CN112509643B - Quantitative analysis model construction method, quantitative analysis method, device and system - Google Patents
Quantitative analysis model construction method, quantitative analysis method, device and system Download PDFInfo
- Publication number
- CN112509643B CN112509643B CN202110146449.9A CN202110146449A CN112509643B CN 112509643 B CN112509643 B CN 112509643B CN 202110146449 A CN202110146449 A CN 202110146449A CN 112509643 B CN112509643 B CN 112509643B
- Authority
- CN
- China
- Prior art keywords
- quantitative analysis
- analysis model
- sample
- sub
- 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
- 238000004445 quantitative analysis Methods 0.000 title claims abstract description 260
- 238000000034 method Methods 0.000 title claims abstract description 92
- 238000010276 construction Methods 0.000 title claims abstract description 38
- 238000001228 spectrum Methods 0.000 claims abstract description 94
- 239000011159 matrix material Substances 0.000 claims abstract description 44
- 230000003595 spectral effect Effects 0.000 claims abstract description 41
- 238000012549 training Methods 0.000 claims abstract description 30
- 238000012417 linear regression Methods 0.000 claims abstract description 21
- 238000002790 cross-validation Methods 0.000 claims description 9
- 238000004364 calculation method Methods 0.000 claims description 3
- 238000001514 detection method Methods 0.000 abstract description 14
- 230000000694 effects Effects 0.000 abstract description 4
- 238000012795 verification Methods 0.000 description 17
- 235000019750 Crude protein Nutrition 0.000 description 9
- 238000002329 infrared spectrum Methods 0.000 description 9
- 230000008569 process Effects 0.000 description 9
- 235000019764 Soybean Meal Nutrition 0.000 description 8
- 238000007781 pre-processing Methods 0.000 description 8
- 239000004455 soybean meal Substances 0.000 description 8
- 238000012545 processing Methods 0.000 description 6
- 238000012986 modification Methods 0.000 description 5
- 230000004048 modification Effects 0.000 description 5
- 238000004458 analytical method Methods 0.000 description 4
- 238000012937 correction Methods 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 230000002452 interceptive effect Effects 0.000 description 4
- 238000004891 communication Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000010183 spectrum analysis Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 2
- 238000011835 investigation Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 239000002994 raw material Substances 0.000 description 2
- 239000000126 substance Substances 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- 238000010200 validation analysis Methods 0.000 description 2
- 238000004497 NIR spectroscopy Methods 0.000 description 1
- 235000002637 Nicotiana tabacum Nutrition 0.000 description 1
- 244000061176 Nicotiana tabacum Species 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 239000006227 byproduct Substances 0.000 description 1
- 239000003153 chemical reaction reagent Substances 0.000 description 1
- 238000012569 chemometric method Methods 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 229940079593 drug Drugs 0.000 description 1
- 239000003814 drug Substances 0.000 description 1
- 235000013601 eggs Nutrition 0.000 description 1
- 238000004836 empirical method Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 235000013305 food Nutrition 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 239000003348 petrochemical agent Substances 0.000 description 1
- 238000002203 pretreatment Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
- G16C20/00—Chemoinformatics, i.e. ICT specially adapted for the handling of physicochemical or structural data of chemical particles, elements, compounds or mixtures
- G16C20/20—Identification of molecular entities, parts thereof or of chemical compositions
-
- 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
- G01N21/17—Systems in which incident light is modified in accordance with the properties of the material investigated
- G01N21/25—Colour; Spectral properties, i.e. comparison of effect of material on the light at two or more different wavelengths or wavelength bands
- G01N21/31—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry
- G01N21/35—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light
- G01N21/359—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light using near infrared light
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
- G16C20/00—Chemoinformatics, i.e. ICT specially adapted for the handling of physicochemical or structural data of chemical particles, elements, compounds or mixtures
- G16C20/70—Machine learning, data mining or chemometrics
Landscapes
- Physics & Mathematics (AREA)
- Spectroscopy & Molecular Physics (AREA)
- Chemical & Material Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Crystallography & Structural Chemistry (AREA)
- Theoretical Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computing Systems (AREA)
- Evolutionary Computation (AREA)
- Investigating Or Analysing Materials By Optical Means (AREA)
- Databases & Information Systems (AREA)
- Software Systems (AREA)
- Medical Informatics (AREA)
- Data Mining & Analysis (AREA)
- Immunology (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Pathology (AREA)
- Artificial Intelligence (AREA)
Abstract
The invention provides a quantitative analysis model construction method, a quantitative analysis method, a device and a system, wherein the construction method comprises the following steps: training based on a preset linear regression method and a sample set to obtain a first quantitative analysis model; the sample set comprises the target component content of each sample and the spectrum matrix of all samples; dividing a coverage range of the content of the target component into k continuous and non-overlapping sub-coverage ranges, dividing a sample set into k sub-samples based on the k sub-coverage ranges and corresponding spectral matrixes, wherein k is more than or equal to 2; respectively training and obtaining k second quantitative analysis models based on a preset nonlinear regression method and each subsample; according to the quantitative analysis model construction method, when the quantitative analysis model is constructed, the first quantitative analysis model is constructed on the basis of the linear regression method to cover the whole detection range, the k second quantitative analysis models are constructed on the basis of the nonlinear regression method to respectively cover the part of the detection range, and the method has good prediction effect and accuracy for the globally linear but locally nonlinear samples.
Description
Technical Field
The invention belongs to the technical field of spectral analysis and detection, and particularly relates to a quantitative analysis model construction method, a quantitative analysis method, a device and a system.
Background
The spectral analysis technology can utilize the optical characteristics of organic substances in a molecular spectrum region to quickly analyze the content of one or more chemical components in a sample, has the characteristics of high analysis speed, no need of sample pretreatment, no chemical reagent consumption, multi-component synchronous analysis and the like, and is widely applied to the fields of agriculture, petrochemicals, medicines, tobacco, food and the like. However, this technique is an empirical method and requires the construction of a quantitative analysis model by means of chemometric methods. The construction of a quantitative analysis model is one of the core techniques of spectral analysis.
A common method for constructing a quantitative analysis model is to construct a quantitative analysis model by a regression method. The regression method includes a linear regression method and a nonlinear regression method. Unlike linear regression methods, these non-linear regression methods have more than one parameter corresponding to a characteristic factor.
The distribution of the real samples has global linearity, but under the condition of large disturbance of local intervals, the distribution has non-linearity, especially for feed raw materials derived from byproducts of grain and oil industry, such as soybean meal and the like. The content range of the crude protein in the soybean meal is 41-48%, but the crude protein content of the soybean meal and the near infrared spectrum show a nonlinear relation in a specific content range (such as a crude protein 43% specification, a crude protein 46% specification and the like) and a linear relation in the whole content range under the regulation and control of factory quality.
