CN111487211A - Incoherent broadband cavity enhanced absorption spectrum fitting waveband selection method - Google Patents
Incoherent broadband cavity enhanced absorption spectrum fitting waveband selection method Download PDFInfo
- Publication number
- CN111487211A CN111487211A CN202010394233.XA CN202010394233A CN111487211A CN 111487211 A CN111487211 A CN 111487211A CN 202010394233 A CN202010394233 A CN 202010394233A CN 111487211 A CN111487211 A CN 111487211A
- Authority
- CN
- China
- Prior art keywords
- fitting
- band
- sub
- uncertainty
- fit
- 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.)
- Granted
Links
- 238000000862 absorption spectrum Methods 0.000 title claims abstract description 21
- 238000010187 selection method Methods 0.000 title claims abstract description 10
- 238000000034 method Methods 0.000 claims abstract description 22
- 238000001228 spectrum Methods 0.000 claims abstract description 20
- 238000013528 artificial neural network Methods 0.000 claims abstract description 18
- 230000003595 spectral effect Effects 0.000 claims abstract description 9
- 238000010801 machine learning Methods 0.000 claims abstract description 6
- 238000010521 absorption reaction Methods 0.000 claims description 16
- 238000010606 normalization Methods 0.000 claims description 14
- 238000012360 testing method Methods 0.000 claims description 6
- 238000005259 measurement Methods 0.000 claims description 5
- 238000012545 processing Methods 0.000 claims description 2
- 238000012549 training Methods 0.000 claims description 2
- 238000002310 reflectometry Methods 0.000 description 7
- 230000003287 optical effect Effects 0.000 description 5
- 230000005855 radiation Effects 0.000 description 4
- 238000001514 detection method Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 230000035945 sensitivity Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/17—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
- G01N21/3103—Atomic absorption analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Physics & Mathematics (AREA)
- Molecular Biology (AREA)
- Data Mining & Analysis (AREA)
- Spectroscopy & Molecular Physics (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Computing Systems (AREA)
- Chemical & Material Sciences (AREA)
- Biochemistry (AREA)
- Analytical Chemistry (AREA)
- Pathology (AREA)
- Immunology (AREA)
- Investigating Or Analysing Materials By Optical Means (AREA)
Abstract
The invention discloses a fitting waveband selection method for an incoherent broadband cavity enhanced absorption spectrum, which is characterized by determining a nonlinear relation between two parameters, namely the width and the central wavelength of a fitting waveband, and a relative fitting uncertainty and a fitting residual standard deviation by using a BP neural network machine learning method, and then determining the fitting waveband according to the expected relative fitting uncertainty and the fitting residual standard deviation, and specifically comprises the following steps of: determining a broad spectral band; dividing the broad spectral band into a plurality of sub-bands having different widths and center wavelengths; fitting the sub-band absorption spectrum; normalizing the fitting result to form a data sample; establishing a BP neural network and performing network learning; and obtaining the width and the central wavelength of the wave band to be selected by utilizing the learned BP neural network, and finally obtaining the spectrum fitting wave band. Compared with the existing method, the method can obtain the optimal spectrum fitting wave band, and avoids the possibility of larger deviation of the fitting result of the existing method.
Description
Technical Field
The invention relates to a fitting waveband selection method of a broadband absorption spectrum, in particular to a fitting waveband selection method of an incoherent broadband cavity enhanced absorption spectrum.
