2-0.4 cm/s; is provided withThe resolution of the Raman spectrometer is 8cm-1The laser power is 440mW, the integration time is 10S, and the incidence direction of the spectrum acquisition probe and the flow direction of the edible oil sample form an incidence angle of 45 degrees; the average of three measurements for each edible oil sample was taken as the final collected secondary raman spectrum.
Preferably, in the fourth step, the sparse dictionary learning is adopted to extract the characteristic variables of the acquired primary raman spectrogram, and the balance error parameter is set to be 1.122Setting a weight parameter lambda to be 30/sigma, setting the sparsity L of sparse dictionary atoms to be 25, sequentially adopting an orthogonal matching pursuit method and a K-singular value decomposition method to carry out iterative optimization, setting the iteration frequency to be 2-20 times, and obtaining a primary Raman characteristic variable when the root mean square error value is minimum.
Preferably, in the fourth step, a competitive adaptive re-weighting sampling method is adopted to extract characteristic variables of the acquired secondary raman spectrogram, 10-fold partial least square method cross validation modeling is adopted when a wavelength variable subset is selected, monte carlo sampling times are set to be 1-100 times, and when the root mean square error value of a partial least square method cross validation model is minimum, the secondary raman characteristic variables are obtained.
Preferably, the method for analyzing fusion by canonical correlation in the fifth step is as follows: extracting two groups of different feature vectors of the primary Raman feature variable and the secondary Raman feature variable to form sample spaces A and B; computing an overall covariance matrix S for sample spaces A and Baa、SbbAnd a cross-covariance matrix S between A and Bab(ii) a Calculating non-zero eigenvalues of the discrimination criterion matrix and sequencing to obtain a typical projection vector; and (5) utilizing the typical projection vector to form a transformation matrix to extract the characteristics, and obtaining the fusion Raman spectrum variable.
Preferably, the method for establishing the optimized qualitative model in the sixth step is as follows: collecting a number of different known types of edible oil samples, the different known types of edible oil samples comprising: soybean oil, peanut oil, rapeseed oil, tea seed oil, rice oil, corn oil, sunflower seed oil and olive oil; collecting primary Raman spectrograms and secondary Raman spectrograms of a plurality of different known edible oil samples, wherein the collecting conditions of the primary Raman spectrograms are described in the step two, and the collecting conditions of the secondary Raman spectrograms are described in the step three; respectively performing characteristic extraction on the primary Raman spectrogram and the secondary Raman spectrogram of a plurality of collected edible oil samples of different known types to obtain primary Raman characteristic variables and secondary Raman characteristic variables of the edible oil samples of different known types, wherein the method for extracting the spectral characteristics is as described in the fourth step; performing typical correlation analysis and fusion on the primary Raman characteristic variables and the secondary Raman characteristic variables of a plurality of different known types of edible oil samples to obtain fusion Raman spectrum variables of the plurality of different known types of edible oil samples, wherein the typical correlation analysis and fusion method is as described in the fifth step; the method comprises the steps of taking the fusion Raman spectrum variables of a plurality of edible oil samples of different known types as input variables of a qualitative model, establishing the qualitative model of the edible oil samples of the different known types by a multi-core learning support vector machine classification method, and optimizing parameters in the qualitative model by adopting a particle swarm optimization algorithm to obtain an optimized qualitative model.
Preferably, fusion Raman spectrum variables of a plurality of different known types of edible oil samples are used as input variables of the qualitative model, data of the fusion Raman spectrum variables are divided into 10 groups according to characteristic dimensions by a multi-core learning support vector machine classification method, and the 10 groups establish the single-core support vector machine qualitative model by using a Gaussian core; and optimizing a penalty factor matrix [ C ] and a kernel function parameter matrix [ g ] in the qualitative model by adopting a particle swarm optimization algorithm, wherein [ C ] and [ g ] are both matrixes of 10x8, the number of population particles is set to be 25 during optimization, the dimension of each particle is 2, the iterative evolution frequency is 150, the initial value of a learning factor is set to be C1 to be 1, and C2 to be 2, so that 10 groups of parameters ([ C ], [ g ]) are obtained, and the 10 groups of parameters ([ C ], [ g ]) are subjected to weighted voting, so that the optimized qualitative model is obtained.