In order to improve the accuracy of the quantitative analysis model of the materials, effective modeling strategies and methods are urgently needed to be developed.
Disclosure of Invention
In order to solve the technical problems, the invention provides a quantitative analysis model construction method, a quantitative analysis method, a device and a system, which can effectively improve the prediction accuracy in all detection ranges.
The technical scheme provided by the invention is as follows:
in a first aspect, a method for constructing a quantitative analysis model is provided, the method comprising the following steps:
training based on a preset linear regression method and a preset sample set to obtain a first quantitative analysis model; the sample set comprises the target component content of each sample and the spectrum matrix of all samples;
dividing the coverage range of the target component content into k continuous and non-overlapping sub-coverage ranges, and dividing the sample set into k sub-samples based on the k sub-coverage ranges and corresponding spectral matrixes, wherein k is more than or equal to 2;
and respectively training and obtaining k second quantitative analysis models based on a preset nonlinear regression method and each subsample.
In a preferred embodiment, the dividing the coverage of the target component content into k consecutive and non-overlapping sub-coverage includes:
obtaining the number k of the second quantitative analysis models according to the coverage range of the sample set and the cross validation root-mean-square error of the first quantitative analysis model;
the coverage of the sample set is equally divided into k consecutive and non-overlapping k sub-coverage.
In a preferred embodiment, the k value is the largest integer value that satisfies the following formula:
2≤k≤(ymax-ymin)/(8×RMSECV)
wherein, ymax-yminIs the coverage of the content of the target component, ymaxIs the maximum content of the target component, yminIs the minimum content of the target component; RMSECV represents the cross-validation root mean square error of the first quantitative analysis model.
In a preferred embodiment, the dividing the sample set into k sub-samples based on the k sub-coverage ranges and the corresponding spectral matrices includes:
correspondingly obtaining a corresponding sub-spectrum matrix according to the content of the target component in each sub-coverage range;
and constructing corresponding k sub-samples based on the k sub-coverage ranges and the corresponding sub-spectrum matrixes.
In a preferred embodiment, the constructing method further includes: obtaining a preset sample set, comprising the following substeps:
collecting n samples containing target components, and obtaining the content of the target components in each sample, wherein n is more than or equal to 3;
obtaining a spectral matrix X (n, m) of the sample based on the spectra of all samples, wherein m represents the number of spectral wavelengths;
constructing the sample set based on the target component content in each sample and a spectral matrix X (n, m).
In a second aspect, a quantitative analysis method based on the quantitative analysis model is provided, and the quantitative analysis method comprises:
inputting the spectrum of the sample to be tested into the first quantitative analysis model to obtain a first predicted value of the content of the target component;
determining a second quantitative analysis model from k second quantitative analysis models corresponding to the first quantitative analysis model according to the first predicted value;
and inputting the spectrum of the sample to be detected into a corresponding second quantitative analysis model to obtain the content of the target component in the sample to be detected.
In a preferred embodiment, the determining a second quantitative analysis model from the k second quantitative analysis models corresponding to the first quantitative analysis model according to the first predicted value includes the following sub-steps:
selecting z sensitive spectral peaks related to the target component, wherein z is more than or equal to 2 and z is less than or equal to m;
calculating the Mahalanobis distance of z sensitive spectrum peaks of the sample to be detected based on all samples of the sample set;
when the Mahalanobis distance does not exceed a preset value q, searching a sub-coverage range to which the first predicted value belongs and acquiring a second quantitative analysis model corresponding to the sub-coverage range, wherein q is greater than 0;
and when the Mahalanobis distance exceeds a preset value q, searching the spectrum matrix to which the sensitive spectrum peak with the lowest Mahalanobis distance belongs and acquiring a second quantitative analysis model corresponding to the spectrum matrix.
In a third aspect, there is provided a quantitative analysis model construction apparatus, including:
the first quantitative analysis model building module is used for training based on a preset linear regression method and a preset sample set to obtain a first quantitative analysis model; the sample set comprises the target component content of each sample and the spectrum matrix of all samples;
the sub-sample construction module is used for dividing the coverage range of the target component content into k continuous and non-overlapping sub-coverage ranges, and dividing the sample set into k sub-samples based on the k sub-coverage ranges and corresponding spectral matrixes, wherein k is more than or equal to 2;
and the second quantitative analysis model building module is used for respectively training and obtaining k second quantitative analysis models based on a preset nonlinear regression method and each subsample.
In a fourth aspect, there is provided a quantitative analysis device based on the quantitative analysis model construction device, the quantitative analysis device including:
the first prediction module is used for inputting the spectrum of the sample to be tested into the first quantitative analysis model to obtain a first predicted value of the content of the target component;
the matching module is used for determining a second quantitative analysis model from k second quantitative analysis models corresponding to the first quantitative analysis model according to the first predicted value;
and the second prediction module is used for inputting the spectrum of the sample to be detected into a corresponding second quantitative analysis model to obtain the content of the target component in the sample to be detected.
In a fifth aspect, there is provided a computer system comprising:
one or more processors; and
a memory associated with the one or more processors for storing program instructions that, when read and executed by the one or more processors, perform operations comprising:
training based on a preset linear regression method and a preset sample set to obtain a first quantitative analysis model; the sample set comprises the target component content of each sample and the spectrum matrix of all samples;
dividing the coverage range of the target component content into k continuous and non-overlapping sub-coverage ranges, and dividing the sample set into k sub-samples based on the k sub-coverage ranges and corresponding spectral matrixes, wherein k is more than or equal to 2;
and respectively training and obtaining k second quantitative analysis models based on a preset nonlinear regression method and each subsample.