Background
The incoherent wideband cavity reinforced absorption spectrum technology is one high sensitivity optical detection method, and has optical resonant cavity comprising high reflectivity lens to increase the absorption optical path and raise the detection sensitivity of the detected gas. The technology selects an absorption spectrum with a certain wave band, and uses a least square method to fit a gas absorption cross section to obtain a measured absorption coefficient so as to obtain the molecular number concentration of the measured gas. In selecting the spectrum fitting band, the band in which the reflectivity of the lens is located, the band in which the radiation spectrum of the light source is located, and the band in which the strong absorption of the gas exists are generally considered. If there are overlapping regions in these three bands, the overlapping regions are typically used as spectral fit bands. At present, the incoherent broadband cavity enhanced absorption spectrum technology mainly adopts a light emitting diode as a light source, generally, the full width at half maximum of the light emitting diode is only 20-30 nm, the reflectivity of a lens of an optical resonant cavity is a function of wavelength, a large difference between the peak wavelength of the radiation spectrum of the light emitting diode light source and the peak wavelength of the reflectivity of the lens often occurs, and the full width at half maximum of the light emitting diode is very narrow, so that the overlapping degree of the wave bands of the light emitting diode and the lens is not high. For this situation, the existing spectrum fitting band selection method is a trial-and-error method, i.e. empirically selecting several fitting bands considered to be better, performing fitting comparison, and then using the band with the smallest uncertainty or fitting residual amplitude of the fitting result as the final spectrum fitting band. Because the number of the tried wave bands is limited, when the spectrum fitting wave band selected by the existing method is used for fitting the absorption spectrum, the uncertainty of the obtained fitting result or the fitting residual amplitude value does not necessarily reach the expected value of the measurement system, and especially when the overlapping area of the wave bands where the reflectivity of the lens, the radiation spectrum of the light source and the strong absorption of the gas exist is small, the fitting result of the spectrum fitting wave band selected by the existing method for the concentration of the gas to be measured may generate large deviation.
Disclosure of Invention
The invention aims to solve the problem that the fitting result of the existing spectrum fitting waveband selection method for the incoherent broadband cavity enhanced absorption spectrum technology to the concentration of a gas to be measured may generate large deviation, and provides a spectrum fitting waveband selection method based on BP neural network machine learning.
In order to achieve the purpose, the invention adopts the following technical scheme:
the invention discloses a fitting waveband selection method of an incoherent broadband cavity enhanced absorption spectrum, which is characterized in that a BP neural network machine learning method is utilized to determine a nonlinear relation between two parameters of the width and the center wavelength of a fitting waveband and a relative fitting uncertainty and a fitting residual standard deviation, and then the fitting waveband is determined according to an expected relative fitting uncertainty and the fitting residual standard deviation, and specifically comprises the following steps:
the first step is as follows: determining a wide spectral band according to an incoherent broadband cavity enhanced absorption spectrum measurement system; assuming that the wave band is [ a, b ], wherein a is the minimum wavelength, b is the maximum wavelength, and the difference between b and a is not less than 30nm and is an even number;
the second step is that: in the wave bands [ a, b]Internally divided into a plurality of sub-bandsWherein, BWjIs the width of the sub-band or sub-bands,λijis the center wavelength of the sub-band,
the third step: selecting waveletsSegment ofAs a spectrum fitting wave band, fitting the absorption cross section of the gas to be measured to the measured absorption coefficient by adopting a least square method to obtain a fitting concentration value N corresponding to all sub-wave bandsijUncertainty of fit EijAnd standard deviation of fit residual Dij(ii) a Calculating relative fit uncertainty REij,
The fourth step: sample data, i.e. sub-band width BW, by using maximum-minimum normalization methodjCentral wavelength of sub-band lambdaijRelative fit uncertainty REijAnd fitted residual standard deviation DijNormalization processing is carried out, and the normalized results are respectivelyAnd
the fifth step: establishing BP neural network, the number of nodes of network input layer is 2, and the input quantity is respectivelyAndthe number of nodes of the network output layer is 2, and the output quantity isThe number of nodes of the network hidden layer can be adjusted according to the training result; dividing a data sample corresponding to input-output quantity into a learning set and a testing set, and learning and testing the BP neural network;
and a sixth step: uncertainty RE of expected relative fit using max-min normalizationEAnd standard deviation of fit residual DEGo on to returnIs subjected to a normalization treatment to obtainAndobtaining the expected sub-band width with normalized output quantity of the network as the input quantity of the learned BP neural networkSum sub-band center wavelength
The seventh step: for normalized expected sub-band widthSum sub-band center wavelengthPerforming inverse normalization to obtain the desired sub-band width BWESum sub-band center wavelength λE(ii) a The finally determined spectral fit band is
The invention has the advantages and beneficial effects that:
(1) according to the method, a plurality of sub-wave bands with different widths and central wavelengths are selected in a specific wave band to serve as the spectrum fitting wave band, so that a large number of data samples representing the relevance between the fitting wave band and the fitting result are obtained, the relation between the data samples is mined, and the relevance between the data samples is more scientific than that obtained by the conventional method through experience.