Preferably, the first and second light sources are, preferably,
when [ C ] is equal to
[g] Is equal to
And the weight is [0.0235, 0.0582, 0.1417, 0.1329, 0.0622, 0.0381, 0.0940, 0.1248, 0.1193, 0.0874], an optimized qualitative model is obtained.
Compared with the prior art, the invention has the beneficial effects that: according to the edible oil type rapid identification method based on Raman spectrum typical correlation analysis fusion, the influence of external conditions on a Raman spectrogram can be effectively eliminated by adopting primary spectrum collection and secondary spectrum collection in different ranges, setting the same temperature, humidity and different flow rate of an edible oil sample during the secondary spectrum collection, and setting different resolutions, laser power, integration time and the same incident angle of a Raman spectrometer during the secondary spectrum collection; by fusing the primary Raman characteristic variable and the secondary Raman characteristic variable through typical correlation analysis, useless variables in the spectrogram can be effectively compressed, and effective information can be highlighted; by adopting a multi-core learning support vector machine classification method and combining a particle swarm optimization algorithm to establish an optimized qualitative model, the identification accuracy can be remarkably improved.
Drawings
FIG. 1 is a flow chart of the method for rapidly identifying the type of edible oil by Raman spectrum typical correlation analysis fusion.
Detailed Description
The present invention is further described in detail below with reference to the attached drawings so that those skilled in the art can implement the invention by referring to the description text.
As shown in FIG. 1, the invention provides a method for rapidly identifying the type of edible oil by combining Raman spectrum typical correlation analysis, which comprises the following steps:
selecting an unknown type edible oil sample to be identified, wherein the selected unknown type edible oil sample is any one of soybean oil, peanut oil, rapeseed oil, tea seed oil, rice oil, corn oil, sunflower seed oil and olive oil;
step two,Placing the edible oil sample in a flow cell, setting the temperature of the edible oil sample to be 60 +/-5 ℃, the humidity to be 30 +/-5% RH and the flow rate of the edible oil sample to be 0.8-1 cm/s; setting the resolution of the Raman spectrometer to be 4cm, the laser power to be 220mW, the integration time to be 5S, and forming an incident angle of 45 degrees by the incident direction of the spectrum acquisition probe and the flow direction of the edible oil sample; collecting an unknown type of edible oil sample at 1050-1350 cm-1(ii) a raman spectrogram of the range, wherein the average value of three measurements taken for each edible oil sample is taken as a finally acquired primary raman spectrogram;
placing the edible oil sample in a flow cell, setting the temperature of the edible oil sample to be 60 +/-5 ℃, the humidity to be 30 +/-5% RH and the flow rate of the edible oil sample to be 0.2-0.4 cm/s; setting the resolution of the Raman spectrometer to be 8cm-1The laser power is 440mW, the integration time is 10S, and the incidence direction of the spectrum acquisition probe and the flow direction of the edible oil sample form an incidence angle of 45 degrees; collecting an unknown edible oil sample at 1400-1700 cm-1(ii) a raman spectrogram of the range, wherein the average value of three measurements taken for each edible oil sample is taken as a finally acquired secondary raman spectrogram;
step four, extracting characteristic variables of the acquired primary Raman spectrogram by adopting sparse dictionary learning, and setting a balance error parameter to be 1.122Setting a weight parameter lambda to be 30/sigma, setting the sparsity L of sparse dictionary atoms to be 25, sequentially adopting an orthogonal matching pursuit method and a K-singular value decomposition method to carry out iterative optimization, setting the iteration frequency to be 2-20 times, and obtaining a primary Raman characteristic variable when a root mean square error value is minimum; extracting characteristic variables of the acquired secondary Raman spectrogram by adopting a competitive adaptive reweighting sampling method, selecting a wavelength variable subset, performing cross validation modeling by adopting a 10-fold partial least square method, setting the Monte Carlo sampling times to be 1-100 times, and when the root mean square error value of the cross validation model by the partial least square method is minimum, acquiring secondary Raman characteristic variables;