Compared with the prior art, the invention has the beneficial effects that:
the invention provides a method for constructing a quantitative analysis model, which comprises the following steps: training based on a preset linear regression method and a preset sample set to obtain a first quantitative analysis model; the sample set comprises the target component content of each sample and the spectrum matrix of all samples; dividing a coverage range of the content of the target component into k continuous and non-overlapping sub-coverage ranges, and dividing a sample set into k sub-samples based on the k sub-coverage ranges and corresponding spectral matrixes, wherein k is more than or equal to 2; respectively training and obtaining k second quantitative analysis models based on a preset nonlinear regression method and each subsample; when a master-slave quantitative analysis model is constructed, a first quantitative analysis model is constructed on the basis of a linear regression method to cover all detection ranges, k second quantitative analysis models are constructed on the basis of a nonlinear regression method to cover all the detection ranges respectively, and the superposition of the detection ranges of all the second quantitative analysis models is equal to the detection range of the first quantitative analysis model, so that the coverage range of the quantitative analysis models is listed as the investigation content of the model construction, the conditions of global linearity but local nonlinearity of samples are fully considered, and the method has better prediction effect and accuracy on the construction of a near-infrared quantitative analysis model of feed raw materials from the grain and oil industry;
furthermore, the invention also provides a quantitative analysis method, which is realized based on the quantitative analysis model, and when quantitative analysis is carried out, the Mahalanobis distance of a sensitive peak is introduced to realize the incidence relation between the first quantitative analysis model and the second quantitative analysis model, and a prediction result is finally obtained, so that the detection of global linearity and local nonlinearity in the actual analysis process is realized, and the detection accuracy is further improved;
the embodiments of the present application only need to achieve any technical effect.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed to be used in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.
FIG. 1 is a flow chart of a method for constructing a quantitative analysis model according to an embodiment of the present invention;
FIG. 2 is a flow chart of a quantitative analysis method according to a second embodiment of the present invention;
FIG. 3 is a flow chart of a quantitative analysis method for detection according to a third embodiment of the present invention;
FIG. 4 is a near infrared spectrum of all samples in example three of the present invention;
FIG. 5 is a scatter plot of reference values and near-IR prediction values for a sample set in accordance with a third embodiment of the present invention;
FIG. 6 is a scatter plot of the reference values and the near-IR prediction values of the independent validation set samples in the third embodiment of the present invention;
FIG. 7 is a schematic structural diagram of a quantitative analysis model construction apparatus according to a fourth embodiment of the present invention;
FIG. 8 is a schematic structural view of a quantitative analysis device in accordance with a fifth embodiment of the present invention;
FIG. 9 is a block diagram of a computer system according to a sixth embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The embodiment provides a quantitative analysis model construction method and a quantitative analysis method aiming at the situation that the actual content distribution of a detected sample has global linearity and local nonlinearity in the spectrum quantitative analysis technology, which can effectively overcome the problems and provide a quantitative analysis method with higher accuracy.
The method for constructing a quantitative analysis model, the method for quantitative analysis, the apparatus and the system will be further described with reference to the specific embodiments.
Example one
Referring to fig. 1, the present embodiment provides a method for constructing a quantitative analysis model, where the method includes the following steps:
s1, training based on a preset linear regression method and a preset sample set to obtain a first quantitative analysis model; the sample set includes the target component content of each sample and the spectral matrix obtained from the spectral scan of all samples.
The spectrum-based quantitative analysis method provided in this embodiment is generally implemented based on a functional relationship between a spectrum matrix of a constructed sample and a reference value matrix. By constructing the model, the content of the target component in any sample to be detected can be obtained as long as the spectral spectrum of the sample is obtained. Therefore, before step S1, the construction method further includes: s100, obtaining a preset sample set, wherein the preset sample set comprises the following substeps:
s101, collecting n samples containing target components, and obtaining the content of the target components in each sample, wherein n is more than or equal to 3. The method comprises the following steps:
collecting n representative samples, and obtaining the content (i.e. reference value) of the target component of each sample (y)iI =1, 2, …, n, n ≧ 3). So the coverage detection range of the sample in the model construction is ymax-yminWherein y ismaxRepresents the maximum value of the reference values, yminRepresents the minimum of the reference values;
s102, acquiring a spectrum matrix X (n, m) of the sample based on the spectrums of all the samples, wherein m represents the number of spectrum wavelengths. The method comprises the following steps:
the spectra of all samples were scanned and the spectral matrix for all the samples was X (n, m).
In a preferred embodiment, each sample may be repeatedly filled several times to obtain a corresponding plurality of spectra and averaged to obtain the spectra of the sample to improve configuration construction accuracy.
S103, constructing the sample set based on the target component content in each sample and the spectrum matrix X (n, m).
In one embodiment, the sample set is divided into a calibration set and an independent verification set according to a preset proportion, the calibration set is used as a sample for constructing the first quantitative model, and the independent verification set is used for model verification.
Preferably, the scaling set and the independent verification set are divided by using an SPXY sample division method, and if the scaling set and the independent verification set are divided, the following sample sets are all the scaling sets. Of course, the present embodiment does not limit this.
After obtaining the sample set, executing step S1, wherein step S1 specifically includes:
s11, spectrum preprocessing including a combination of one or more of smoothing, derivative processing, multivariate scatter correction, and variable normalization. Such pre-treatment processes are conventional means of spectral processing and will not be described in detail here. By this preprocessing step, the accuracy of model construction can be further improved.
And S12, training based on a preset linear regression method and a preset sample set to obtain a first quantitative analysis model. The embodiment is not limited to a specific linear regression method, and may be a partial least squares linear regression method, and the like.
In the modeling process, the model is subjected to interactive verification, and the training is stopped when the cross-verification Root Mean Square Error (RMSECV) and the component number meet preset thresholds (RMSECV does not exceed 0.365, and the component number does not exceed 15) to obtain a first quantitative analysis model. Preferably, when the model is interactively verified, an optimal first quantitative analysis model is determined on the basis of the minimum RMSECV and the minimum component number of an interactive verification set.
In one embodiment, the spectral preprocessing method for the optimal first quantitative analysis model includes first derivative (17 points), multivariate scatter correction, where 17 points represent the size of the window in the derivative process. In this manner, RMSECV is 0.365 with an optimal composition score of 14.
S2, dividing the coverage range of the target component content into k continuous and non-overlapping sub-coverage ranges, and dividing the sample set into k sub-samples based on the k sub-coverage ranges and corresponding spectrum matrixes, wherein k is larger than or equal to 2.
Specifically, dividing the coverage of the target component content into k consecutive and non-overlapping sub-coverage, including:
s21, obtaining the number k of the second quantitative analysis models according to the coverage of the sample set and the cross validation root mean square error of the first quantitative analysis models, wherein k is larger than or equal to 2.
Specifically, the value k is the largest integer value that satisfies the following formula:
2≤k≤(ymax-ymin)/(8×RMSECV)
wherein, ymax-yminIs the coverage of the content of the target component, ymaxIs the maximum content of the target component, yminIs the minimum content of the target component; RMSECV represents the cross-validation root mean square error of the first quantitative analysis model.
And S22, equally dividing the coverage range of the sample set into k continuous k sub-coverage ranges which are not overlapped.
In the embodiment, when the sample set is divided, the sample set is divided into a plurality of continuous and non-overlapping sub-coverage ranges, so that the coverage ranges of the subsequently constructed second quantitative analysis models are different, and the sum of the coverage ranges of all the second quantitative analysis models is the same as the coverage range of the first quantitative analysis model.
