CN114739959A - Device and method for online rapid detection of total nitrogen of aliasing fluorescence spectrum based on low-rank representation - Google Patents

Device and method for online rapid detection of total nitrogen of aliasing fluorescence spectrum based on low-rank representation Download PDF

Info

Publication number
CN114739959A
CN114739959A CN202210247492.9A CN202210247492A CN114739959A CN 114739959 A CN114739959 A CN 114739959A CN 202210247492 A CN202210247492 A CN 202210247492A CN 114739959 A CN114739959 A CN 114739959A
Authority
CN
China
Prior art keywords
low
rank
total nitrogen
aliasing
fluorescence 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.)
Pending
Application number
CN202210247492.9A
Other languages
Chinese (zh)
Inventor
阳春华
耿静轩
李勇刚
桂卫华
张凤雪
蓝丽娟
韩洁
周灿
朱红求
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.)
Central South University
Original Assignee
Central South 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 Central South University filed Critical Central South University
Priority to CN202210247492.9A priority Critical patent/CN114739959A/en
Publication of CN114739959A publication Critical patent/CN114739959A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
    • G01N21/62—Systems in which the material investigated is excited whereby it emits light or causes a change in wavelength of the incident light
    • G01N21/63—Systems in which the material investigated is excited whereby it emits light or causes a change in wavelength of the incident light optically excited
    • G01N21/64—Fluorescence; Phosphorescence
    • 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
    • G01N21/01—Arrangements or apparatus for facilitating the optical investigation
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
    • G01N21/62—Systems in which the material investigated is excited whereby it emits light or causes a change in wavelength of the incident light
    • G01N21/63—Systems in which the material investigated is excited whereby it emits light or causes a change in wavelength of the incident light optically excited
    • G01N21/64—Fluorescence; Phosphorescence
    • G01N2021/6417—Spectrofluorimetric devices
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
    • Y02A20/00—Water conservation; Efficient water supply; Efficient water use
    • Y02A20/20—Controlling water pollution; Waste water treatment

Landscapes

  • Health & Medical Sciences (AREA)
  • Physics & Mathematics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Chemical & Material Sciences (AREA)
  • Analytical Chemistry (AREA)
  • Biochemistry (AREA)
  • General Health & Medical Sciences (AREA)
  • General Physics & Mathematics (AREA)
  • Immunology (AREA)
  • Pathology (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Investigating, Analyzing Materials By Fluorescence Or Luminescence (AREA)

Abstract

The invention discloses an aliasing fluorescence spectrum total nitrogen online rapid detection method based on low-rank representation, which changes the light intensity of an excitation light source by adjusting a wide-spectrum excitation light source, so as to excite a solution to be detected to generate an aliasing fluorescence spectrum under different light intensities; meanwhile, a micro fluorescence spectrometer is used for recording the whole change process of the fluorescence spectrogram of the solution to be detected under the change of the excitation light intensity in an in-situ high-frequency manner; by combining a low-rank characteristic feature extraction algorithm, according to different excitation characteristics of the fluorescence spectrum under different light intensities, redundant noise is removed, and a low-dimensional atomic feature characteristic subspace is adaptively established; and adding a sparsity constraint hypothesis, introducing a least square regression term, and constructing a total nitrogen spectrum detection model based on variable light intensity aliasing fluorescence spectrum to realize rapid online detection of the total nitrogen concentration of the water sample to be detected.