and fifthly, performing typical correlation analysis and fusion on the primary Raman characteristic variable and the secondary Raman characteristic variable, and extracting the primary Raman characteristic variable and the secondary Raman characteristic variableTwo groups of different feature vectors form sample spaces A and B; computing an overall covariance matrix S for sample spaces A and Baa、SbbAnd a cross-covariance matrix S between A and Bab(ii) a Calculating non-zero eigenvalues of the discrimination criterion matrix and sequencing to obtain a typical projection vector; utilizing typical projection vectors to form a transformation matrix to extract features, and obtaining a fusion Raman spectrum variable;
sixthly, identifying the type of the edible oil sample of unknown type by adopting an optimized qualitative model according to the fusion Raman spectrum variable of the edible oil sample of unknown type obtained in the fifth step;
the method for establishing the optimized qualitative model comprises the following steps:
collecting a number of different known types of edible oil samples, the different known types of edible oil samples comprising: soybean oil, peanut oil, rapeseed oil, tea seed oil, rice oil, corn oil, sunflower seed oil and olive oil;
collecting primary Raman spectrograms and secondary Raman spectrograms of a plurality of different known edible oil samples, wherein the collecting conditions of the primary Raman spectrograms are described in the step two, and the collecting conditions of the secondary Raman spectrograms are described in the step three;
respectively performing characteristic extraction on the primary Raman spectrogram and the secondary Raman spectrogram of a plurality of collected edible oil samples of different known types to obtain primary Raman characteristic variables and secondary Raman characteristic variables of the edible oil samples of different known types, wherein the method for extracting the spectral characteristics is as described in the fourth step;
performing typical correlation analysis and fusion on the primary Raman characteristic variables and the secondary Raman characteristic variables of a plurality of different known types of edible oil samples to obtain fusion Raman spectrum variables of the plurality of different known types of edible oil samples, wherein the typical correlation analysis and fusion method is as described in the fifth step;
taking the fusion Raman spectrum variables of a plurality of different known edible oil samples as input variables of a qualitative model, dividing the data of the fusion Raman spectrum variables into 10 groups according to characteristic dimensions by a multi-core learning support vector machine classification method, and establishing a single-core support vector machine qualitative model by using Gaussian cores in the 10 groups; optimizing a penalty factor matrix [ C ] and a kernel function parameter matrix [ g ] in a qualitative model by adopting a particle swarm optimization algorithm, wherein [ C ] and [ g ] are both matrixes of 10x8, the number of population particles is set to be 25 during optimization, the dimension of each particle is 2, the iterative evolution frequency is 150, the initial value of a learning factor is set to be C1 to be 1, C2 to be 2, 10 groups of parameters ([ C ], [ g ]) are obtained, and the 10 groups of parameters ([ C ], [ g ]) are subjected to weighted voting and optimized,
when [ C ] is equal to
[g] Is equal to
And the weight is [0.0235, 0.0582, 0.1417, 0.1329, 0.0622, 0.0381, 0.0940, 0.1248, 0.1193, 0.0874], an optimized qualitative model is obtained.
Examples
1. Sample selection
468 parts of soybean oil, peanut oil, rapeseed oil, tea seed oil, rice oil, corn oil, sunflower seed oil and olive oil 8-class edible oil samples are collected, 351 parts of training set edible oil samples and 117 parts of prediction set edible oil samples are selected according to the ratio of 3: 1 by adopting an SPXY algorithm; wherein, the number distribution of the prediction set samples and the training set samples of the 8 types of edible oil samples is shown in the following table 1.
TABLE 1
2. Primary spectrum collection
Placing the edible oil sample in a flow cell, setting the temperature of the edible oil sample at 60 ℃, the humidity at 30% RH and the flow rate of the edible oil sample at 0.9 cm/s; setting the resolution of the Raman spectrometer to be 4cm-1The laser power is 220mW, the integration time is 5S, and the incidence direction of the spectrum acquisition probe and the flow direction of the edible oil sample form an incidence angle of 45 degrees; collecting 468 parts of edible oil sample at 1050-1350 cm-1Raman spectra of the range, the average of three measurements taken for each edible oil sample was taken as the final primary raman spectrum collected.