Dividing the sample set into k subsamples based on the k sub-coverage ranges and the corresponding spectral matrices, including:
and S23, correspondingly obtaining corresponding sub-spectrum matrixes according to the content of the target component in each sub-coverage range.
S24, constructing corresponding k sub samples based on the k sub coverage ranges and the corresponding sub spectrum matrixes.
And S3, respectively training and obtaining k second quantitative analysis models based on a preset nonlinear regression method and each subsample.
In one embodiment, a support vector machine method nonlinear regression method is used for model training. Also preferably, the training process comprises a spectral pre-processing process.
Of course, on the premise of dividing the independent verification set, the independent verification set can be adopted to verify the quantitative analysis model.
When the method for constructing the quantitative analysis model is used for constructing the master-slave quantitative analysis model, the coverage range of the quantitative analysis model is listed in the investigation content of model construction, the condition that the sample is globally linear but locally nonlinear is fully considered, and the method has better prediction effect and accuracy.
Example two
Referring to fig. 2, the present embodiment provides a quantitative analysis method based on the quantitative analysis model constructed in the first embodiment, where the quantitative analysis method includes:
s10, inputting the spectrum of the sample to be tested into a first quantitative analysis model to obtain a first predicted value of the content of the target component;
s20, determining a second quantitative analysis model from the k second quantitative analysis models corresponding to the first quantitative analysis model according to the first predicted value;
step S20 specifically includes the following substeps:
s201, selecting z sensitive spectral peaks related to a target component, wherein z is more than or equal to 2 and z is less than or equal to m;
s202, calculating the Mahalanobis distances of z sensitive spectrum peaks of the sample to be detected based on all samples of the sample set;
when the Mahalanobis distance does not exceed a preset value q, searching a sub-coverage range to which the first predicted value belongs, acquiring a second quantitative analysis model corresponding to the sub-coverage range, and constructing an incidence relation between the first quantitative analysis model and the second quantitative analysis model, wherein q is greater than 0;
and when the Mahalanobis distance exceeds a preset value q, searching the spectrum matrix to which the sensitive spectrum peak with the lowest Mahalanobis distance belongs, acquiring a second quantitative analysis model corresponding to the spectrum matrix, and constructing an incidence relation between the first quantitative analysis model and the second quantitative analysis model.
And S30, inputting the spectrum of the sample to be detected into a corresponding second quantitative analysis model to obtain the content of the target component in the sample to be detected.
The quantitative analysis method is realized based on a quantitative analysis model, and when quantitative analysis is carried out, the Mahalanobis distance of a sensitive peak is introduced to realize the incidence relation between a first quantitative analysis model and a second quantitative analysis model, and a prediction result is finally obtained, so that the detection of global linearity and local nonlinearity in the actual analysis process is realized, and the detection accuracy is further improved.
EXAMPLE III
In order to further explain the quantitative analysis model construction method and the quantitative analysis method in the embodiment, the following quantitative analysis based on near infrared spectroscopy is taken as an example, and the scheme is further exemplified by combining a specific scene.
The crude protein content of the soybean meal exhibits a non-linear relationship with the near infrared spectrum over a specified content range (e.g., crude protein 43% specification, crude protein 46% specification, etc.), and a linear relationship over the entire content range. In order to improve the accuracy of the near-infrared quantitative analysis model of the material, the embodiment provides a quantitative analysis model construction method and a quantitative analysis method, wherein the quantitative analysis model construction method comprises the following steps:
s100', obtaining a preset sample set, specifically including:
s101', collecting 961 soybean meal samples, and detecting to obtain the crude protein content of all the samples by adopting a method for determining crude eggs in GB/T6432 feed, wherein the maximum value is 50.56%, the minimum value is 40.84%, and the reference value range is 9.72%.
S102', acquiring a near infrared spectrum of all soybean meal samples, and acquiring a spectrum matrix X (n, m) of all soybean meal samples based on the near infrared spectrum of all the samples. Specifically, each sample was repeatedly filled 2 times, each time 1 spectrum was obtained, and the sample spectrum was taken as the average spectrum of the 2 spectra, as shown in particular with reference to fig. 4. Near infrared spectral resolution of 16 cm-1The spectral range is 10000-4000 cm-1The total number of spectral wavelength variations is 750.
S103', constructing a sample set based on the target component content in each sample and the spectrum matrix X (n, m).
In this example, 961 samples were divided into a sample set and an independent validation set by the SPXY sample division method. And setting a sample set for model training and an independent verification set for model verification according to the ratio of 7:3, wherein the sample set totals 672 samples, and the independent verification set totals 289 samples.
After obtaining a sample set, constructing a quantitative analysis model, which specifically comprises the following steps:
s1', training based on a preset linear regression method and a preset sample set to obtain a first quantitative analysis model, which specifically comprises the following substeps:
s11', spectrum preprocessing, wherein the spectrum preprocessing is performed by adopting a first derivative (17 points) and multivariate scattering correction.
S12', a first quantitative analysis model is obtained based on the partial least squares linear regression method and sample set training.
Based on the principle that the RMSECV minimum and the component number minimum (maximum is not more than 15) of the interactive verification set, an optimal first quantitative analysis model M1 is determined, and the optimal spectrum preprocessing method of the model is first-order derivative (17 points) + multivariate scattering correction, wherein the 17 points represent the size of a window in derivative processing.
The first quantitative analysis model constructed in this example had an RMSECV of 0.365 and an optimal composition number of 14. The reference values and the near-ir prediction scattergrams of the sample sets in this example are shown in detail in fig. 5.
S2', dividing the coverage range of the target component content into k continuous and non-overlapping sub-coverage ranges, and dividing the sample set into k sub-samples based on the k sub-coverage ranges and corresponding spectrum matrixes, wherein k is larger than or equal to 2.
S21', obtaining the number k of the second quantitative analysis models according to the coverage range of the sample set and the cross validation root mean square error of the first quantitative analysis models, wherein the k value is the maximum integer value meeting the following formula:
2≤k≤(ymax-ymin)/(8×RMSECV),
i.e., 2. ltoreq. k.ltoreq.9.72%/(8X 0.365%), so k is 3.
S22', evenly divide the coverage of the sample set [40.84%, 50.56% ] into 3 consecutive and non-overlapping sub-coverage: [40.84%,44.08% ], [44.08%,47.32% ], [47.32,50.56% ].
S23', obtaining corresponding sub-spectrum matrixes according to the content of the target component in each sub-coverage range.
S24', constructing respective 3 subsamples based on the 3 sub-coverage ranges and the respective sub-spectral matrices.