Description

Device and method for online rapid detection of total nitrogen of aliasing fluorescence spectrum based on low-rank representation
Technical Field
The invention belongs to the field of automatic detection, and particularly relates to an aliasing fluorescence spectrum total nitrogen online rapid detection device and method based on low-rank representation.
Background
The shortage of water resources can bring great influence to social production and life of people, and sewage treatment is taken as a key link of cyclic utilization of the water resources, so that the stable and efficient normal operation of the sewage treatment system is of great significance. The water quality parameters are used as parameters for representing physical, chemical and biochemical characteristics of various substances in the water body, are characteristic indexes for measuring the quality degree and the change trend of the water body, and are also important basis for optimizing and controlling the sewage treatment process. The total nitrogen is used as a main water quality pollution parameter, the pollution degree of nitrogen-containing substances in the water body can be intuitively reflected, and the rapid and stable detection of the total nitrogen has important significance for preventing water quality environmental pollution disasters such as water eutrophication and the like. However, the conventional total nitrogen online detection technology needs a long oxidation digestion process, so that the speed and the precision of the total nitrogen online detection are limited, and multiple chemical reagents used in the detection process, such as poor treatment, can cause secondary pollution to the environment. The invention provides an aliasing fluorescence spectrum total nitrogen online rapid detection method based on low-rank representation, which comprises the steps of obtaining a fluorescence spectrogram of a sample liquid to be detected under different excitation light intensities by changing excitation light intensity of a broadband excitation light source, extracting fluorescent substance information in a variable light intensity fluorescence aliasing spectrum in a self-adaptive manner based on a low-rank representation feature extraction algorithm, and finally constructing a total nitrogen rapid detection model based on the variable light intensity fluorescence spectrum to realize the purpose of rapidly detecting the total nitrogen concentration of a water sample to be detected.
The fluorescence spectrum is an excitation spectrum, the principle is that a substance receives excitation after absorbing electromagnetic radiation, and molecules or atoms emit radiation with the same or different wavelength as excitation radiation waves in the de-excitation process. The traditional fluorescence water sample detection is usually based on that a monochromatic laser excites a water body to be detected to generate fluorescence, but considering that the actual water body substance components are complex, and a fluorescence spectrum line generated by single excitation light intensity cannot reflect rich substance information in the water body, the invention provides a method for detecting the water body substance by using a more general broad spectrum light source to excite an aliasing fluorescence spectrum: the main fluorescent substance components in the water body are qualitatively and quantitatively analyzed by utilizing the fluorescence expression difference of different water quality components under different excitation light intensities and assisting a characteristic extraction algorithm, so that the defects of aliasing of fluorescence excited by different wavelengths and difficulty in substance information extraction caused by fluorescence excited by a broad spectrum are overcome.
As a new high-latitude data analysis and processing method, low rank characterization has been used in a number of popular fields such as machine vision, statistical analysis, signal processing, etc. The low-rank representation can be regarded as popularization and development of a compressed sensing theory, the rank of the matrix is used as a sparse measure, and the low-latitude feature subspace is found from the high-latitude observation sample, so that the feature information of the observation sample can be effectively extracted, and noise and redundancy are eliminated. However, in the existing technical scheme, a good optimization mode for applying low-rank representation to water sample detection is not available, and the rapidity and the accuracy of water sample detection have wide market demands and are valued by various research groups and institutions.
Disclosure of Invention
Aiming at the problems in the prior art, the first purpose of the invention is to provide the device for rapidly detecting the total nitrogen of the aliasing fluorescence spectrum based on the low-rank representation, the wide-spectrum fluorescence excitation light source is used for exciting the fluorescence aliasing spectrum of the sample liquid to be detected, the light intensity of the excitation light source is adjusted by changing the light transmission area of the collimating mirror, and the fluorescent substance of the water body to be detected is induced to generate different fluorescence excitation reactions, so that the water sample information is collected more comprehensively. The device is wide in spectrum measurement range, high in sensitivity, high in response speed and accurate in test result, and can meet the requirements of rapidness and accuracy of water body detection.