3. Secondary spectrum collection
Placing the edible oil sample in a flow cell, setting the temperature of the edible oil sample at 60 ℃, the humidity at 30% RH and the flow rate of the edible oil sample at 0.3 cm/s; setting the resolution of the Raman spectrometer to be 8cm-1The laser power is 440mW, the integration time is 10S, and the incidence direction of the spectrum acquisition probe and the flow direction of the edible oil sample form an incidence angle of 45 degrees; collecting 468 parts of edible oil sample at 1400-1700 cm-1(ii) raman spectra of the ranges, taking the average of three measurements for each edible oil sample as the final collected secondary raman spectra.
4. Spectral feature extraction
Performing characteristic variable extraction on the primary Raman spectrogram of 351 edible oil samples in the acquired training set by adopting sparse dictionary learning, and setting a balance error parameter as 1.122Setting a weight parameter lambda to be 30/sigma, setting the sparsity L of sparse dictionary atoms to be 25, sequentially performing iterative optimization by adopting an orthogonal matching pursuit method and a K-singular value decomposition method, setting the iteration frequency to be 2-20 times, obtaining a minimum root mean square error value of 0.1683 when sigma is 28 and the iteration frequency is 12 times, and extracting primary Raman feature variables to be 39;
extracting characteristic variables of secondary Raman spectrograms of 351 edible oil samples in a training set acquired by adopting a competitive adaptive re-weighting sampling method, selecting a wavelength variable subset, adopting a 10-fold partial least square method cross validation modeling, setting the Monte Carlo sampling frequency to be 1-100 times, and when the Monte Carlo sampling frequency is the 58 th time, minimizing the root mean square error value of the partial least square method cross validation model to 0.3429, and extracting to obtain 46 secondary Raman characteristic variables.
5. Canonical correlation analysis fusion
Performing typical correlation analysis and fusion on 32 primary Raman characteristic variables and 46 secondary Raman characteristic variables of 351 edible oil samples in a training set, and extracting two groups of different characteristic vectors of the primary Raman characteristic variables and the secondary Raman characteristic variables to form sample spaces A and B; computing an overall covariance matrix S for sample spaces A and Baa、SbbAnd a cross-covariance matrix S between A and Bab(ii) a Calculating non-zero eigenvalues of the discrimination criterion matrix and ordering, wherein eigenvalues larger than 0.001 are respectively 0.9997, 0.7912, 0.5437, 0.1974, 0.1675, 0.0763, 0.0542, 0.0270, 0.0238, 0.0192, 0.0157, 0.0113, 0.0081, 0.0050, 0.0049, 0.0038, 0.0029, 0.0025, 0.0024, 0.0021, 0.0018, 0.0016, 0.0015, 0.0014, 0.0013, 0.0012, 0.0011 and 0.0010, and obtaining a typical projection vector; and (4) utilizing the typical projection vectors to form a transformation matrix to extract the characteristics, and obtaining 28 fused Raman spectrum variables.
6. Optimized qualitative model building
Taking 28 fused Raman spectrum variables of 351 edible oil samples in a training set as input variables of a qualitative model, dividing the 28 fused Raman spectrum variables into 10 groups according to characteristic dimensions by a multi-core learning support vector machine classification method, and establishing a single-core support vector machine qualitative model by using Gaussian cores in the 10 groups; optimizing a penalty factor matrix [ C ] and a kernel function parameter matrix [ g ] in a qualitative model by adopting a particle swarm optimization algorithm, wherein [ C ] and [ g ] are both matrixes of 10x8, the number of population particles is set to be 25 during optimization, the dimension of each particle is 2, the iterative evolution frequency is 150, the initial value of a learning factor is set to be C1 to be 1, C2 to be 2, 10 groups of parameters ([ C ], [ g ]) are obtained, and the 10 groups of parameters ([ C ], [ g ]) are subjected to weighted voting and optimized,
when [ C ] is equal to
[g] Is equal to
And when the weight is [0.0235, 0.0582, 0.1417, 0.1329, 0.0622, 0.0381, 0.0940, 0.1248, 0.1193, 0.0874], an optimized qualitative model is obtained, and the identification accuracy of 351 edible oil samples in the training set is 96.87%.