S3', respectively training to obtain 3 second quantitative analysis models (e.g., "second quantitative analysis model 1", "second quantitative analysis model 2", and "second quantitative analysis model 3" in fig. 3) based on each subsample by using a support vector machine method nonlinear regression method in combination with a spectral preprocessing method. Similarly, an optimal second quantitative analysis model is determined based on the minimum RMSECV of the interactive verification set.
Calculating evaluation parameters of the second quantitative analysis model: absolute coefficient R2p is 0.95, RMSEP is 0.302, RPDpIs 4.78. The constructed quantitative analysis model is verified by adopting an independent verification set, and a scatter diagram of the reference value and the near-infrared predicted value of the sample of the independent verification set is shown in detail in figure 6. Therefore, the accuracy of the quantitative analysis model constructed by the embodiment is high.
And after the model construction is completed, performing crude protein content determination on the soybean meal sample to be determined based on the quantitative analysis model. As shown in fig. 3, the quantitative analysis method specifically includes the following steps:
s10', inputting the near infrared spectrum of the sample to be detected into a first quantitative analysis model to obtain a first predicted value;
s20', determining a second quantitative analysis model from the 3 second quantitative analysis models corresponding to the first quantitative analysis model based on the first predicted value. Specifically, step S20' includes:
s201', selecting 5 sensitive spectrum peaks (4878 cm) related to target components-1、4761 cm-1、4587 cm-1、5737 cm-1And 6756 cm-1);
S202', calculating the Mahalanobis distance of 5 sensitive spectrum peaks of the sample to be detected based on all samples of the sample set;
when the Mahalanobis distance does not exceed the preset value of 3.0, searching the sub-coverage range to which the first predicted value belongs, acquiring a second quantitative analysis model corresponding to the sub-coverage range, and constructing an incidence relation between the first quantitative analysis model and the second quantitative analysis model;
and when the Mahalanobis distance exceeds the preset value of 3.0, searching the spectrum matrix to which the sensitive spectrum peak with the lowest Mahalanobis distance belongs, acquiring a second quantitative analysis model corresponding to the spectrum matrix, and constructing the incidence relation between the first quantitative analysis model and the second quantitative analysis model.
S30', acquiring the near infrared spectrum of the sample to be detected and inputting the near infrared spectrum of the sample to be detected into the corresponding second quantitative analysis model to obtain the content of the target component in the sample to be detected.
It should be noted that, in this embodiment, the technical solution of the present invention is described by way of example only by using a quantitative analysis technique based on near infrared spectrum, and is not limited to the present invention.
Example four
In order to implement the method for constructing a quantitative analysis model in the first embodiment, this embodiment provides a corresponding apparatus for constructing a quantitative analysis model, as shown in fig. 7, the apparatus includes:
the first quantitative analysis model building module is used for training based on a preset linear regression method and a preset sample set to obtain a first quantitative analysis model; the sample set comprises the target component content of each sample and a spectrum matrix obtained by the spectrum scanning of all samples;
the sub-sample construction module is used for dividing the coverage range of the target component content into k continuous and non-overlapping sub-coverage ranges, and dividing the sample set into k sub-samples based on the k sub-coverage ranges and corresponding spectral matrixes, wherein k is more than or equal to 2;
and the second quantitative analysis model building module is used for respectively training and obtaining k second quantitative analysis models based on a preset nonlinear regression method and each subsample.
The sub-sample construction module comprises:
the first calculation unit is used for acquiring the number k of the second quantitative analysis models according to the coverage range of the sample set and the cross validation root-mean-square error of the first quantitative analysis model;
and the second computing unit is used for averagely dividing the coverage range of the sample set into k continuous k sub-coverage ranges which are not overlapped.
The k value is the largest integer value that satisfies the following formula:
2≤k≤(ymax-ymin)/(8×RMSECV)
wherein, ymax-yminIs the coverage of the content of the target component, ymaxIs the maximum content of the target component, yminIs the minimum content of the target component; RMSECV represents the cross-validation root mean square error of the first quantitative analysis model.
The sub-sample construction module further comprises:
the first acquisition unit is used for correspondingly acquiring corresponding sub-spectrum matrixes according to the content of the target component in each sub-coverage range;
a first constructing unit, configured to construct k corresponding sub-samples based on the k sub-coverage ranges and the corresponding sub-spectrum matrices.
The construction apparatus further includes: a preset sample set acquisition module comprising:
the acquisition unit is used for acquiring n samples containing target components to obtain the content of the target components in each sample, wherein n is more than or equal to 3;
a second acquisition unit for acquiring a spectral matrix X (n, m) of the sample based on the spectra of all samples, where m represents the number of spectral wavelengths;
a second construction unit for constructing the sample set based on the target component content in each sample and the spectral matrix X (n, m).
It should be noted that: the quantitative analysis model building device provided in the foregoing embodiment is only illustrated by dividing the functional modules when triggering a quantitative analysis model building service, and in practical applications, the function distribution may be completed by different functional modules according to needs, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. In addition, the embodiment of the quantitative analysis model construction device provided in the above embodiment and the embodiment of the quantitative analysis model construction method provided in the first embodiment belong to the same concept, that is, the device is based on the method, and the specific implementation process thereof is described in the method embodiment in detail, and is not described herein again.
EXAMPLE five
In order to implement the quantitative analysis method in the second embodiment, the present embodiment provides a quantitative analysis device corresponding to the second embodiment, and the quantitative analysis device is implemented based on the quantitative analysis model construction device in the fourth embodiment, as shown in fig. 8, the quantitative analysis device includes:
the first prediction module is used for inputting the spectrum of the sample to be tested into the first quantitative analysis model to obtain a first predicted value of the content of the target component;
the matching module is used for determining a second quantitative analysis model from k second quantitative analysis models corresponding to the first quantitative analysis model according to the first predicted value; the matching module comprises:
a selecting unit for selecting z sensitive spectral peaks related to the target component, wherein z is more than or equal to 2 and z is less than or equal to m;
the third calculation unit is used for calculating the Mahalanobis distances of the z sensitive spectrum peaks of the sample to be detected based on all samples in the sample set;
a matching unit for:
when the Mahalanobis distance does not exceed a preset value q, searching a sub-coverage range to which the first predicted value belongs and acquiring a second quantitative analysis model corresponding to the sub-coverage range;
and when the Mahalanobis distance exceeds a preset value q, searching the spectrum matrix to which the sensitive spectrum peak with the lowest Mahalanobis distance belongs and acquiring a second quantitative analysis model corresponding to the spectrum matrix.