The detection device provided by the invention is used for intelligently extracting fluorescence aliasing spectrum substance characteristic information under the change of excitation light intensity through a low-rank representation algorithm, and establishing a total nitrogen concentration detection model based on fluorescence aliasing spectrum to realize the rapid online detection of the total nitrogen concentration of the actual water sample.
In order to achieve the technical purpose, the invention provides an aliasing fluorescence spectrum total nitrogen online rapid detection device based on low-rank representation, which comprises: the device comprises a light source, a backlight condenser, a condenser prism, a shading baffle, a collimating mirror, a cuvette and a CCD probe; the backlight condenser, the light source, the condensing prism, the shading baffle, the collimating mirror and the cuvette are sequentially distributed on the same axis, and the CCD probe is distributed on one side of the smooth surface of the cuvette in a manner of being perpendicular to the axis.
As a preferred scheme, the light source is a broad spectrum excitation light source; the shading baffle is adjustable. By adjusting the area of the wide-spectrum excitation light source penetrating through the collimating mirror and changing the excitation light intensity of the light source from weak to strong, the aliasing fluorescence spectrum change of the solution to be detected under different excitation light intensities in situ detection can be obtained, and a plurality of groups of calibration water sample aliasing fluorescence change maps with different total nitrogen concentrations are collected. In order to cooperate with the subsequent light splitting process, the traditional light source generally adopts a narrow light source, so that although the testing precision is ensured, a large amount of water sample information is lost. Compared with the traditional spectrum excitation light source, the wide spectrum excitation light source is adopted to excite more collected water sample information, and then the subsequent algorithm optimization is carried out, so that the water sample information can be better utilized, and the model is more accurate and rapid.
In a preferred embodiment, the CCD probe pixel matrix is 4K × 4K or 3K × 3K, and the pixel size is 108 μm or 140 μm.
The invention also provides a low-rank representation-based method for rapidly detecting total nitrogen of aliasing fluorescence spectrum on line, which mainly comprises the following steps: 1) equally dividing a sample to be detected into two parts, acquiring a true value from one part, and matrixing the other part by an aliasing fluorescence spectrum to obtain a high-dimensional aliasing matrix; 2) extracting a low-rank matrix in a high-dimensional aliasing matrix and optimizing the low-rank matrix; 3) and constructing a total nitrogen detection SVM model according to the optimized low-rank matrix and training to obtain the total nitrogen detection SVM model.
The low-rank representation algorithm is used for feature extraction of fluorescence aliasing spectra under variable light intensity, spectral information of different fluorescent substances in a water sample excited by the excitation light intensity under the change can be effectively extracted, and redundant information irrelevant to total nitrogen concentration detection is further eliminated through sparse weight distribution, so that a total nitrogen detection model is constructed, and the accuracy and speed of total nitrogen online detection are improved
As a preferred scheme, the liquid to be detected is urban sewage with different concentrations; the true value is the total nitrogen national standard detection value of the liquid to be detected. Dividing total nitrogen sample liquid to be detected with different concentrations into two parts in equal proportion, wherein one part is detected by using a total nitrogen national standard detection method GB11894-89 as a true value; and the other part is used for collecting fluorescence spectrum change data in the process of exciting light intensity change. Diluting the solution to be detected before adding the solution to be detected into the cuvette, wherein the diluted solution adopts ultrapure water and the resistivity is not less than 18.3 megohm.
As a preferable scheme, the aliasing fluorescence spectrum is the aliasing of fluorescence change spectrograms of the solution to be measured in situ under different intensities of excitation light to be measured, the measurement range of the fluorescence spectrum is 200 nm-1100 nm, the scanning time interval t is 0.1-0.5 s, and the moving speed of the shading plate is 0.02-0.1 cm/s.
As a preferred solution, the matrixing is a process of collecting fluorescence changes in situ by using the intensity change of the excitation light and constructing a high-dimensional aliasing matrix.
As a preferred scheme, the low rank matrix extraction process is: and constructing a low-rank characteristic extraction target loss function through a high-dimensional aliasing matrix, and introducing low-rank constraint and structured sparse constraint on the low-rank matrix.
As a preferred scheme, the optimization mode is an augmented lagrange multiplier method, and the optimization is performed until low-rank constraint and sparse constraint convergence.
As a preferred scheme, the target loss function of the low rank feature extraction is a least square loss function of a product of a low rank matrix and a sparse representation and a high dimensional matrix. The low rank target loss function is:
Figure BDA0003545616190000031
p, X, W are derived from the matrixing process, where P is the original high-dimensional sample data, X is the isolated low-rank token subspace, W is the sparse representation of the sample data,
as a preferred scheme, the low-rank constraint of the low-rank matrix is introduced by a nuclear norm, and the sparsity constraint of the low-rank matrix is introduced by a combined norm. After introducing the low rank constraint and the sparse constraint, the low rank target loss function becomes:
Figure BDA0003545616190000032
Figure BDA0003545616190000033
λ is the loss weight. Effective information of the low-rank constraint separation aliasing fluorescence spectrum is introduced, low-rank representation of different nitrogen-containing substances in the high-dimensional aliasing spectrum is mined, the low-rank constraint fundamentally aims at constraining the rank of the matrix, and the nuclear norm convex is similar to the rank of the matrix, so that the low-rank constraint is introduced through the nuclear norm. And the sparsity constraint of the low-rank matrix is introduced to ensure the sparsity of the matrix, and the effect is to ignore elements in the matrix with low or no correlation on the calculation result.
As a preferred scheme, the penalty term obtaining manner of the augmented lagrange multiplier method is at least one of an inner product method, an outer product method and a hybrid method.
As a preferred scheme, the total nitrogen detection model construction process is as follows: and constructing a target loss function through the total nitrogen concentration, introducing the sparse constraint of the time regression weight and the sparse constraint of the space regression weight, and optimizing through a linear alternating direction multiplier method. Because the function contains two variables of time regression weight and space regression weight, the direct adoption of the augmented Lagrange multiplier method is complex, and the optimization of the linear alternating direction multiplier method can well solve the optimization problem of the multivariate variables.
As a preferred scheme, the total nitrogen concentration is used as a true value, and the constructed target loss function is a time regression weight multiplied by a sparse constraint multiplied byA least squares loss function of the spatial regression weight and total nitrogen concentration. The target loss function for total nitrogen concentration is:
Figure BDA0003545616190000041
y is the total nitrogen concentration value of the liquid to be detected, W is the sparse representation of sample data, T is the sparse representation time regression weight of the regression model, and D is the sparse representation space regression weight of the regression model.
As a preferred scheme, the sparsity constraint of the temporal regression weight and the sparsity constraint of the spatial regression weight are introduced by a two-norm method. After introducing the constraint term, the following results are obtained:
Figure BDA0003545616190000042
λDand λTThe sparse constraint for T and D loses weight. L is a radical of an alcohol2The norm limits the size of the parameter and thus the size of the component, and therefore, L2The norm can improve the generalization capability of the model. Moreover, when the condition number is large, it is difficult to directly perform matrix inversion, and moreover, too large cond value will result in λ -gradient constant, but L passes2The norm can effectively solve the problems.
The invention has the beneficial technical effects that:
1. according to the invention, the fluorescence aliasing spectrum excited by the broad spectrum light source under the condition of light intensity change is utilized, and the total nitrogen rapid detection model is established by introducing sparse constraint based on the low-rank representation trait extraction algorithm, so that the long oxidation digestion reaction step in the traditional total nitrogen detection process is avoided, and the speed and the precision of total nitrogen online detection are improved.
2. According to the invention, through algorithm optimization, redundant information in a spectrum is removed, the dimension and the calculated amount of the model are reduced while the model precision is ensured, so that the rapid detection is realized, and in addition, the model can be independently learned and adjusted along with the increase of detection data, so that the model precision is further improved.
Drawings
FIG. 1 is a flow chart of an aliasing fluorescence spectrum total nitrogen online rapid detection method based on low rank characterization;
FIG. 2 is a schematic structural diagram of a total nitrogen-to-light intensity fluorescence aliasing spectrum measurement system of the method;
FIG. 3 is a schematic diagram of a low rank feature representation subspace feature extraction algorithm;
FIG. 4 is a variable intensity fluorescence aliased spectrum of an actual water sample.
Detailed Description
Model establishment and training:
the first step is as follows: changing the excitation light intensity of a light source from weak to strong by adjusting the transmission area of a wide-spectrum excitation light source collimating mirror, simultaneously detecting the aliasing fluorescence spectrum change of the solution to be detected under different excitation light intensities in situ by using a micro fluorescence spectrometer, and collecting a plurality of groups of calibration water sample aliasing fluorescence change maps with different total nitrogen concentrations;
step 1: dividing the total nitrogen sample solution to be detected with different concentrations into two parts in equal proportion, and detecting one part by using a total nitrogen national standard detection method GB11894-89 as a true value; the other part is used for collecting fluorescence spectrum change data in the excitation light intensity change process;
and 2, step: injecting 0.4ml of solution to be detected into a quartz detection vessel shown in FIG. 2, and adding water to dilute the solution to 4 ml;
wherein the diluting water is ultrapure water prepared by a water purifying instrument, and the resistivity of the ultrapure water meets the national standard and is more than 18.3 megaohms;
and 3, step 3: detecting the fluorescence spectrum change of the diluted liquid to be detected at time intervals of t, and simultaneously moving the shading baffle plate shown in the figure 2 according to the speed v to change the light intensity of the excitation light source from weak to strong until the excitation light intensity reaches a rated value;
wherein the measurement range of the fluorescence spectrum is 200 nm-1100 nm, the scanning time interval t is 0.5s, and the moving speed of the shading plate is 0.1 cm/s;
the second step is that: simultaneously extracting the characteristics of the multiple groups of calibration water samples collected in the step 1 by adopting a low-rank characteristic extraction algorithm, removing redundant information and noise in the original variable light intensity fluorescence aliasing spectrum, adaptively learning a low-dimensional characteristic representation subspace under the current water body, and extracting main material information of the aliasing fluorescence spectrum;
step 1: dividing total nitrogen sample liquid to be detected with different concentrations into two parts in equal proportion, and detecting one part by using a total nitrogen national standard detection method GB11894-89 as a true value; the other part is used for collecting fluorescence spectrum change data in the excitation light intensity change process;
step 2: injecting 0.4ml of solution to be detected into a quartz detection vessel shown in FIG. 2, and adding water to dilute the solution to 4 ml;
wherein the diluting water is ultrapure water prepared by a water purifying instrument, and the resistivity of the ultrapure water meets the national standard and is more than 18.3 megaohms;
and step 3: detecting the change of the fluorescence spectrum of the diluted liquid to be detected at time intervals of t, moving the shading baffle plate shown in the figure 2 according to the speed v, and changing the light intensity of the excitation light source from weak to strong until the excitation light intensity reaches a rated value;
wherein the measurement range of the fluorescence spectrum is 200 nm-1100 nm, the scanning time interval t is 0.5s, and the movement speed of the light shielding plate is 0.1 cm/s;
the third step: introducing sparsity constraint and a least square regression term, constructing a total nitrogen rapid detection model based on variable light intensity fluorescence aliasing spectrum based on the total nitrogen concentration value of the calibrated water sample, and realizing online detection of total nitrogen based on the aliasing fluorescence spectrum.
Step 1: constructing a total nitrogen detection fluorescence spectrum regression model by taking a least square loss function of a formula (4) as a main target loss function;
wherein the target loss function of least squares is:
Figure BDA0003545616190000051
y is a calibration sample total nitrogen concentration value, T is a regression model sparse representation time regression weight D is a regression model sparse representation space regression weight;
step 2: introducing sparse feature constraint terms to ensure the sparsity of regression weights T and D;
wherein the sparse constraint is introduced through a two-norm, and a target loss function for introducing the sparse constraint is as follows:
Figure BDA0003545616190000061
λDand λTLoss of weight for sparse constraints of T and D;
and 3, step 3: and respectively solving the optimal values of T and D by using a linear alternating direction multiplier optimization algorithm to minimize a target loss function, and then constructing a total nitrogen online detection model based on the variable light intensity aliasing fluorescence spectrum.
The embodiment is as follows:
in order to make the objects, technical solutions and advantages of the embodiments 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 drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. 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.
Take the rapid online detection of the total nitrogen concentration of a sewage pipeline in city of a certain community as an example. Firstly, collecting spectrum data (figure 4) by using the variable light intensity fluorescence aliasing spectrum measuring system shown in figure 2, and determining the total nitrogen concentration of a sample liquid to be measured as a true value according to a national standard method; secondly, constructing a low-rank atomic subspace and sparse feature representation thereof by using a low-rank representation algorithm based on the variable light intensity fluorescence aliasing spectrum data collected in the first step, eliminating noise interference in an observation high-dimensional spectrogram, and extracting material information in the variable light intensity fluorescence aliasing spectrum; and thirdly, taking a least square loss function as a core, introducing a sparse constraint term, constructing a regression model target loss function, and further optimizing a target function loss value through a linear alternating direction multiplier method, so as to establish a total nitrogen rapid detection model based on a variable light intensity fluorescence aliasing spectrum. The method is compared with the traditional total nitrogen online detection method in terms of detection error, repeatability, detection duration and the like, the obtained results are shown in table 1, and as can be seen from table 1, the method is superior to the traditional detection method in terms of relative error, repeatability, detection duration and the like.
TABLE 1 comparison of total nitrogen rapid detection method and traditional on-line detection method
Figure BDA0003545616190000062

Claims (9)

1. An aliasing fluorescence spectrum total nitrogen online rapid detection device based on low-rank characterization is characterized by comprising: the device comprises a light source, a backlight condenser, a condenser prism, a shading baffle, a collimating mirror, a cuvette and a CCD probe; the backlight condenser, the light source, the condensing prism, the shading baffle, the collimating mirror and the cuvette are sequentially distributed on the same axis, and the CCD probe is distributed on one side of the smooth surface of the cuvette in a manner of being perpendicular to the axis.
2. The device for online rapid detection of total nitrogen of aliased fluorescence spectrum based on low-rank characterization according to claim 1, wherein: the light source is a wide-spectrum excitation light source; the shading baffle is adjustable.
3. The device for online rapid detection of total nitrogen of aliased fluorescence spectrum based on low-rank characterization according to claim 1, wherein: the pixel matrix of the CCD probe is 4 Kx 4K or 3 Kx 3K, and the pixel size is 108 mu m or 140 mu m.
4. The method for rapidly detecting total nitrogen on line based on the aliasing fluorescence spectrum with the low rank characterization as claimed in any one of claims 1-3, characterized by comprising the following steps: 1) equally dividing a sample to be detected into two parts, acquiring a true value from one part, and matrixing the other part by an aliasing fluorescence spectrum to obtain a high-dimensional aliasing matrix; 2) extracting a low-rank matrix in a high-dimensional aliasing matrix and optimizing the low-rank matrix; 3) and constructing a total nitrogen detection SVM model according to the optimized low-rank matrix and training to obtain the total nitrogen detection SVM model.
5. The method for on-line rapid detection of total nitrogen of aliasing fluorescence spectrum based on low rank characterization according to claim 4, characterized in that: the liquid to be detected is urban sewage with different concentrations; the true value is the national standard detection value of the total nitrogen of the liquid to be detected; the aliasing fluorescence spectrum is obtained by aliasing the fluorescence change spectrogram of the in-situ solution to be detected under different excitation light intensities to be detected, the measurement range of the fluorescence spectrum is 200 nm-1100 nm, the scanning time interval t is 0.1-0.5 s, and the movement speed of the light shielding plate is 0.02-0.1 cm/s; the matrixing is a process of collecting fluorescence changes in situ by using the light intensity change of the exciting light and constructing a high-dimensional aliasing matrix.
6. The method for on-line rapid detection of total nitrogen of aliasing fluorescence spectrum based on low rank characterization according to claim 4, characterized in that: the low-rank matrix extraction process comprises the following steps: constructing a low-rank characteristic extraction target loss function through a high-dimensional aliasing matrix, and introducing low-rank constraint and structured sparse constraint on the low-rank matrix; the optimization mode is an augmented Lagrange multiplier method, and the optimization is carried out until low-rank constraint and sparse constraint convergence.
7. The method for the online rapid detection of total nitrogen based on the low-rank-characterization aliasing fluorescence spectrum according to any one of claim 6, wherein the method comprises the following steps: the target loss function extracted by the low-rank characteristic is a least square loss function of a product of a low-rank matrix and sparse representation and a high-dimensional matrix; the low-rank constraint of the low-rank matrix is introduced through a nuclear norm, and the sparse constraint of the low-rank matrix is introduced through a combined norm; the penalty term obtaining mode of the augmented Lagrange multiplier method is at least one of an inner product method, an outer product method and a mixed method.
8. The method for on-line rapid detection of total nitrogen of aliasing fluorescence spectrum based on low rank characterization according to claim 4, characterized in that: the total nitrogen detection model construction process comprises the following steps: and constructing a target loss function through the total nitrogen concentration, introducing the sparse constraint of the time regression weight and the sparse constraint of the space regression weight, and optimizing through a linear alternating direction multiplier method.
9. The method for on-line rapid detection of total nitrogen of aliasing fluorescence spectrum based on low rank characterization according to claim 8, characterized in that: the target loss function constructed through the total nitrogen concentration is a least square loss function of time regression weight multiplied by sparse constraint multiplied by space regression weight and the total nitrogen concentration; the sparse constraint of the temporal regression weight and the sparse constraint of the spatial regression weight are introduced by a two-norm.
CN202210247492.9A 2022-03-14 2022-03-14 Device and method for online rapid detection of total nitrogen of aliasing fluorescence spectrum based on low-rank representation Pending CN114739959A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210247492.9A CN114739959A (en) 2022-03-14 2022-03-14 Device and method for online rapid detection of total nitrogen of aliasing fluorescence spectrum based on low-rank representation

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210247492.9A CN114739959A (en) 2022-03-14 2022-03-14 Device and method for online rapid detection of total nitrogen of aliasing fluorescence spectrum based on low-rank representation

Publications (1)

Publication Number Publication Date
CN114739959A true CN114739959A (en) 2022-07-12

Family

ID=82275581

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210247492.9A Pending CN114739959A (en) 2022-03-14 2022-03-14 Device and method for online rapid detection of total nitrogen of aliasing fluorescence spectrum based on low-rank representation

Country Status (1)

Country Link
CN (1) CN114739959A (en)

Similar Documents

Publication Publication Date Title
US7570357B2 (en) Visible/near-infrared spectrometry and its device
CN103983595B (en) A kind of water turbidity calculation method processed based on ultraviolet-visible spectrum
Fontana et al. Raman spectroscopic sensors for inorganic salts
CN103175805B (en) Method for determining indexes of COD and BOD5 in sewage through near infrared spectrometry
CN101750024A (en) Near infrared spectrum detection method for insulation coating layer thickness of silicon steel
CN109001182B (en) Raman spectrum nondestructive testing method for alcohol content in closed container
CN101701911A (en) Method for detecting content of reducing sugar in fermentation materials
CN101943658A (en) Method for continuously monitoring water quality by combining principal component analysis theory and spectrum analysis technology
Li et al. A feasibility study on quantitative analysis of low concentration methanol by FT-NIR spectroscopy and aquaphotomics
CN100443883C (en) Method for detecting hydrgenated tail-oil paraffin composition using near-infrared spectrum
Cai et al. Quantitative analysis of stibnite content in raw ore by Raman spectroscopy and chemometric tools
Bürck Spectrophotometric determination of uranium and nitric acid by applying partial least-squares regression to uranium (VI) absorption spectra
Zhou et al. Detection of chemical oxygen demand in water based on UV absorption spectroscopy and PSO-LSSVM algorithm
CN104914065A (en) Sewage spectrum analyzer
CN115615976A (en) Water quality chemical oxygen demand detection method based on laser spectrum technology
CN114739959A (en) Device and method for online rapid detection of total nitrogen of aliasing fluorescence spectrum based on low-rank representation
CN116399836A (en) Cross-talk fluorescence spectrum decomposition method based on alternating gradient descent algorithm
CN115656144A (en) Method and device for measuring seawater salinity in situ based on photoacoustic information fusion and application
Wang et al. Research Article Quantitative Analysis of Multiple Components in Wine Fermentation using Raman Spectroscopy
CN103398966A (en) Method for detecting TMC concentration in organic solution by using spectrometer
CN109799225B (en) Method for measuring element content based on resonance and non-resonance double lines
CN113588597A (en) Method for improving analysis precision of furnace slag
CN101701910A (en) Method for detecting brix, volatile acid and alcohol degree of fermenting materials
CN112595706A (en) Laser-induced breakdown spectroscopy variable selection method and system
CN111829997A (en) Crude oil excitation path system based on fluorescence yield index

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