7. Species discrimination
And (3) taking the fused Raman spectrum variable (the fusion method is described in the step five) of 157 edible oil samples in the prediction set as an input variable of a qualitative model, adopting the penalty factor matrix [ C ] and the kernel function parameter matrix [ g ] which are obtained through optimization, and carrying out prediction identification on the types of 157 edible oil samples in the prediction set by using the optimized qualitative model with the weight of [0.0235, 0.0582, 0.1417, 0.1329, 0.0622, 0.0381, 0.0940, 0.1248, 0.1193 and 0.0874], wherein the prediction identification accuracy is 91.08%.
While embodiments of the invention have been described above, it is not limited to the applications set forth in the description and the embodiments, which are fully applicable in various fields of endeavor to which the invention pertains, and further modifications may readily be made by those skilled in the art, it being understood that the invention is not limited to the details shown and described herein without departing from the general concept defined by the appended claims and their equivalents.
Claims (5)
1. A method for rapidly identifying the type of edible oil by Raman spectrum canonical correlation analysis fusion is characterized by comprising the following steps:
step one, selecting a sample: selecting an unknown type of edible oil sample to be identified;
step two, primary spectrum acquisition: miningCollecting the unknown type of edible oil sample at 1050-1350 cm-1Obtaining a primary Raman spectrogram through a Raman spectrogram of the range;
the conditions for primary spectrum acquisition were as follows:
placing the edible oil sample in a flow cell, setting the temperature of the edible oil sample to be 60 +/-5 ℃, the humidity to be 30 +/-5% RH and the flow rate of the edible oil sample to be 0.8-1 cm/s;
setting the resolution of the Raman spectrometer to be 4cm-1The laser power is 220mW, the integration time is 5S, and the incidence direction of the spectrum acquisition probe and the flow direction of the edible oil sample form an incidence angle of 45 degrees;
taking the average value of three measurements of each edible oil sample as a finally collected primary Raman spectrogram;
step three, secondary spectrum acquisition: collecting an unknown edible oil sample at 1400-1700 cm-1Obtaining a secondary Raman spectrogram through a Raman spectrogram of the range;
the conditions for secondary spectrum collection were as follows:
placing the edible oil sample in a flow cell, setting the temperature of the edible oil sample to be 60 +/-5 ℃, the humidity to be 30 +/-5% RH and the flow rate of the edible oil sample to be 0.2-0.4 cm/s;
setting the resolution of the Raman spectrometer to be 8cm-1The laser power is 440mW, the integration time is 10S, and the incidence direction of the spectrum acquisition probe and the flow direction of the edible oil sample form an incidence angle of 45 degrees;
taking the average value of three measurements of each edible oil sample as a finally collected secondary Raman spectrogram;
step four, spectral feature extraction: respectively extracting the characteristics of the acquired primary Raman spectrogram and the acquired secondary Raman spectrogram to obtain a primary Raman characteristic variable and a secondary Raman characteristic variable;
step five, typical correlation analysis fusion: performing typical correlation analysis fusion on the primary Raman characteristic variable and the secondary Raman characteristic variable to obtain a fusion Raman spectrum variable;
step six, species identification: according to the fusion Raman spectrum variable of the edible oil sample of unknown type obtained in the fifth step, performing type identification on the edible oil sample of unknown type by adopting an optimized qualitative model;
the establishment method of the optimized qualitative model in the sixth step comprises the following steps:
collecting a number of different known types of edible oil samples, the different known types of edible oil samples comprising: soybean oil, peanut oil, rapeseed oil, tea seed oil, rice oil, corn oil, sunflower seed oil and olive oil;
collecting primary Raman spectrograms and secondary Raman spectrograms of a plurality of different known edible oil samples, wherein the collecting conditions of the primary Raman spectrograms are described in the step two, and the collecting conditions of the secondary Raman spectrograms are described in the step three;