And the second prediction module is used for inputting the spectrum of the sample to be detected into a corresponding second quantitative analysis model to obtain the content of the target component in the sample to be detected.
It should be noted that: in the quantitative analysis device provided in the above embodiment, when triggering the quantitative analysis service, only the division of the functional modules is illustrated, and in practical applications, the function distribution may be completed by different functional modules according to needs, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. In addition, the quantitative analysis device provided in the above embodiment and the embodiment of the quantitative analysis method provided in the second embodiment belong to the same concept, that is, the device is based on the method, and the specific implementation process thereof is described in the method embodiment, and will not be described herein again.
EXAMPLE six
Corresponding to the above method and apparatus, the present embodiment provides a computer system, including:
one or more processors; and
a memory associated with the one or more processors for storing program instructions that, when read and executed by the one or more processors, perform operations comprising:
training based on a preset linear regression method and a preset sample set to obtain a first quantitative analysis model; the sample set comprises the target component content of each sample and a spectrum matrix obtained by the spectrum scanning of all samples;
dividing the coverage range of the target component content into k continuous and non-overlapping sub-coverage ranges, and dividing the sample set into k sub-samples based on the k sub-coverage ranges and corresponding spectral matrixes, wherein k is more than or equal to 2;
and respectively training and obtaining k second quantitative analysis models based on a preset nonlinear regression method and each subsample.
Fig. 9 illustrates an architecture of a computer system 1500 that may include, in particular, a processor 1510, a video display adapter 1511, a disk drive 1512, an input/output interface 1513, a network interface 1514, and a memory 1520. The processor 1510, video display adapter 1511, disk drive 1512, input/output interface 1513, network interface 1514, and memory 1520 may be communicatively coupled via a communication bus 1530.
The processor 1510 may be implemented by using a general CXU (Central processing Unit), a microprocessor, an Application Specific Integrated Circuit (ASIC), or one or more Integrated circuits, and is configured to execute a relevant program to implement the technical solution provided by the present application.
The Memory 1520 may be implemented in the form of a ROM (Read Only Memory), a RAM (Random Access Memory), a static storage device, a dynamic storage device, or the like. The memory 1520 may store an operating system 1521 for controlling the operation of the computer system 1500, a Basic Input Output System (BIOS)1522 for controlling low-level operations of the computer system 1500. In addition, a web browser 1523, a data storage management system 1524, an icon font processing system 1525, and the like can also be stored. The icon font processing system 1525 may be an application program that implements the operations of the foregoing steps in this embodiment of the application. In summary, when the technical solution provided by the present application is implemented by software or firmware, the relevant program codes are stored in the memory 1520 and called for execution by the processor 1510.
The input/output interface 1513 is used for connecting an input/output device to input and output information. The input/output devices may be disposed as components within the device (not shown) or may be external to the device to provide corresponding functionality. The input devices may include a keyboard, a mouse, a touch screen, a microphone, various sensors, etc., and the output devices may include a display, a speaker, a vibrator, an indicator light, etc.
The network interface 1514 is used to connect network devices (not shown) for communicative interaction with the present device. The network device may implement communication in a wired manner (e.g., USB, network cable, etc.), or may implement communication in a wireless manner (e.g., mobile network, WIFI, bluetooth, etc.).
The bus 1530 includes a path to transfer information between the various components of the device, such as the processor 1510, the video display adapter 1511, the disk drive 1512, the input/output interface 1513, the network interface 1514, and the memory 1520.
In addition, the computer system 1500 may also obtain information of specific pickup conditions from a virtual resource object pickup condition information database for performing condition judgment, and the like.
It should be noted that although the above devices only show the processor 1510, the video display adapter 1511, the disk drive 1512, the input/output interface 1513, the network interface 1514, the memory 1520, the bus 1530, etc., in a specific implementation, the devices may also include other components necessary for proper operation. Furthermore, it will be understood by those skilled in the art that the apparatus described above may also include only the components necessary to implement the solution of the present application, and not necessarily all of the components shown in the figures.
From the above description of the embodiments, it is clear to those skilled in the art that the present application can be implemented by software plus necessary general hardware platform. Based on such understanding, the technical solutions of the present application may be embodied in the form of a software product, which may be stored in a storage medium, such as a ROM/RAM, a magnetic disk, an optical disk, or the like, and includes several instructions for enabling a computer device (which may be a personal computer, a cloud server, or a network device) to execute the method according to the embodiments or some parts of the embodiments of the present application.
The embodiments in the present specification are described in a progressive manner, and the same and similar parts among the embodiments are referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, the system or system embodiments are substantially similar to the method embodiments and therefore are described in a relatively simple manner, and reference may be made to some of the descriptions of the method embodiments for related points. The above-described system and system embodiments are only illustrative, wherein the units described as separate parts may or may not be physically separate, and the parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment. One of ordinary skill in the art can understand and implement the data without inventive effort.
While preferred embodiments of the present invention have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. Therefore, it is intended that the appended claims be interpreted as including preferred embodiments and all such alterations and modifications as fall within the scope of the embodiments of the invention.
It will be apparent to those skilled in the art that various changes and modifications may be made in the present invention without departing from the spirit and scope of the invention. Thus, if such modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include such modifications and variations.
Claims (9)
1. A method for constructing a quantitative analysis model is characterized by comprising the following steps:
training based on a preset linear regression method and a preset sample set to obtain a first quantitative analysis model; the sample set comprises the target component content of each sample and the spectrum matrix of all samples;
dividing the coverage range of the target component content into k continuous and non-overlapping sub-coverage ranges, and dividing the sample set into k sub-samples based on the k sub-coverage ranges and corresponding spectral matrixes, wherein k is more than or equal to 2;
respectively training and obtaining k second quantitative analysis models based on a preset nonlinear regression method and each subsample;
constructing an incidence relation between the first quantitative analysis model and the second quantitative analysis model;
wherein the constructing of the incidence relation between the first quantitative analysis model and the second quantitative analysis model comprises:
setting a preset value q of the Mahalanobis distance so that after quantitative analysis is carried out on a sample to be detected and a first predicted value is obtained through a first quantitative analysis model:
when the Mahalanobis distance of the sample to be detected does not exceed a preset value q, searching a sub-coverage range to which the first predicted value belongs, acquiring a second quantitative analysis model corresponding to the sub-coverage range, and constructing a correlation relation between the first quantitative analysis model and the second quantitative analysis model, wherein q is greater than 0;
when the Mahalanobis distance of the sample to be detected exceeds a preset value q, searching the spectrum matrix of the sensitive spectrum peak with the lowest Mahalanobis distance, acquiring a second quantitative analysis model corresponding to the spectrum matrix, and constructing a correlation relation between the first quantitative analysis model and the second quantitative analysis model;
the Mahalanobis distance of the sample to be detected is obtained by calculating the selected z sensitive spectral peaks of the sample to be detected and a preset sample set.
