CN114528872A - Evaluation method for spectrum standardization condition - Google Patents

Evaluation method for spectrum standardization condition Download PDF

Info

Publication number
CN114528872A
CN114528872A CN202210067342.XA CN202210067342A CN114528872A CN 114528872 A CN114528872 A CN 114528872A CN 202210067342 A CN202210067342 A CN 202210067342A CN 114528872 A CN114528872 A CN 114528872A
Authority
CN
China
Prior art keywords
spectrum
principal component
evaluation method
standardization
calculating
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.)
Pending
Application number
CN202210067342.XA
Other languages
Chinese (zh)
Inventor
田静
陈斌
梁振楠
陆道礼
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Jiangsu University
Original Assignee
Jiangsu University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Jiangsu University filed Critical Jiangsu University
Priority to CN202210067342.XA priority Critical patent/CN114528872A/en
Publication of CN114528872A publication Critical patent/CN114528872A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
    • G06F2218/12—Classification; Matching
    • 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
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
    • G06F18/2135—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on approximation criteria, e.g. principal component analysis

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • General Engineering & Computer Science (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Chemical & Material Sciences (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Signal Processing (AREA)
  • Spectroscopy & Molecular Physics (AREA)
  • Health & Medical Sciences (AREA)
  • Evolutionary Computation (AREA)
  • Analytical Chemistry (AREA)
  • Biochemistry (AREA)
  • General Health & Medical Sciences (AREA)
  • Immunology (AREA)
  • Pathology (AREA)
  • Investigating Or Analysing Materials By Optical Means (AREA)

Abstract

The invention provides an evaluation method aiming at a spectrum standardization condition, which comprises the following steps of 1, calculating a score matrix: from the host spectrum XmDecomposing the principal component load matrix PmAnd a principal component score matrix TmFurther, through the slave spectrum X'tAnd PmCalculating to obtain a master component score matrix T 'of the slave't(ii) a Step 2, calculating the score error rate of the principal component: through TmAnd T'tCalculating a principal component score error rate PCSER; finally, judging whether the spectral standardization is good or not according to the size of the PCSER value, wherein the spectral standardization is better when the value of the PCSER is smaller. The spectral standardization evaluation method can evaluate the difference between the spectrums according to the correction model established by the partial least square method and improve the similarity of the spectrums, thereby realizing the mode among different instrumentsType sharing; in addition, compared with the traditional evaluation means, the evaluation method of the invention does not need to completely execute a whole set of model prediction work, thereby saving a large amount of time cost.

Description

Evaluation method for spectrum standardization condition
Technical Field
The invention belongs to the technical field of spectral analysis, and particularly relates to an evaluation method for spectral standardization.
Background
The near infrared spectrum analysis technology has the advantages of rapidness, accuracy and greenness, and is widely applied to the fields of food, medicines and the like. However, in the using process of the near infrared spectrum analysis model, when the detection conditions, the detection environment or the instrument and equipment change, the absorbance of the near infrared spectrum will be different, so that the established analysis model is invalid, and a large amount of manpower and material resources are consumed for reestablishing the model.
In view of the above problems, it is currently common to employ a spectral normalization method to cope therewith. The spectrum standardization is to correct the spectrum data of the same sample collected by different instruments or different conditions, so that the difference caused by different external factors such as instruments, environments and the like in the measurement process is eliminated, the spectra collected under different conditions can be suitable for the same model, and the model sharing is realized.
However, a better evaluation method is lacked for the good and bad performance of the spectral standardization work at present. In current applications, only the standardized spectrum is actually input into the spectrum analysis model, and then the prediction result (equivalent to the measured value) is output through the model, and the prediction result is compared with the real result (namely the real value), and the spectrum standardization is indirectly deduced in a reverse manner according to the error between the measured value and the real value. The evaluation method cannot evaluate in advance, and the quality can be known only by completely executing all the work once, so that the time needed is long; besides standardization, many factors influence the final prediction error, that is, the size of the prediction error cannot necessarily indicate the quality of standardization.
Disclosure of Invention
Aiming at the defects in the prior art, the invention provides an evaluation method aiming at the spectrum standardization condition, which is used for solving the evaluation problem of the spectrum standardization.
The present invention achieves the above technical objects by the following technical means.
An evaluation method for spectrum standardization, comprising the following steps:
step 1, calculating a score matrix: from the host spectrum XmDecomposing the principal component load matrix PmAnd a principal component score matrix TmFrom the machine spectrum X'tAnd said PmSubstituting formula X ═ TPT+EXIn which EXFor the residual matrix, the slave is calculatedOf the principal component score matrix T't；
Step 2, calculating the score error rate of the principal component: will be the TmAnd T'tCalculating the principal component score error rate PCSER by substituting the following formula:
Figure BDA0003480708740000011
wherein T ism,iIs TmFraction of the ith main component, T't,iIs T'tFraction of the ith principal component, WiThe contribution rate of the ith principal component in the spectral analysis model, and n is the number of the principal components.
Further, the spectrum standardization is judged to be good or bad according to the value of the PCSER value, and the spectrum standardization is better when the value of the PCSER is smaller.
Further, the host spectrum X is analyzed based on principal component analysis methodmDecomposition is carried out.
Further, multiple sets of data are tested and the mean value PCSERave or the maximum value PCSERmax is taken among multiple PCSER results.
Further, the spectral band range is the near infrared light band.
Further, the spectral analysis model is established by a partial least squares method.
Further, the spectra were preprocessed by normalization and multivariate scatter correction.
The invention has the beneficial effects that:
the invention provides an evaluation method aiming at the spectrum standardization condition, which can evaluate the difference among spectra according to a correction model established by a partial least square method and improve the similarity of the spectra, thereby realizing the model sharing among different instruments. In addition, compared with the traditional evaluation means, the evaluation method of the invention does not need to completely execute a whole set of model prediction work, thereby saving a large amount of time cost.
Detailed Description
Reference will now be made in detail to the embodiments of the present invention, examples of which are illustrated, wherein like reference numerals refer to the same or similar elements or elements having the same or similar functions throughout. The embodiments described below by reference are exemplary and are intended to be illustrative of the invention, but are not to be construed as limiting the invention.
Description of the basic concept
The system is provided with two spectrometers which are a host and a slave, wherein a spectral analysis model is established by taking spectral data of the host as a reference. In order to apply the spectral analysis model (established based on the host) directly to the spectral data of the slave, the spectral data of the slave needs to be standardized.
It is to be noted that: the above-described concepts of master and slave are not narrowly construed as two distinct machines; if the measured spectrum data of the same machine is changed in two different use environments, the concepts of the master machine and the slave machine can be applied, namely the machine in one use environment is considered as the master machine, and the machine in the other use environment is considered as the slave machine.
For convenience of description, let subscripts m and t denote the master and slave, respectively, and let the master spectrum be X accordinglymFrom the original spectrum of the machine as XtX 'is the spectrum normalized from the machine'tIn particular Xm、XtAnd X'tThe data form of (a) is a matrix and thus may also be referred to as a spectral matrix. Obviously, the most desirable normalized result is X't＝Xm. In addition, how to perform the spectrum standardization, namely, the standardization method of the spectrum, is not the content of the invention; any known spectral normalization method can be used by those skilled in the art to perform the corresponding spectral normalization operation. The invention provides an evaluation and judgment method only for the quality of a spectrum standardization result.
In the field of spectral analysis, the spectrum of the near infrared light band is most often selected to detect the components of the analyte, so the following embodiments also illustrate the technical solution of the present invention based on the near infrared spectrum.
Second, evaluation method
Step 1, calculating score matrix
Decomposing the spectrum matrix based on principal component analysis method to obtain host spectrum XmPrincipal component load matrix P in (1)mAnd a principal component score matrix Tm。
The principal component load matrix P, the principal component score matrix T and the spectrum matrix X have the following operational relationship:
X＝TPT+EX (1)
wherein EXIs a residual matrix, also obtained when decomposing the principal component, PTIs the transpose of P.
Then from the normalized spectrum X'tAnd principal component load matrix P of the hostmSubstituting the formula (1) into the formula (1) to calculate and obtain a master component score matrix T 'of the slave't。
Step 2, calculating the score error rate of the principal component
Scoring principal components of the host by a matrix TmAnd a master component score matrix T 'of the slave'tSubstituting into the following equation:
Figure BDA0003480708740000031
wherein PCSER is the principal component score error rate defined in the invention; t ism,iPrincipal component scoring matrix T for a hostmFraction of the ith principal component, similarly, T't,iIs T'tThe score of the ith main component; wiThe contribution rate of the ith principal component in the spectral analysis model; n is the number of principal components.
Finally, the quality of the spectral standardization is judged according to the value of the PCSER value, and the smaller the value of the PCSER is, the better the spectral standardization is; in order to ensure the reliability of the result, a plurality of samples can be collected, a plurality of groups of data are correspondingly tested, and the average value PCSEVAve or the maximum value PCSEMax is taken from a plurality of PCSER results for judgment.
Third, effect test
1) Collecting the sample
Collecting 154 parts of wheat flour of different varieties or brands, including high gluten wheat flour, medium gluten wheat flour, low gluten wheat flour, self-raising flour and whole wheat flour. The test takes the content of crude protein in wheat flour as a detection object.
2) Near infrared spectral collection
Two spectrometers with different models are respectively selected as a host and a slave, wherein the host is a raster scanning type S450 near infrared spectrometer produced by Shanghai prism optical technology Limited, the waveband range is 900-2500 nm, the wavelength interval is 2nm, the number of wavelength points is 801, the wavelength points are consistent with the waveband of the slave, and the waveband range is cut into 1550-1950 nm during actual use; the Fabry-Perot interferometer type N500 near infrared spectrometer is produced by Jinan sea energy instruments, Inc., and has a wave band range of 1550-1950 nm, a wavelength interval of 2nm and 201 wavelength points.
154 parts of wheat flour was subjected to near infrared spectrum collection using the master and slave described above, respectively.
3) Determination of crude protein content in wheat flour
The crude protein content of 154 parts of wheat flour samples is detected according to the method specified in GB/T31578-.
4) Establishing a spectral analysis model
Establishing a spectral analysis model by a partial least square method based on spectral data acquired by a host and crude protein content data measured by corresponding wheat flour; in order to eliminate irrelevant information and noise interference in spectral data, normalization and multivariate scattering correction are used for preprocessing the spectrum.
5) Spectral normalization
The method comprises the steps of standardizing the spectrums of the slave machines by using three Standardization methods, namely Direct Standardization (DS), Piecewise Direct Standardization (PDS) and Single Linear Regression Direct Standardization (SLRDS), and accordingly obtaining the spectrums after three groups of slave machines are standardized.
6) Calculating a principal component score error rate
According to the above steps 1 to 2, the principal component score error rates PCSER corresponding to the three normalization processing results are calculated respectively, and for increasing the contrast, the principal component score error rates PCSER are calculated according to the steps 1 to 2 from the raw spectra of the slave machines without normalization processing. The corresponding results are shown in the following table:
table 1: principal component score error rate
Figure BDA0003480708740000041
7) Model prediction
The established spectrum analysis models are respectively applied to the slave spectra before standardization and the slave spectra after standardization in three modes, the crude protein content of the wheat flour is predicted by using the models, the prediction errors are compared with actual measured values, and the final result is shown in table 2, wherein Rp is a prediction correlation coefficient, RMSEP is a prediction standard deviation, RPD is a relative prediction deviation, and the higher the value of Rp is close to 1, the lower the value of RMSEP is, the higher the value of RPD is, the higher the corresponding model prediction precision is.
Table 2: model predicted effect
Figure BDA0003480708740000051
The data in tables 1 and 2 are comprehensively compared, and the DS algorithm has the best standardization effect in the test no matter the standard evaluation method is adopted or the actual model prediction result is tested, and then the PDS algorithm standardization and the SLRDS algorithm standardization are sequentially carried out. Therefore, the standardized evaluation method can effectively and accurately judge the standard quality of the spectrum.
Based on the same inventive concept as the above-described method of spectral normalized assessment, the present invention also provides an electronic device comprising one or more processors and one or more memories having computer readable code stored therein, wherein the computer readable code, when executed by the one or more processors, performs the calculation of the respective principle component score error rate, PCSER. Wherein, the memory may include a nonvolatile storage medium and an internal memory; the non-volatile storage medium may store an operating system and computer readable code. The computer readable code includes program instructions that, when executed, cause the processor to perform a test question similarity calculation. The processor is used for providing calculation and control capability and supporting the operation of the whole electronic equipment. The memory provides an environment for the execution of computer readable code in the non-volatile storage medium, which when executed by the processor, causes the processor to perform the test question similarity calculation method based on the solution idea and the knowledge points of the present invention. It should be understood that the Processor may be a Central Processing Unit (CPU), and the Processor may be other general purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field Programmable Gate Arrays (FPGAs) or other Programmable logic devices, discrete Gate or transistor logic devices, discrete hardware components, etc. Wherein a general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.
In the description of the present invention, it is to be understood that the terms "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc., indicate orientations or positional relationships based on those shown, and are used merely for convenience of description and simplicity of description, but do not indicate or imply that the device or element being referred to must have a particular orientation, be constructed and operated in a particular orientation, and thus, should not be construed as limiting the present invention.
The present invention is not limited to the above-described embodiments, and any obvious modifications, substitutions or alterations can be made by those skilled in the art without departing from the spirit of the present invention.

Claims (7)

1. An evaluation method for spectrum standardization, which is characterized by comprising the following steps:
step 1, calculating a score matrix: from the host spectrum XmDecomposing the principal component load matrix PmAnd a principal component score matrix TmFrom the machine spectrum X'tAnd said PmSubstituting formula X ═ TPT+EXIn which EXCalculating to obtain a master component score matrix T 'of the slave machine as a residual matrix't；
Step 2, calculating the score error rate of the principal component: will be the TmAnd T'tCalculating the principal component score error rate PCSER by substituting the following formula:
Figure FDA0003480708730000011
wherein T ism，iIs TmFraction of the ith main component, T't，iIs T'tFraction of the ith principal component, WiThe contribution rate of the ith principal component in the spectral analysis model, and n is the number of the principal components.
2. The evaluation method for spectrum normalization according to claim 1, wherein: the spectrum standardization is judged to be good or bad according to the size of the PCSER value, and the spectrum standardization is better when the value of the PCSER is smaller.
3. The evaluation method for spectrum normalization according to claim 1, wherein: main machine spectrum X based on principal component analysis methodmDecomposition is carried out.
4. The evaluation method for spectral normalization according to claim 2, characterized in that: multiple sets of data are tested and the mean PCSERave or maximum PCSERmax value is taken among multiple PCSER results.
5. The evaluation method for spectrum normalization according to claim 1, wherein: the spectral band range is the near infrared light band.
6. The evaluation method for spectrum normalization according to claim 5, wherein: the spectral analysis model is established by a partial least squares method.
7. The evaluation method for spectrum normalization according to claim 6, wherein: spectra were preprocessed by normalization and multivariate scatter correction.
CN202210067342.XA 2022-01-20 2022-01-20 Evaluation method for spectrum standardization condition Pending CN114528872A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210067342.XA CN114528872A (en) 2022-01-20 2022-01-20 Evaluation method for spectrum standardization condition

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210067342.XA CN114528872A (en) 2022-01-20 2022-01-20 Evaluation method for spectrum standardization condition

Publications (1)

Publication Number Publication Date
CN114528872A true CN114528872A (en) 2022-05-24

Family

ID=81621154

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210067342.XA Pending CN114528872A (en) 2022-01-20 2022-01-20 Evaluation method for spectrum standardization condition

Country Status (1)

Country Link
CN (1) CN114528872A (en)

Similar Documents

Publication Publication Date Title
CN108362662B (en) Near infrared spectrum similarity calculation method and device and substance qualitative analysis system
Shao et al. Visible/near-infrared spectra for linear and nonlinear calibrations: a case to predict soluble solids contents and pH value in peach
CN102564993B (en) Method for identifying rice varieties by using Fourier transform infrared spectrum and application of method
CN104897607A (en) Food modeling and rapid detecting integration method and system adopting portable NIRS (near infrared spectroscopy)
CN110687072B (en) Calibration set and verification set selection and modeling method based on spectral similarity
CN106706553A (en) Method for quick and non-destructive determination of content of amylase in corn single grains
CN102590129B (en) Method for detecting content of amino acid in peanuts by near infrared method
Li et al. Wavelength selection method for near-infrared spectroscopy based on standard-sample calibration transfer of mango and apple
CN108613943B (en) Near-infrared single-grain crop component detection method based on spectrum morphology transfer
CN109211829A (en) A method of moisture content in the near infrared spectroscopy measurement rice based on SiPLS
Duan et al. Sensitive variables extraction, non-destructive detection and visualization of total viable count (TVC) and pH in vacuum packaged lamb using hyperspectral imaging
WO2020248961A1 (en) Method for selecting spectral wavenumber without reference value
CN110376154A (en) Fruit online test method and system based on spectrum correction
CN108169168A (en) Test and analyze rice grain protein content mathematical model and construction method and application
CN106706554A (en) Method for rapidly and nondestructively determining content of straight-chain starch of corn single-ear grains
CN102519903B (en) Method for measuring whiteness value of Agaricus bisporus by using near infrared spectrum
CN114528872A (en) Evaluation method for spectrum standardization condition
CN104181125A (en) Method for rapidly determining Kol-bach value of beer malt
CN115436315A (en) Near infrared spectrum-based cement additive detection method
CN114324233A (en) Near-infrared nondestructive online quality detection method and system for nutritional ingredients of agricultural products
Liu et al. Feasibility of nondestructive detection of apple crispness based on spectroscopy and machine vision
CN113984683A (en) Hyperspectrum-based method for measuring starch content of potato whole flour noodles
Yu et al. A weighted ensemble method based on wavelength selection for near-infrared spectroscopic calibration
Zhao et al. Study on Near-Infrared Spectroscopy Non-Destructive Testing of Strawberry Quality
CN112763448A (en) ATR-FTIR technology-based method for rapidly detecting content of polysaccharides in rice bran

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