respectively performing characteristic extraction on the primary Raman spectrogram and the secondary Raman spectrogram of a plurality of collected edible oil samples of different known types to obtain primary Raman characteristic variables and secondary Raman characteristic variables of the edible oil samples of different known types, wherein the method for extracting the spectral characteristics is as described in the fourth step;
performing typical correlation analysis and fusion on the primary Raman characteristic variables and the secondary Raman characteristic variables of a plurality of different known types of edible oil samples to obtain fusion Raman spectrum variables of the plurality of different known types of edible oil samples, wherein the typical correlation analysis and fusion method is as described in the fifth step;
the method comprises the steps of taking a fusion Raman spectrum variable of a plurality of edible oil samples of different known types as an input variable of a qualitative model, establishing the qualitative model of the edible oil samples of the different known types by a multi-core learning support vector machine classification method, and optimizing parameters in the qualitative model by adopting a particle swarm optimization algorithm to obtain an optimized qualitative model;
taking the fusion Raman spectrum variables of a plurality of different known edible oil samples as input variables of a qualitative model, dividing the data of the fusion Raman spectrum variables into 10 groups according to characteristic dimensions by a multi-core learning support vector machine classification method, and establishing a single-core support vector machine qualitative model by using Gaussian cores in the 10 groups; optimizing a penalty factor matrix [ C ] and a kernel function parameter matrix [ g ] in a qualitative model by adopting a particle swarm optimization algorithm, wherein [ C ] and [ g ] are both matrixes of 10x8, the number of population particles is set to be 25 during optimization, the dimension of each particle is 2, the iterative evolution frequency is 150, the initial value of a learning factor is set to be C1 to be 1, C2 to be 2, 10 groups of parameters ([ C ], [ g ]) are obtained, and the 10 groups of parameters ([ C ], [ g ]) are subjected to weighted voting, so that an optimized qualitative model is obtained;
after the optimization, the method has the advantages of high efficiency,
when [ C ] is equal to
[g] Is equal to
And the weight is [0.0235, 0.0582, 0.1417, 0.1329, 0.0622, 0.0381, 0.0940, 0.1248, 0.1193, 0.0874], an optimized qualitative model is obtained.
2. The method for rapidly identifying the type of edible oil through raman spectroscopy canonical correlation analysis fusion according to claim 1, wherein the edible oil sample of unknown type selected in the first step is any one of soybean oil, peanut oil, rapeseed oil, tea seed oil, rice oil, corn oil, sunflower seed oil, and olive oil.
3. The method for rapidly identifying the type of the edible oil through Raman spectrum typical correlation analysis fusion as claimed in claim 1, wherein in the fourth step, sparse dictionary learning is adopted to extract characteristic variables of the acquired primary Raman spectrogram, and a balance error parameter is set to be 1.122Setting a weight parameter lambda to be 30/sigma, setting the sparsity L of sparse dictionary atoms to be 25, sequentially adopting an orthogonal matching pursuit method and a K-singular value decomposition method to carry out iterative optimization, setting the iteration frequency to be 2-20 times, and obtaining a primary Raman characteristic variable when the root mean square error value is minimum.
4. The method for rapidly identifying the types of the edible oil through Raman spectrum typical correlation analysis fusion as claimed in claim 1, wherein a competitive adaptive re-weighting sampling method is adopted in the fourth step to extract characteristic variables of the acquired secondary Raman spectrogram, 10-fold partial least square method cross validation modeling is adopted when a wavelength variable subset is selected, Monte Carlo sampling times are set to be 1-100 times, and when a root mean square error value of a partial least square method cross validation model is minimum, the secondary Raman characteristic variables are obtained.