2. The method for constructing a quantitative analysis model according to claim 1, wherein the dividing the coverage of the target component content into k continuous and non-overlapping sub-coverage comprises:
obtaining the number k of the second quantitative analysis models according to the coverage range of the sample set and the cross validation root-mean-square error of the first quantitative analysis model;
the coverage of the sample set is equally divided into k consecutive and non-overlapping k sub-coverage.
3. The method of constructing a quantitative analysis model according to claim 2, wherein the k value is a maximum integer value satisfying the following formula:
2≤k≤(ymax-ymin)/(8×RMSECV)
wherein, ymax-yminIs the coverage of the content of the target component, ymaxIs the maximum content of the target component, yminIs the minimum content of the target component;
RMSECV represents the cross-validation root mean square error of the first quantitative analysis model.
4. The method of claim 1, wherein the dividing the sample set into k subsamples based on the k sub-coverage ranges and corresponding spectral matrices comprises:
correspondingly obtaining a corresponding sub-spectrum matrix according to the content of the target component in each sub-coverage range;
and constructing corresponding k sub-samples based on the k sub-coverage ranges and the corresponding sub-spectrum matrixes.
5. The quantitative analysis model building method according to claim 1, further comprising: obtaining a preset sample set, comprising the following substeps:
collecting n samples containing target components, and obtaining the content of the target components in each sample, wherein n is more than or equal to 3;
obtaining a spectral matrix X (n, m) of the sample based on the spectra of all samples, wherein m represents the number of spectral wavelengths;
constructing the sample set based on the target component content in each sample and a spectral matrix X (n, m).
6. A quantitative analysis method based on the quantitative analysis model of any one of claims 1 to 5, wherein the quantitative analysis method comprises:
inputting the spectrum of the sample to be tested into the first quantitative analysis model to obtain a first predicted value of the content of the target component;
determining a second quantitative analysis model from k second quantitative analysis models corresponding to the first quantitative analysis model according to the first predicted value;
inputting the spectrum of the sample to be detected into a corresponding second quantitative analysis model to obtain the content of the target component in the sample to be detected;
the determining a second quantitative analysis model from k second quantitative analysis models corresponding to the first quantitative analysis model according to the first predicted value comprises the following sub-steps:
selecting z sensitive spectral peaks related to the target component, wherein z is more than or equal to 2 and z is less than or equal to m;
calculating the Mahalanobis distance of z sensitive spectrum peaks of the sample to be detected based on all samples of the sample set;
when the Mahalanobis distance does not exceed a preset value q, searching a sub-coverage range to which the first predicted value belongs and acquiring a second quantitative analysis model corresponding to the sub-coverage range, wherein q is greater than 0;
and when the Mahalanobis distance exceeds a preset value q, searching the spectrum matrix to which the sensitive spectrum peak with the lowest Mahalanobis distance belongs and acquiring a second quantitative analysis model corresponding to the spectrum matrix.
7. A quantitative analysis model construction apparatus for performing the quantitative analysis model construction method according to any one of claims 1 to 5, characterized in that the construction apparatus comprises:
the first quantitative analysis model building module is used for training based on a preset linear regression method and a preset sample set to obtain a first quantitative analysis model; the sample set comprises the target component content of each sample and the spectrum matrix of all samples;
the sub-sample construction module is used for dividing the coverage range of the target component content into k continuous and non-overlapping sub-coverage ranges, and dividing the sample set into k sub-samples based on the k sub-coverage ranges and corresponding spectral matrixes, wherein k is more than or equal to 2;
the second quantitative analysis model building module is used for respectively training and obtaining k second quantitative analysis models based on a preset nonlinear regression method and each subsample;
the construction device is also used for constructing an incidence relation between the first quantitative analysis model and the second quantitative analysis model;
wherein the constructing of the incidence relation between the first quantitative analysis model and the second quantitative analysis model comprises:
setting a preset value q of the Mahalanobis distance so that after quantitative analysis is carried out on a sample to be detected and a first predicted value is obtained through a first quantitative analysis model:
when the Mahalanobis distance of the sample to be detected does not exceed a preset value q, searching a sub-coverage range to which the first predicted value belongs, acquiring a second quantitative analysis model corresponding to the sub-coverage range, and constructing a correlation relation between the first quantitative analysis model and the second quantitative analysis model, wherein q is greater than 0;
when the Mahalanobis distance of the sample to be detected exceeds a preset value q, searching the spectrum matrix of the sensitive spectrum peak with the lowest Mahalanobis distance, acquiring a second quantitative analysis model corresponding to the spectrum matrix, and constructing a correlation relation between the first quantitative analysis model and the second quantitative analysis model;
the Mahalanobis distance of the sample to be detected is obtained by calculating the selected z sensitive spectral peaks of the sample to be detected and a preset sample set.
8. A quantitative analysis apparatus for executing the quantitative analysis model building apparatus according to claim 7, characterized by comprising:
the first prediction module is used for inputting the spectrum of the sample to be tested into the first quantitative analysis model to obtain a first predicted value of the content of the target component;
the matching module is used for determining a second quantitative analysis model from k second quantitative analysis models corresponding to the first quantitative analysis model according to the first predicted value;
the second prediction module is used for inputting the spectrum of the sample to be detected into a corresponding second quantitative analysis model to obtain the content of the target component in the sample to be detected;
the matching module includes:
a selecting unit for selecting z sensitive spectral peaks related to the target component, wherein z is more than or equal to 2 and z is less than or equal to m;
the third calculation unit is used for calculating the Mahalanobis distances of the z sensitive spectrum peaks of the sample to be detected based on all samples in the sample set;
a matching unit for:
when the Mahalanobis distance does not exceed a preset value q, searching a sub-coverage range to which the first predicted value belongs and acquiring a second quantitative analysis model corresponding to the sub-coverage range;
and when the Mahalanobis distance exceeds a preset value q, searching the spectrum matrix to which the sensitive spectrum peak with the lowest Mahalanobis distance belongs and acquiring a second quantitative analysis model corresponding to the spectrum matrix.
9. A computer system, comprising:
one or more processors; and
a memory associated with the one or more processors for storing program instructions that, when read and executed by the one or more processors, perform operations comprising:
training based on a preset linear regression method and a preset sample set to obtain a first quantitative analysis model; the sample set comprises the target component content of each sample and the spectrum matrix of all samples;
dividing the coverage range of the target component content into k continuous and non-overlapping sub-coverage ranges, and dividing the sample set into k sub-samples based on the k sub-coverage ranges and corresponding spectral matrixes, wherein k is more than or equal to 2;
respectively training and obtaining k second quantitative analysis models based on a preset nonlinear regression method and each subsample;
constructing an incidence relation between the first quantitative analysis model and the second quantitative analysis model;
wherein the constructing of the incidence relation between the first quantitative analysis model and the second quantitative analysis model comprises:
setting a preset value q of the Mahalanobis distance so that after quantitative analysis is carried out on a sample to be detected and a first predicted value is obtained through a first quantitative analysis model:
when the Mahalanobis distance of the sample to be detected does not exceed a preset value q, searching a sub-coverage range to which the first predicted value belongs, acquiring a second quantitative analysis model corresponding to the sub-coverage range, and constructing a correlation relation between the first quantitative analysis model and the second quantitative analysis model, wherein q is greater than 0;
when the Mahalanobis distance of the sample to be detected exceeds a preset value q, searching the spectrum matrix of the sensitive spectrum peak with the lowest Mahalanobis distance, acquiring a second quantitative analysis model corresponding to the spectrum matrix, and constructing a correlation relation between the first quantitative analysis model and the second quantitative analysis model;
the Mahalanobis distance of the sample to be detected is obtained by calculating the selected z sensitive spectral peaks of the sample to be detected and a preset sample set.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110146449.9A CN112509643B (en) | 2021-02-03 | 2021-02-03 | Quantitative analysis model construction method, quantitative analysis method, device and system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110146449.9A CN112509643B (en) | 2021-02-03 | 2021-02-03 | Quantitative analysis model construction method, quantitative analysis method, device and system |
Publications (2)
Publication Number | Publication Date |
---|---|
CN112509643A CN112509643A (en) | 2021-03-16 |
CN112509643B true CN112509643B (en) | 2021-07-09 |
Family
ID=74952656
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110146449.9A Active CN112509643B (en) | 2021-02-03 | 2021-02-03 | Quantitative analysis model construction method, quantitative analysis method, device and system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112509643B (en) |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109060771A (en) * | 2018-07-26 | 2018-12-21 | 温州大学 | A kind of common recognition model building method based on spectrum different characteristic collection |
CN110514619A (en) * | 2019-08-29 | 2019-11-29 | 山东科技大学 | Near-Infrared Quantitative Analysis model building method based on Biased estimator |
Family Cites Families (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP2323081A1 (en) * | 2008-08-05 | 2011-05-18 | Fujitsu Limited | Method of creating estimation model by means of multiple regression analysis, creation system and creation program |
CN107991264A (en) * | 2017-11-17 | 2018-05-04 | 西南大学 | A kind of wheat flour protein matter and wet gluten content quick determination method |
CN109115692B (en) * | 2018-07-04 | 2021-06-25 | 北京格致同德科技有限公司 | Spectral data analysis method and device |
CN112285056B (en) * | 2020-10-14 | 2022-02-08 | 山东大学 | Method for selecting and modeling personalized correction set of spectrum sample |
-
2021
- 2021-02-03 CN CN202110146449.9A patent/CN112509643B/en active Active
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109060771A (en) * | 2018-07-26 | 2018-12-21 | 温州大学 | A kind of common recognition model building method based on spectrum different characteristic collection |
CN110514619A (en) * | 2019-08-29 | 2019-11-29 | 山东科技大学 | Near-Infrared Quantitative Analysis model building method based on Biased estimator |
Also Published As
Publication number | Publication date |
---|---|
CN112509643A (en) | 2021-03-16 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Padarian et al. | Using deep learning to predict soil properties from regional spectral data | |
Angelopoulou et al. | From laboratory to proximal sensing spectroscopy for soil organic carbon estimation—A review | |
Rossel | ParLeS: Software for chemometric analysis of spectroscopic data | |
Basile et al. | Use of artificial neural networks and NIR spectroscopy for non-destructive grape texture prediction | |
S. Veum et al. | Predicting profile soil properties with reflectance spectra via Bayesian covariate-assisted external parameter orthogonalization | |
Kalivas | Multivariate calibration, an overview | |
Pham et al. | Soil profile analysis using interactive visualizations, machine learning, and deep learning | |
JP7076463B2 (en) | Spectrum analyzer and spectrum analysis method | |
Zhao et al. | Calibration transfer based on affine invariance for NIR without transfer standards | |
Shen et al. | Rapid and real-time detection of moisture in black tea during withering using micro-near-infrared spectroscopy | |
Yu et al. | Prediction of soil properties based on characteristic wavelengths with optimal spectral resolution by using Vis-NIR spectroscopy | |
Jianqiang et al. | Characterization of deep green infection in tobacco leaves using a hand-held digital light projection based near-infrared spectrometer and an extreme learning machine algorithm | |
Wang et al. | A Lightweight convolutional neural network for nicotine prediction in tobacco by near-infrared spectroscopy | |
Reis et al. | Prediction of profiles in the process industries | |
Metz et al. | RoBoost-PLS2-R: an extension of RoBoost-PLSR method for multi-response | |
Tan et al. | Mutual information-induced interval selection combined with kernel partial least squares for near-infrared spectral calibration | |
CN112509643B (en) | Quantitative analysis model construction method, quantitative analysis method, device and system | |
Spiers et al. | Calibration model updating to novel sample and measurement conditions without reference values | |
Zhang et al. | Rapid analysis of apple leaf nitrogen using near infrared spectroscopy and multiple linear regression | |
Bowler et al. | Domain adaptation for in-line allergen classification of agri-food powders using near-infrared spectroscopy | |
Grassi et al. | Electric drive supervisor for milling process 4.0 automation: A process analytical approach with IIoT NIR devices for common wheat | |
CN110632024B (en) | Quantitative analysis method, device and equipment based on infrared spectrum and storage medium | |
CN113406057A (en) | Method and system for rapidly detecting pesticide residues in food | |
Xue et al. | The validity domain of sensor fusion in sensing soil quality indicators | |
He et al. | An active learning representative subset selection method using net analyte signal |
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 | ||
CB03 | Change of inventor or designer information |
Inventor after: Jiang Xunpeng Inventor after: Li Feng Inventor after: Chu Chunling Inventor after: Yang Yan Inventor before: Jiang Xunpeng Inventor before: Li Feng Inventor before: Ji Kun Inventor before: Chu Chunling Inventor before: Yang Yan Inventor before: Pu Junjie |
|
CB03 | Change of inventor or designer information | ||
GR01 | Patent grant | ||
GR01 | Patent grant |