(2) According to the method, after the nonlinear relation between the fitting wave band and the fitting result is obtained through a machine learning method, the optimal spectrum fitting wave band can be obtained according to the uncertainty of the fitting result and the specific requirements of the fitting residual error, and the possibility that the fitting result of the existing method generates larger deviation is avoided.
Drawings
FIG. 1 is a flow chart of the operation of the present invention.
Detailed Description
The method determines the nonlinear relation between two parameters of the width and the center wavelength of the fitting wave band and the relative fitting uncertainty and the standard deviation of the fitting residual by using a BP neural network machine learning method, and then determines the fitting wave band according to the expected relative fitting uncertainty and the standard deviation of the fitting residual.
As shown in FIG. 1, the method for selecting the fitting band of the incoherent broadband cavity enhanced absorption spectrum of the present invention has seven steps, namely step 1 to step 7. In step 1, a wide spectrum wave band [ a, b ], [ a, b ] is determined according to a wave band where the reflectivity of a lens of the incoherent broadband cavity enhanced absorption spectrum measurement system is located, a wave band where a light source radiation spectrum is located and a wave band where strong absorption exists in gas, the three wave bands are covered as far as possible, and the difference value between b and a is not less than 30nm and is an even number.
In step 2, the spectral bands [ a, b ]]The method comprises the following steps of dividing the optical fiber into sub-bands with different widths and central wavelengths: the sub-band width is selected first, starting at 10nm and one step every 2nm until b-a. Thus, the width BW of a sub-bandjCan be expressed as:then, in [ a, b ]]All possible center wavelengths are selected for each width sub-band in 1nm steps between. For example, if a is 430 and b is 460, the sub-band widths may range from 10,12,14, …, and 30 nm. For a sub-band of 10nm width, all possible center wavelengths are 435,436,437, …, 455nm, i.e. all sub-bands divided into 10nm width are: [430,440],[431,441],[432,442],…,[450,460]And (5) nm. Sub-bands of other widths, and so on. Thus, the center wavelength λ of the sub-bandijCan be expressed as:
in step 3, the absorption cross section of the measured gas can be obtained by performing convolution operation on the instrument function of the spectrometer adopted by the incoherent broadband cavity enhanced absorption spectrum measurement system and the high-resolution absorption cross section of the measured gas given in the literature. The absorption coefficient is enhanced by the incoherent broadband cavity and the absorption spectrum measuring system according to the formula(Note: α (. lamda.) is the absorption coefficient of the measured gas; I (. lamda.) and I0(λ) is the gas absorption spectrum and the reference spectrum, respectively; r (lambda) is the lens reflectivity; d is the length of the optical cavity). Taking all the sub-bands obtained by division in the step 2 as spectrum fitting bands, fitting the absorption cross section of the gas to be measured to the absorption coefficient by adopting a least square fitting method, and obtaining a fitting concentration value NijUncertainty of fit EijAnd standard deviation of fit residual Dij(ii) a According to the formulaCalculating to obtain relative fitting uncertainty REij。
In step 4, BW is normalized by max-min normalizationj、λij、REijAnd DijNormalization is carried out to obtain Andthe method specifically comprises the following steps: wherein min represents the minimum value of the data set, and max represents the maximum value of the data set.
In step 5, a BP neural network containing an input layer, a hidden layer and an output layer is built by using MAT L AB, and the BP neural network obtained in step 4 is usedAndas an input to the network, a network interface is provided,as a network output. And (4) dividing all data samples corresponding to the network input-output quantity obtained in the step (3) and the step (4) into a learning set and a testing set according to a certain proportion, and learning and testing the BP neural network.
In step 6, the expected relative fit uncertainty RE is normalized using a maximum-minimum methodEAnd standard deviation of fit residual DENormalization is carried out to obtainAndthe method specifically comprises the following steps:will be provided withAndinputting the data into the learned BP neural network obtained in the step 5, and outputting the data through the BP neural network to obtainAnd
in step 7, the normalized expected sub-band width obtained in step 6 is usedSum sub-band center wavelengthPerforming inverse normalization to obtain the desired sub-band width BWESum sub-band center wavelength λEThe method specifically comprises the following steps: thus, the final determined spectral fit band isAnd ending the whole incoherent broadband cavity enhanced absorption spectrum fitting waveband selection process.
Claims (1)
1. The fitting wave band selection method of the incoherent broadband cavity enhanced absorption spectrum is characterized by comprising the following steps: determining a nonlinear relation between two parameters of the width and the center wavelength of a fitting waveband and a relative fitting uncertainty and a fitting residual standard deviation by using a BP neural network machine learning method, and then determining the fitting waveband according to the expected relative fitting uncertainty and the fitting residual standard deviation, wherein the method specifically comprises the following steps:
the first step is as follows: determining a wide spectral band according to an incoherent broadband cavity enhanced absorption spectrum measurement system; assuming that the wave band is [ a, b ], wherein a is the minimum wavelength, b is the maximum wavelength, and the difference between b and a is not less than 30nm and is an even number;
the second step is that: in the wave bands [ a, b]Internally divided into a plurality of sub-bandsWherein, BWjIs the width, BW, of a sub-bandj=10+2j,λijIs the center wavelength of the sub-band,i=0,1,2,…,b-a-BWj;
the third step: selecting sub-bandsAs a spectrum fitting wave band, fitting the absorption cross section of the gas to be measured to the measured absorption coefficient by adopting a least square method to obtain a fitting concentration value N corresponding to all sub-wave bandsijUncertainty of fit EijAnd standard deviation of fit residual Dij(ii) a Calculating relative fit uncertainty REij,
The fourth step: sample data, i.e. sub-band width BW, by using maximum-minimum normalization methodjCentral wavelength of sub-band lambdaijRelative fit uncertainty REijAnd fitted residual standard deviation DijNormalization processing is carried out, and the normalized results are respectivelyAnd
the fifth step: establishing BP neural network, the number of nodes of network input layer is 2, and the input quantity is respectivelyAndthe number of nodes of the network output layer is 2, and the output quantity isThe number of nodes of the network hidden layer can be adjusted according to the training result; dividing a data sample corresponding to input-output quantity into a learning set and a testing set, and learning and testing the BP neural network;
and a sixth step: uncertainty RE of expected relative fit using max-min normalizationEAnd standard deviation of fit residual DECarrying out normalization treatment to obtainAndobtaining the expected sub-band width with normalized output quantity of the network as the input quantity of the learned BP neural networkSum sub-band center wavelength
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010394233.XA CN111487211B (en) | 2020-05-11 | 2020-05-11 | Incoherent broadband cavity enhanced absorption spectrum fitting waveband selection method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010394233.XA CN111487211B (en) | 2020-05-11 | 2020-05-11 | Incoherent broadband cavity enhanced absorption spectrum fitting waveband selection method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN111487211A true CN111487211A (en) | 2020-08-04 |
CN111487211B CN111487211B (en) | 2022-09-30 |
Family
ID=71811959
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010394233.XA Active CN111487211B (en) | 2020-05-11 | 2020-05-11 | Incoherent broadband cavity enhanced absorption spectrum fitting waveband selection method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111487211B (en) |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103712939A (en) * | 2013-12-30 | 2014-04-09 | 张显超 | Pollutant concentration fitting method based on ultraviolet-visible spectrum |
CN109001136A (en) * | 2018-09-20 | 2018-12-14 | 杭州绿洁水务科技股份有限公司 | A kind of COD on-line monitoring method based on ultraviolet visible light absorption spectrum |
CN110414729A (en) * | 2019-07-19 | 2019-11-05 | 西北农林科技大学 | The potential maximum photosynthetic capacity prediction technique of plant based on characteristic wavelength |
CN110991064A (en) * | 2019-12-11 | 2020-04-10 | 广州城建职业学院 | Soil heavy metal content inversion model generation method and system, storage medium and inversion method |
-
2020
- 2020-05-11 CN CN202010394233.XA patent/CN111487211B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103712939A (en) * | 2013-12-30 | 2014-04-09 | 张显超 | Pollutant concentration fitting method based on ultraviolet-visible spectrum |
CN109001136A (en) * | 2018-09-20 | 2018-12-14 | 杭州绿洁水务科技股份有限公司 | A kind of COD on-line monitoring method based on ultraviolet visible light absorption spectrum |
CN110414729A (en) * | 2019-07-19 | 2019-11-05 | 西北农林科技大学 | The potential maximum photosynthetic capacity prediction technique of plant based on characteristic wavelength |
CN110991064A (en) * | 2019-12-11 | 2020-04-10 | 广州城建职业学院 | Soil heavy metal content inversion model generation method and system, storage medium and inversion method |
Non-Patent Citations (3)
Title |
---|
NEERAJ PRAKASH ET AL.: "Near-infrared incoherent broadband cavity enhanced absorption spectroscopy (NIR-IBBCEAS) for detection and quantification of natural gas components", 《ANALYST》 * |
唐科等: "基于Stacking集成学习模型的气态亚硝酸预测", 《中国环境科学》 * |
薛宇等: "基于RBF神经网络的非色散红外SF_6气体传感器", 《光子学报》 * |
Also Published As
Publication number | Publication date |
---|---|
CN111487211B (en) | 2022-09-30 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109493287B (en) | Deep learning-based quantitative spectral data analysis processing method | |
CN104089911B (en) | Spectral model transmission method based on one-variable linear regression | |
CN104020135B (en) | Calibration model modeling method based near infrared spectrum | |
CN105372198B (en) | Infrared spectroscopy Wavelength selecting method based on integrated L1 regularization | |
CN111563436B (en) | Infrared spectrum measuring instrument calibration migration method based on CT-CDD | |
CN102305772A (en) | Method for screening characteristic wavelength of near infrared spectrum features based on heredity kernel partial least square method | |
CN110736707B (en) | Spectrum detection optimization method for transferring spectrum model from master instrument to slave instrument | |
Zhang et al. | Accurate and nondestructive detection of apple brix and acidity based on visible and near-infrared spectroscopy | |
Wen et al. | Ptychography imaging by 1-D scanning with a diffuser | |
CN111487211B (en) | Incoherent broadband cavity enhanced absorption spectrum fitting waveband selection method | |
CN111398200B (en) | Method for determining optimal inversion waveband of incoherent broadband cavity enhanced absorption spectrum | |
Doner et al. | Machine learning models for binary molecular classification using VUV absorption spectra | |
Huang et al. | Optimal wavelength selection for hyperspectral scattering prediction of apple firmness and soluble solids content | |
CN105004707A (en) | CCD array on-line Raman spectrometer spectra standardization method | |
CN114062306B (en) | Near infrared spectrum data segmentation preprocessing method | |
CN115015120B (en) | Fourier infrared spectrometer and temperature drift online correction method thereof | |
Shang et al. | Pink noise removal and spectral distortion correction based fiber Bragg grating demodulation algorithm | |
CN112834448B (en) | Spectral decomposition method for water pollutant analysis | |
Fade et al. | Minimum description length approach for unsupervised spectral unmixing of multiple interfering gas species | |
Zhang et al. | A Laplace sensitivity operator enhances the calculation efficiency of OCD metrology | |
CN117388198B (en) | Water quality COD detection method and detection equipment based on multi-source spectrum fusion | |
Ren et al. | Application of a wavelet packet transform based radial basis function neural network to analyze overlapping spectra | |
Ren et al. | Resolution of overlapping signals in spectrometry using a wavelet packet transform and an Elman recurrent neural network | |
CN115329809A (en) | Spectrum smoothing method suitable for infrared spectrum data | |
CN118035649A (en) | Spectral data processing and analyzing method based on spectrogram |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
EE01 | Entry into force of recordation of patent licensing contract | ||
EE01 | Entry into force of recordation of patent licensing contract |
Application publication date: 20200804 Assignee: Hefei Zhongke Guangbo Quantum Technology Co.,Ltd. Assignor: Anhui University of Science and Technology Contract record no.: X2023980043040 Denomination of invention: Band selection method for fitting incoherent broadband cavity enhanced absorption spectra Granted publication date: 20220930 License type: Exclusive License Record date: 20231010 |