5. The method for rapidly identifying the type of edible oil fused with Raman spectrum canonical correlation analysis according to claim 1, wherein the canonical correlation analysis fusion method in the fifth step is as follows:
extracting two groups of different feature vectors of the primary Raman feature variable and the secondary Raman feature variable to form sample spaces A and B;
computing an overall covariance matrix S for sample spaces A and Baa、SbbAnd a cross-covariance matrix S between A and Bab；
Calculating non-zero eigenvalues of the discrimination criterion matrix and sequencing to obtain a typical projection vector;
and (5) utilizing the typical projection vector to form a transformation matrix to extract the characteristics, and obtaining the fusion Raman spectrum variable.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810979738.5A CN109001181B (en) | 2018-08-24 | 2018-08-24 | Method for rapidly identifying type of edible oil by combining Raman spectrum typical correlation analysis |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810979738.5A CN109001181B (en) | 2018-08-24 | 2018-08-24 | Method for rapidly identifying type of edible oil by combining Raman spectrum typical correlation analysis |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109001181A CN109001181A (en) | 2018-12-14 |
| CN109001181B true CN109001181B (en) | 2020-11-03 |
Family
ID=64592274
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810979738.5A Active CN109001181B (en) | 2018-08-24 | 2018-08-24 | Method for rapidly identifying type of edible oil by combining Raman spectrum typical correlation analysis |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109001181B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109993054A (en) * | 2019-02-14 | 2019-07-09 | 广东电网有限责任公司 | A kind of transformer insulation oil category identification method, apparatus, equipment and storage medium |
| JP2020201174A (en) * | 2019-06-12 | 2020-12-17 | 国立研究開発法人物質・材料研究機構 | Component identification device for spectrum analyzer, method thereof, and computer program |
| CN114113035B (en) * | 2021-11-18 | 2024-02-02 | 北京理工大学 | Identification method of transgenic soybean oil |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060160065A1 (en) * | 1998-06-01 | 2006-07-20 | Roger Timmis | Method for classifying plant embryos using Raman spectroscopy |
| CN101788459B (en) * | 2010-02-08 | 2012-07-18 | 暨南大学 | Quasi-continuous spectroscopic wavelength combination method |
| CN102495343B (en) * | 2011-12-30 | 2014-04-16 | 重庆大学 | Partial discharge detection identification method based on ultrasound and ultraviolet information fusion and system thereof |
| CN102779240B (en) * | 2012-06-21 | 2015-05-20 | 哈尔滨工程大学 | Inherent irregular protein structure forecasting method based on kernel canonical correlation analysis |
| CN103940800B (en) * | 2014-03-10 | 2016-08-17 | 北京理工大学 | Confocal laser Brillouin-method for measuring Raman spectrum and device |
-
2018
- 2018-08-24 CN CN201810979738.5A patent/CN109001181B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN109001181A (en) | 2018-12-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Li et al. | A combination of chemometrics methods and GC–MS for the classification of edible vegetable oils | |
| CN109001181B (en) | Method for rapidly identifying type of edible oil by combining Raman spectrum typical correlation analysis | |
| CN108362659B (en) | Edible oil type rapid identification method based on multi-source spectrum parallel fusion | |
| CN103534578B (en) | Determine spectral device and the method for composition present in sample | |
| CN101819141B (en) | Maize variety identification method based on near infrared spectrum and information processing | |
| CN105486655B (en) | The soil organism rapid detection method of model is intelligently identified based on infrared spectroscopy | |
| Lamanna et al. | Geographical origin of durum wheat studied by 1H‐NMR profiling | |
| CN101738373A (en) | Method for distinguishing varieties of crop seeds | |
| CN104374739A (en) | Identification method for authenticity of varieties of seeds on basis of near-infrared quantitative analysis | |
| CN110361356A (en) | A kind of near infrared spectrum Variable Selection improving wheat water content precision of prediction | |
| Wang et al. | Extraction and classification of origin characteristic peaks from rice Raman spectra by principal component analysis | |
| CN109409350A (en) | A kind of Wavelength selecting method based on PCA modeling reaction type load weighting | |
| Lu et al. | Fluorescence hyperspectral image technique coupled with HSI method to predict solanine content of potatoes | |
| CN112098358A (en) | Near infrared spectrum parallel fusion quantitative modeling method based on quaternion convolution neural network | |
| CN109142265B (en) | Method for rapidly identifying type of edible oil by near infrared spectrum wavelet fusion | |
| CN103278467A (en) | Rapid nondestructive high-accuracy method with for identifying abundance degree of nitrogen element in plant leaf | |
| CN109001148B (en) | Tea seed oil multi-component adulteration detection method based on near infrared spectrum typical correlation analysis fusion | |
| CN108303406B (en) | Camellia seed oil adulteration detection method based on Raman spectrum | |
| CN108267424B (en) | Method for rapidly identifying type of edible oil based on multiple characteristic components | |
| CN108613965B (en) | Method for measuring whey protein in dairy product based on data-driven Raman spectrum | |
| CN108509840B (en) | Hyperspectral remote sensing image waveband selection method based on quantum memory optimization mechanism | |
| CN108169204B (en) | Raman spectrum preprocessing method based on database | |
| CN106950193B (en) | Based on the near infrared spectrum Variable Selection from weight variable combination cluster analysis | |
| CN115795225A (en) | Method and device for screening near infrared spectrum correction set | |
| CN105067558A (en) | Infrared qualitative discrimination feature extraction method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |