CN114936513A - Method and system for improving detection precision of laser-induced breakdown spectroscopy - Google Patents

Method and system for improving detection precision of laser-induced breakdown spectroscopy Download PDF

Info

Publication number
CN114936513A
CN114936513A CN202210434868.7A CN202210434868A CN114936513A CN 114936513 A CN114936513 A CN 114936513A CN 202210434868 A CN202210434868 A CN 202210434868A CN 114936513 A CN114936513 A CN 114936513A
Authority
CN
China
Prior art keywords
libs
particle
dimension
algorithm
sample
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
CN202210434868.7A
Other languages
Chinese (zh)
Inventor
王�琦
王海萍
张鹏飞
徐琢频
吴跃进
詹玥
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.)
Hefei Institutes of Physical Science of CAS
Original Assignee
Hefei Institutes of Physical Science of CAS
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 Hefei Institutes of Physical Science of CAS filed Critical Hefei Institutes of Physical Science of CAS
Priority to CN202210434868.7A priority Critical patent/CN114936513A/en
Publication of CN114936513A publication Critical patent/CN114936513A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
    • 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
    • G01N21/71—Systems in which the material investigated is excited whereby it emits light or causes a change in wavelength of the incident light thermally excited
    • G01N21/718—Laser microanalysis, i.e. with formation of sample plasma
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/25—Design optimisation, verification or simulation using particle-based methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/004—Artificial life, i.e. computing arrangements simulating life
    • G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
    • 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
    • G01N2021/0106—General arrangement of respective parts
    • G01N2021/0112—Apparatus in one mechanical, optical or electronic block
    • 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
    • Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
    • Y02E30/00—Energy generation of nuclear origin
    • Y02E30/30—Nuclear fission reactors

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Health & Medical Sciences (AREA)
  • Evolutionary Computation (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • General Engineering & Computer Science (AREA)
  • General Health & Medical Sciences (AREA)
  • Computer Hardware Design (AREA)
  • Chemical & Material Sciences (AREA)
  • Software Systems (AREA)
  • Pathology (AREA)
  • Artificial Intelligence (AREA)
  • Immunology (AREA)
  • Biochemistry (AREA)
  • Geometry (AREA)
  • Analytical Chemistry (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Plasma & Fusion (AREA)
  • Medical Informatics (AREA)
  • Optics & Photonics (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Investigating, Analyzing Materials By Fluorescence Or Luminescence (AREA)

Abstract

The invention discloses a method for improving detection precision of laser-induced breakdown spectroscopy, which comprises the following steps: the method comprises the following steps: detecting the content of elements in the sample by ICP; step two: collecting LIBS data of a sample; step three: selecting LIBS emission lines of the analytical elements; step four: selecting an element emission LIBS spectral line wave band by using a variable dimension particle swarm optimization-combined moving window algorithm; step five: and establishing an element quantitative analysis model by using the characteristic wave bands screened by the algorithm. The method has the advantages of accurate detection result, quick, simple and convenient detection process, green and safety, and can effectively improve the detection precision of the laser-induced breakdown spectroscopy.

Description

Method and system for improving detection precision of laser-induced breakdown spectroscopy
Technical Field
The invention relates to the technical field of quantitative analysis of sample elements by laser-induced breakdown spectroscopy, in particular to a method and a system for improving detection precision of laser-induced breakdown spectroscopy.
Background
Laser-Induced Breakdown Spectroscopy (LIBS) is an atomic Spectroscopy technique that focuses pulsed Laser on the surface of a sample to excite the sample to generate a plasma emission spectrum, and the wavelength and intensity of the spectrum can be used to determine the type and relative content of elements in the sample, which can be used to detect the elements. Compared with the traditional element analysis methods, such as Inductively Coupled Plasma Emission Spectrometry (ICPES), Atomic Absorption Spectrophotometry (AAS), hydride generation-atomic fluorescence spectrometry (HG-AFS), ultraviolet-visible spectrophotometry (UV-VIS) and the like, the LIBS has the advantage of high speed of a spectrum detection technology, can directly detect a sample, does not need pretreatment, is simple to operate and low in cost, and can perform remote control and online analysis by simultaneously analyzing multiple components. The technology is suitable for various material forms such as solid, liquid, gas and the like, and has wide application in the fields of nuclear industry, environment, metallurgy, agricultural biology and the like. However, based on the stability and repeatability of the LIBS system, the LIBS spectral data changes due to the difference of the LIBS pulse laser energy, the interference of external background noise and system parameters, and the spectral lines change due to the influence of factors such as self-absorption, inter-element spectral line interference and matrix effect on the analysis of elements, so that the challenge is brought to accurate LIBS element detection. The appropriate characteristic variable algorithm can determine the most relevant variable with the element to be detected and eliminate the irrelevant variable, thereby reducing the influence of other factors on the accurate detection of the element and improving the analysis result.
[1]Method for reducing matrix effect in LIBS quantitative analysis]Journal of the university of Hebei (Nature science edition), 2015,37(2):76-79, [2 ]]Suanlan Xiang, Hai bin, Suzhibo, et alSi[J]Optical journal, 2010,30(09):2757-]Vincent M.,Alexander K.,Gordon R.O.,et al.Quantitative multi-elemental laser-induced breakdown spectroscopy using artificial neural networks[J]Journal of the European Optical Society-Rapid Publications,2008,3:08011-1-08011-5, all disclose the improvement of LIBS detection accuracy for the purpose of reducing the influence of matrix effects during detection. Experts take many ways. A calibration curve can be established according to LIBS element spectral lines, the influence of a matrix effect can be effectively reduced by utilizing a method for improving linear regression by utilizing the calibration curve, but a standard sample with the element type consistent with that of a sample to be measured needs to be configured, and the workload in the measurement process is increased [1] . The neural network calibration method which can utilize the LIBS technology and combines different network input modes can make full use of characteristic information in the spectrum compared with an internal standard method, effectively reduce the influence of matrix effect and weaken the interference between emission spectral lines, but the input mode of the neural network is strictly limited and can improve the repeatability and the accuracy of a measurement signal by reasonable selection [2] . The LIBS technology is combined with an error back propagation neural network (BP-ANN) to effectively improve the quantitative detection precision, but the selection of the initial connection weight and the threshold of the neural network has certain randomness, lacks of theoretical basis and needs to be further improved.
Disclosure of Invention
The invention aims to solve the technical problem of providing a method for improving the detection precision of laser-induced breakdown spectroscopy by combining a variable-dimension particle swarm optimization algorithm with a moving window algorithm.
The invention solves the technical problems through the following technical means:
a method for improving laser-induced breakdown spectroscopy detection comprises the following steps:
s01, collecting samples with element contents in gradient distribution as a modeling sample set, and detecting the element contents of each sample by using ICP (inductively coupled plasma);
s02, collecting LIBS data of the sample;
s03, selecting an LIBS emission spectral line of an analysis element;
s04, selecting LIBS emission spectral line wave bands of elements by adopting a variable dimension particle swarm optimization-combined moving window algorithm; the LIBS emission spectrum line wavelength is a particle, the size of an LIBS emission spectrum line interval is a window, the collection of the LIBS emission spectrum line wavelengths is a group, the LIBS emission spectrum line wave band needing to be screened is a particle swarm, the LIBS emission spectrum line and a chemical value are used as input of an algorithm, and the optimal spectrum interval position is output through calculation of a variable-dimension particle swarm optimization-combined moving window algorithm;
and S05, establishing an element quantitative analysis model by using the screened characteristic wave bands.
The method adopts a method of LIBS spectrum combined with a variable-dimension particle swarm optimization-combined moving window wavelength selection algorithm to accurately detect the content of elements in a sample, improves the traditional particle swarm algorithm, provides a variable-dimension idea, can search data spaces in different dimensions, and reduces the risk of limiting local extreme values and overfitting of the traditional particle swarm algorithm. Meanwhile, an algorithm provides a combined moving window strategy, and the spectral data variable interval is quickly selected through the combined moving window.
Further, the specific process of selecting the LIBS emission spectrum band of the element by using the wiener swarming optimization-combined moving window algorithm in step S04 is as follows:
(1) for M particles, equation D is used i ＝min(D max ,fix((i-1)/(M/D max ) +1) the size of the initialization particles; wherein D is i Is the dimension of the particle i, the position x of the particle i i With uniformly distributed random initialization, the velocity v of the particle i i Having the same dimension as its position and filled with zeros, fix () rounds the argument to the nearest integer, M is the number of particles of the population, D max Is the maximum dimension of the window;
(2) evaluating the value of the objective function for each particle, setting p bi For the current particle position, p g Is the particle position with the best objective function;
(3) updating the velocity of each particle; using equation v i ＝w×v i +c 1 ×rand()×(p bi -x i )+c 2 ×rand()×(p g x i ) Updating the position of each particleAnd a velocity vector; where w is the inertial weight, c 1 And c 2 For the acceleration coefficient, a rand () function is used for generating random numbers which are uniformly divided in an interval of 0-1; compare the optimal value of each particle to its current objective function and update p bi And p g Loop until iteration is complete, p g Is the result of the wavelength ultimately selected.
Further, step S03 is specifically to perform SNV preprocessing on the collected LIBS data to remove irrelevant noise and background, and then use a plurality of atomic emission spectral lines as LIBS emission spectral lines of some element; a number of spectral data points are taken near each LIBS emission band and the bands are fused to form the LIBS emission for the element.
Further, in step S05, an elemental quantitative analysis model is established by using a chemometric algorithm.
Further, the chemometric algorithm adopts a least square method to construct a quantitative analysis model of the elements.
Corresponding to the method, the invention also provides a system for improving the laser-induced breakdown spectroscopy detection, which comprises the following steps:
the sample selection module is used for selecting the element content of the ICP detection sample as a sample;
the sample data acquisition module is used for acquiring LIBS data of the sample;
the LIBS emission line selection module is used for selecting the LIBS emission line of the analysis element;
the LIBS emission spectral line wave band selection module is used for selecting the LIBS emission spectral line wave band of the element by adopting a variable dimension particle swarm optimization-combined moving window algorithm; the LIBS emission spectrum line wavelength is a particle, the size of an LIBS emission spectrum line interval is a window, the collection of the LIBS emission spectrum line wavelength is a group, the LIBS emission spectrum line wave band needing to be screened is a particle swarm, the LIBS emission spectrum line and a chemical value are used as the input of an algorithm, and the optimal spectrum interval position is output through calculation of a variable-dimension particle swarm optimization-combined moving window algorithm;
and the analysis model establishing module is used for establishing an element quantitative analysis model by utilizing the screened characteristic wave bands.
Further, the LIBS emission spectral line band selection module adopts a dimension particle swarm optimization-combined moving window algorithm to select the LIBS emission spectral line band of the element in a specific process that:
(1) for M particles, equation D is used i ＝min(D max ,fix((i-1)/(M/D max ) +1) the size of the initialization particles; wherein D is i Is the dimension of the particle i, the position x of the particle i i With uniformly distributed random initialization, the velocity v of the particles i i Having the same dimension as its position and filled with zeros, fix () rounds the argument to the nearest integer, M is the number of particles of the particle population, D max Is the maximum dimension of the window;
(2) evaluating the value of the objective function for each particle, setting p bi For the current particle position, p g Is the particle position with the best objective function;
(3) updating the velocity of each particle; using equation v i ＝w×v i +c 1 ×rand()×(p bi -x i )+c 2 ×rand()×(p g x i ) Updating the position and velocity vector of each particle; where w is the inertial weight, c 1 And c 2 For the acceleration coefficient, a rand () function is used for generating random numbers which are uniformly divided in an interval of 0-1; compare the optimal value of each particle to its current objective function and update p bi And p g Loop until iteration is complete, p g Is the result of the final selected wavelength.
Further, the LIBS emission line selection module specifically performs SNV preprocessing on the acquired LIBS data to remove irrelevant noise and background, and then uses a plurality of atomic emission lines as LIBS emission lines of some element; and taking a plurality of spectral data points near each LIBS emission line wave band, and fusing the plurality of wave bands to form the LIBS emission line of the element.
Furthermore, a chemometrics algorithm is adopted in the analysis model building module to build an element quantitative analysis model.
Further, the chemometric algorithm adopts a least square method to construct a quantitative analysis model of the elements.
The invention has the advantages that:
(1) the method has the advantages of no need of sample pretreatment in the detection process, simple and quick detection, green and safety, and capability of realizing the improvement of the LIBS detection sample element content precision.
(2) On the basis of quantitative analysis of the LIBS emission spectrum of the elements, the invention further selects the characteristics of the emission spectrum by using a waveband selection algorithm, removes the influence of spectral line self-absorption and interference, improves the precision of the quantitative analysis and is applied to the accurate detection research of the elements of various samples.
(3) The invention provides a method for accurately detecting the content of elements in a sample by a LIBS (laser induced breakdown spectroscopy) combined with a variable-dimension particle swarm optimization-combined moving window wavelength selection algorithm. Meanwhile, an algorithm provides a combined moving window strategy, and the spectral data variable interval is quickly selected through the combined moving window.
Drawings
FIG. 1 is a flowchart of a method for improving the detection accuracy of laser-induced breakdown spectroscopy according to an embodiment of the present invention;
fig. 2 is a diagram of the content of sodium and iron elements in a sorghum root sample, which is a sample material of the method for improving the detection accuracy of laser-induced breakdown spectroscopy disclosed in the embodiment of the present invention;
FIG. 3 shows LIBS emission spectrum lines of sodium and bands selected by a variable dimension particle swarm optimization-combined moving window algorithm according to the method for improving detection accuracy of laser induced breakdown spectroscopy disclosed in the embodiment of the present invention;
FIG. 4 shows an LIBS emission spectrum line of iron element and a band selected by a variable dimension particle swarm optimization-combined moving window algorithm according to the method for improving the detection precision of laser induced breakdown spectroscopy disclosed in the embodiment of the present invention;
FIG. 5 shows the results of a PLS quantitative analysis model established by screening and unseen LIBS spectral lines of sodium elements based on a variable-dimension particle swarm optimization-combined moving window algorithm according to the method for improving the detection precision of laser-induced breakdown spectroscopy disclosed by the embodiment of the invention;
fig. 6 shows the results of a PLS quantitative analysis model established by screening and unseen LIBS spectral lines of iron elements based on a variable-dimension particle swarm optimization-combined moving window algorithm in the method for improving the detection accuracy of laser-induced breakdown spectroscopy disclosed in the embodiment of the present invention.
Detailed Description
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 embodiments of the present invention, and it is obvious that the described embodiments are some embodiments of the present invention, but not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The method for improving the detection precision of the laser-induced breakdown spectroscopy in the embodiment includes the following steps, as shown in fig. 1:
the method comprises the following steps: collecting samples with element content in gradient distribution as a modeling sample set, and detecting the element content of each sample by using ICP (inductively coupled plasma);
step two: collecting LIBS data of a sample;
step three: selecting an LIBS emission line of an analysis element;
step four: selecting an element emission LIBS spectral line wave band by using a variable dimension particle swarm optimization-combined moving window algorithm;
step five: and establishing an element quantitative analysis model by using the characteristic wave bands screened by the algorithm.
The first step, the second step, the third step, the fourth step and the fifth step are implemented on the basis of a LIBS acquisition system and a variable dimension particle swarm optimization-combined moving window algorithm capable of screening wavelengths. ICP is used for detecting the content of elements in the sample, so that the content can be conveniently compared with the model, and the quality of the model can be evaluated; collecting LIBS data of a sample, and preprocessing the spectrum in order to remove influences of independent variables and background noise in the spectrum; selecting an atomic or ionic emission line for the element being analyzed as the LIBS emission line for that element; screening the emission spectral lines by using a variable-dimension particle swarm optimization-combined moving window algorithm to select a characteristic wave band; and establishing an element quantitative analysis model for the screened characteristic wave bands.
In the step, an ICP method is used for carrying out quantitative detection on elements to be analyzed in the sample, and the sample with element concentration gradient is prepared for experiment, so that the accuracy of the model is conveniently verified. And step two, acquiring LIBS data of the sample by using an LIBS device. And step three, selecting the LIBS emission spectral line of the analysis element. Before analyzing the LIBS data of the sample, the influence of irrelevant variables and background noise needs to be removed through pretreatment. Each element emits an atomic or ionic emission line with a specific wavelength for laser ablation, which represents the identity of the element, namely the LIBS emission line of the element. The wavelength of the spectral line is used to distinguish the kind of the element, and the intensity of the spectral line is used to indicate the relative content of the element. The LIBS emission line of the analytical element is selected for accurate detection of the element.
And step four, performing band screening on the LIBS emission spectral line of the element by using a variable-dimension particle swarm optimization-combined moving window algorithm. The particles in this example are the locations (wavelengths) of the spectrum. The window refers to the size of the spectral interval. The population is the collection of spectral positions, and the particle swarm is the emission spectral band needing to be screened. For the algorithm, the spectrum and the chemical value are used as the input of the algorithm, and the optimal spectrum interval position is output through algorithm calculation.
In the four-dimension variable particle swarm optimization-combined moving window algorithm, a combined moving window can automatically select proper interval quantity and width, the main principle is that a combined moving window method divides a spectrum range into N windows with equal width, each window can slide and move in the whole spectrum range, and different windows can be mutually overlapped. And performing performance evaluation on the selected windows by taking a cross validation Root Mean Square Error (RMSECV) as an objective function, and updating the center position of each window through an intelligent optimization algorithm. This allows to obtain an optimal combination of window intervals and to select the variables. And the variable dimension particle swarm optimization algorithm seeks an optimal solution through cooperation and information sharing among individuals in a swarm.
In the basic particle swarm algorithm, each particle in the swarm of particles represents a possible solution to the problem, and each particle has N dimensions (N windows) representing the positions of different windows in the spectral range. Let x i Represents the position of particle i:
x i ＝{w i,1 ,w i,2 ,...w i,j ...w i,N } (1)
w i,j indicating the position of the jth window for particle i. The method of uniform distribution is used for random initialization so that the particles cover the search space as uniformly as possible.
Iteratively changing the velocity v under the guidance of the global optimal solution and the local optimal solution i And position x i The equation is as follows:
v i ＝w×v i '+c 1 ×rand()×(p bi -x i ')+c 2 ×rand()×(p g x i ') (2)
x i ＝x i '+v i (3)
wherein x is i ' position of i-th particle before update, v i ' represents the velocity of the i-th particle before the update; p is a radical of bi For the optimal position of the entire population, w is the inertial weight, c 1 And c 2 To accelerate the coefficients, a rand () function is used to generate random numbers that are evenly divided within the interval 0-1.
The process of the dimension-variable particle swarm is basically the same as that of the classical particle swarm, and two important differences exist in the aspects of initialization of the particle swarm and dimension change of the particles:
(1) each particle in a population of particles has the same dimensions in classical particle swarm algorithms, while in variable-dimension particle swarm algorithms the dimensions of each particle may be different and have a certain distribution. In this work, the particle dimensions are distributed layer by layer. D i Is the dimension of the particle number i, calculated by:
D i ＝min(D max ,fix((i-1)/(M/D max ))+1) (4)
here, fix () is an integer that rounds an argument to the nearest zero. M is a particleNumber of particles of the population, D max Is the maximum dimension of the window.
(2) The particle dimension changes with a certain probability, the probability of the particle dimension change is lower in the initial stage of algorithm iteration, each particle keeps the dimension of the particle unchanged, and the particle can find the optimal value in the dimension of the particle in the mode. And simultaneously sharing the optimal position information with the particles according to the global optimal value of the particle swarm. With the increase of the iteration times, the probability of changing the dimension of the particle is increased, the dimension of the particle is gradually changed into the optimal dimension, and further searching is continued in the optimal dimension until the optimal solution is found. The particles have different dimensions and when using a formula it must be ensured that the particle position, velocity and global optimum position have the same dimensions.
The variable dimension particle swarm algorithm comprises the following specific steps:
(1) for M particles, the size of the particles is initialized using equation (4). X of the particles i i A uniformly distributed random initialization is used. V of the particles i i Has the same dimension as its location and is filled with zeros.
(2) The objective function value of each particle is evaluated. Setting p bi For the current particle position, p g Is the particle position with the best objective function.
(3) The dimensions of each particle are updated. The position and velocity vectors for each particle are updated using equation (2). Compare the optimal value of each particle to its current objective function and update p bi Update p g With an optimal objective function. Looping until the iteration is complete, p g Is the result of the wavelength ultimately selected.
And finally, adopting a chemometrics algorithm to the characteristic wave bands screened by the variable-dimension particle swarm optimization-combined moving window algorithm to establish an element quantitative analysis model. A common chemometric algorithm is the partial least squares algorithm.
The embodiment also provides a system for improving laser-induced breakdown spectroscopy detection corresponding to the method, including:
the sample selection module is used for selecting the element content of the ICP detection sample as a sample;
the sample data acquisition module is used for acquiring LIBS data of the sample;
the LIBS emission line selection module is used for selecting the LIBS emission line of the analysis element;
the LIBS emission spectral line band selection module is used for selecting the LIBS emission spectral line band of the element by adopting a variable-dimension particle swarm optimization-combined moving window algorithm; the LIBS emission spectrum line wavelength is a particle, the size of an LIBS emission spectrum line interval is a window, the collection of the LIBS emission spectrum line wavelength is a group, the LIBS emission spectrum line wave band needing to be screened is a particle swarm, the LIBS emission spectrum line and a chemical value are used as the input of an algorithm, and the optimal spectrum interval position is output through calculation of a variable-dimension particle swarm optimization-combined moving window algorithm;
and the analysis model establishing module is used for establishing an element quantitative analysis model by utilizing the screened characteristic wave bands.
The specific implementation process of the above modules is shown in the steps from one to five.
The present invention is described in detail below with reference to examples of sorghum root samples. As shown in fig. 1, a method for improving LIBS detection accuracy includes the following steps:
s1: the content of the elements in the sample was measured by ICP. For quantitative detection of elements in a sample, the content of the elements needs to be accurately measured, and the prediction result of the model is compared with an accurate value, so that the quality of the LIBS quantitative model is judged.
As shown in fig. 2, the contents of the metallic elements sodium and iron in the sorghum root sample have a certain gradient range, which is substantially within the detection limit of LIBS, and can be quantitatively detected by LIBS. The samples selected were sorghum samples of three varieties "623B", "BJ 602", "BJ 603", all provided by the recent physical research institute of the chinese academy of sciences. The seeds are respectively planted and cultivated in the field under the stress concentration of four soil sodium salt with the concentration of 9-2.2g/kg (0, 0.2%, 0.4% and 0.6%), and 27 sorghum single plants are harvested after the seeds are matured. Cutting the root of each sample, cleaning, drying, grinding, sieving with a 100-mesh sieve, and drying in a drying oven at 60 deg.C for 12h to constant weight. Taking 0.1g of sample, and determining the content of Na and Fe by an inductive coupling plasma method (ICP) according to the national standard method for food safety (GB 5009.91-2017).
S2: LIBS data was collected from the samples. Weighing 1g of the sorghum root powder sample dried to constant weight, and pressing the sample into a round cake with the radius of 10mm and the thickness of 4mm by using a press with the pressure of 12 Pa. And (3) acquiring LIBS data of the sorghum sample by utilizing a DP-LIBS experimental system independently built in a laboratory. The experimental parameters were respectively: YAG laser energy is set to be 50mJ, repetition frequency is 1Hz, 1064nm laser is triggered firstly, 532nm laser is triggered later, and the time interval of two laser pulses is 0.4 mus; the spectrometer has a collection gate width of 1ms and a collection delay of 2 mus. The sample is placed on a three-dimensional platform, the platform is moved by 2mm amplitude to collect sample spectra at different positions once each sample is ablated by laser, 30 spectra are collected for each sample, and then the 30 spectra are averaged to be used as the sample spectrum.
S3: LIBS emission lines of the analytical elements were selected. SNV pretreatment is carried out on the collected LIBS data, and irrelevant noise and background are removed. The emission spectral line wave bands of the sodium element and the iron element are selected, and a plurality of atomic emission spectral lines are used as LIBS emission spectral lines of the elements. The LIBS atomic emission spectrum line of sodium element is Na I588.995 nm, Na I589.592 nm, Na I818.326 nm and Na I819.482 nm. And taking three spectral data points in front of and behind the several wave bands, and fusing the several wave bands to be used as an LIBS emission line of the Na element. The LIBS atomic emission spectrum line of the iron element is Fe I248.327 nm, Fe I248.814 nm, Fe I358.119 nm, Fe I373.486 nm, Fe I374.556 nm, Fe I382.043 nm, Fe I385.991 nm and Fe I404.581 nm. And (4) taking three spectral data points in front of and behind the bands, fusing the bands to serve as an LIBS emission line of the Fe element.
S4: and selecting the wave band of the LIBS spectral line wave band of the emission of the element by using a variable-dimension particle swarm optimization-combined moving window algorithm. The variable dimension particle swarm optimization-combined moving window algorithm is characterized in that the wave bands for screening sodium elements are concentrated near Na I818.326 nm and Na I819.482 nm, and the wave bands for screening iron elements are concentrated near Fe I248.327 nm, Fe I358.119 nm, Fe I373.486 nm, Fe I382.043 nm and Fe I385.991 nm.
As shown in FIG. 3, for emission lines of Na I588.995 nm, Na I589.592 nm, Na I818.326 nm and Na I819.482 nm of the sodium element, the wave bands selected by the dimension-changing particle swarm optimization-combined moving window algorithm are concentrated near Na I818.326 nm and Na I819.482 nm. Because two spectral lines of Na I588.995 nm and Na I589.592 nm are two resonance lines, the spectral line intensity is high, but the self-absorption phenomenon is easily generated, and the influence of spectral line interference is also generated. The wave bands selected by the variable-dimension particle swarm optimization-combined moving window algorithm can be effectively removed from the wave bands generated by self-absorption, and the interference of the wave bands on the result is avoided.
As shown in FIG. 4, for the emission lines of Fe element, Fe I248.327 nm, Fe I248.814 nm, Fe I358.119 nm, Fe I373.486 nm, Fe I374.556 nm, Fe I382.043 nm, Fe I385.991 nm and Fe I404.581 nm, the wave bands selected by the variable dimension particle swarm optimization-combined moving window algorithm are concentrated near Fe I248.327 nm, Fe I358.119 nm, Fe I373.486 nm, Fe I382.043 nm and Fe I385.991 nm. The interference phenomenon of the emission lines Fe I248.814 nm and Fe I404.518 nm of Fe can be found by combining a spectrum database with an LIBS spectrum, and due to the influence of the resolution of a spectrometer, Fe I374.556 nm is influenced by the spectrums of Fe I374.547 nm and Fe I374.589 nm, and is also interfered by the spectrum of Mn I374.661 nm. Therefore, the variable-dimension particle swarm optimization-combined moving window algorithm can select the emission spectral line of the iron element which is less interfered, and the influence of spectral line interference on quantitative analysis is weakened.
S5: and (3) establishing an element quantitative analysis model by using the characteristic wave band screened by the dimension-variable particle swarm optimization-combined moving window algorithm, namely establishing a quantitative analysis model of sodium element and iron element after using the algorithm to screen the characteristic wave band. And establishing a quantitative analysis model for the characteristic wave bands of the sodium element and the iron element selected by the variable-dimension particle swarm optimization-combined moving window algorithm by adopting a chemometrics algorithm. The chemometrics used here were Partial Least Squares (PLS) to model quantitative analysis of sodium and iron elements. Determination of coefficients of Cross Validation (R) is used 2 CV ) To and fromCross Validation Root Mean Square Error (RMSECV), Prediction Coefficient of decision (R) 2 p ) And Root Mean Square Error of Prediction (RMSEP) as an evaluation index for evaluating the model correction set and the Prediction set.
As shown in fig. 5, the graphs (a) and (b) are the results of the correction set and the prediction set of the PLS model constructed based on the characteristic bands of sodium element screened by the variable-dimension particle swarm optimization-combined moving window algorithm; panels (c) and (d) are the calibration and test set results for the PLS model constructed without screening for sodium emission lines. It can be seen that the model parameter R before variable selection of the dimension-variable particle swarm optimization-combined moving window algorithm 2 CV 、RMSECV、R 2 P RMSEP is respectively 0.903, 2.025, 0.846 and 3.807; model parameter R screened by dimension-variable particle swarm optimization-combined moving window algorithm 2 CV 、RMSECV、R 2 P RMSEP was 0.962, 1.261, 0.988, 1.063, respectively. Compared with R of a control experiment, R of a model sodium element test result after optimization of variable-dimension particle swarm optimization and combined moving window algorithm 2 CV 6.53% of the increase, 37.73% of the decrease in RMSECV, R 2 P 16.78% of the RMSEP increase and 72.08% of the RMSEP decrease.
As shown in fig. 6, the graphs (a) and (b) are the results of the correction set and the prediction set of the PLS model constructed based on the iron element characteristic wave band screened by the variable-dimension particle swarm optimization-combined moving window algorithm; graphs (c) and (d) are the results of the calibration set and test set of the PLS model constructed without screening for iron emission lines. It can be seen that the model parameter R before variable selection of the dimension-variable particle swarm optimization-combined moving window algorithm 2 CV 、RMSECV、R 2 P RMSEP is 0.890, 8.098, 0.916, 8.850, respectively; after being screened by a dimension-variable particle swarm optimization-combined moving window algorithm, R of the model 2 CV 、RMSECV、R 2 P RMSEP was 0.956, 5.095, 0.955, 6.438, respectively. The prediction result of the Fe content of the model after the variable-dimension particle swarm optimization-combined moving window optimization is compared with the R of a comparison model 2 CV 7.42% of the increase, 37.08% of the decrease in RMSECV, and R 2 P The increase was 4.26% and the RMSEP was reduced by 27.25%.
Specifically, the first step, the second step, the third step, the fourth step and the fifth step are implemented on the basis of a LIBS spectrum acquisition system and a variable-dimension particle swarm optimization-combined moving window algorithm capable of screening wavelengths.
Through the technical scheme, the method for improving the LIBS detection precision provided by the invention does not need to pretreat the sample, is simple, quick, green and safe, and can realize accurate detection of the sample. The method can be applied to quantitative detection of the metal elements in the plants by detecting the metal elements in the sorghum samples, and provides a scheme for quantitative detection of the plant metal elements by LIBS. The method can detect the contents of sodium element and iron element in the sorghum sample, and can also detect different elements in other samples. Meanwhile, the method can be used in combination with other spectrum detection technologies, such as a near infrared spectrum or Raman spectrum combined molecular detection technology, and the molecular information is used for correcting the matrix effect generated when the LIBS detects the element information, so that the quantitative analysis of the element can realize a more accurate detection result, and the accurate and rapid detection of the content of the element in the sample can be realized.
The above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.

Claims (10)

1. A method for improving laser-induced breakdown spectroscopy detection is characterized by comprising the following steps:
s01, collecting samples with element content in gradient distribution as a modeling sample set, and detecting the element content of each sample by using ICP (inductively coupled plasma);
s02, collecting LIBS data of the sample;
s03, selecting an LIBS emission spectral line of an analysis element;
s04, selecting LIBS emission spectral line wave bands of elements by adopting a variable-dimension particle swarm optimization-combined moving window algorithm; the LIBS emission spectrum line wavelength is a particle, the size of an LIBS emission spectrum line interval is a window, the collection of the LIBS emission spectrum line wavelengths is a group, the LIBS emission spectrum line wave band needing to be screened is a particle swarm, the LIBS emission spectrum line and a chemical value are used as input of an algorithm, and the optimal spectrum interval position is output through calculation of a variable-dimension particle swarm optimization-combined moving window algorithm;
and S05, establishing an element quantitative analysis model by using the screened characteristic wave bands.
2. The method for improving the detection of the laser-induced breakdown spectroscopy according to claim 1, wherein the step S04 of selecting the LIBS emission spectral line band of the element by using the wiener swarm optimization-combined moving window algorithm comprises:
(1) for M particles, equation D is used i ＝min(D max ,fix((i-1)/(M/D max ) +1) the size of the initialization particles; wherein D is i Is the dimension of the particle i, the position x of the particle i i With uniformly distributed random initialization, the velocity v of the particles i i Having the same dimension as its position and filled with zeros, fix () is an integer rounding the argument to the nearest zero, M is the number of particles of the population, D max Is the maximum dimension of the window;
(2) evaluating the objective function value of each particle, setting p bi For the current particle position, p g Is the particle position with the best objective function;
(3) updating the dimension of each particle; using equation v i ＝w×v i +c 1 ×rand()×(p bi -x i )+c 2 ×rand()×(p g x i ) Updating the position and velocity vector of each particle; where w is the inertial weight, c 1 And c 2 For the acceleration factor, a rand () function is used to generate a random uniform division within the 0-1 intervalCounting; compare the optimal value of each particle to its current objective function and update p bi And p g Loop until iteration is complete, p g Is the result of the wavelength ultimately selected.
3. The method for improving the LIBS detection according to claim 1 or 2, wherein the step S03 specifically includes performing SNV preprocessing on the collected LIBS data to remove irrelevant noise and background, and then using a plurality of atomic emission spectral lines as LIBS emission spectral lines of a certain element; a number of spectral data points are taken near each LIBS emission band and the bands are fused to form the LIBS emission for the element.
4. The method for improving detection of laser-induced breakdown spectroscopy according to claim 1 or 2, wherein a chemometrics algorithm is used to build a quantitative analysis model of elements in step S05.
5. The method for improving LIBS detection as claimed in claim 4, wherein the chemometric algorithm is a least squares method for constructing a quantitative analysis model of the element.
6. A system for enhancing detection of laser induced breakdown spectroscopy, comprising:
the sample selection module is used for selecting the element content of the ICP detection sample as a sample;
the sample data acquisition module is used for acquiring LIBS data of the sample;
the LIBS emission line selection module is used for selecting the LIBS emission line of the analysis element;
the LIBS emission spectral line band selection module is used for selecting the LIBS emission spectral line band of the element by adopting a variable-dimension particle swarm optimization-combined moving window algorithm; the LIBS emission spectrum line wavelength is a particle, the size of an LIBS emission spectrum line interval is a window, the collection of the LIBS emission spectrum line wavelength is a group, the LIBS emission spectrum line wave band needing to be screened is a particle swarm, the LIBS emission spectrum line and a chemical value are used as the input of an algorithm, and the optimal spectrum interval position is output through calculation of a variable-dimension particle swarm optimization-combined moving window algorithm;
and the analysis model establishing module is used for establishing an element quantitative analysis model by utilizing the screened characteristic wave bands.
7. The system for improving LIBS emission spectrum detection according to claim 6, wherein the LIBS emission spectrum band selection module adopts a particle-swarm optimization-combined moving window algorithm to select the LIBS emission spectrum band of the element according to a specific process:
(1) for M particles, equation D is used i ＝min(D max ,fix((i-1)/(M/D max ) +1) the size of the initialization particles; wherein D is i Is the dimension of the particle i, the position x of the particle i i With uniformly distributed random initialization, the velocity v of the particle i i Having the same dimension as its position and filled with zeros, fix () is an integer rounding the argument to the nearest zero, M is the number of particles of the population, D max Is the maximum dimension of the window;
(2) evaluating the value of the objective function for each particle, setting p bi For the current particle position, p g Is the particle position with the best objective function;
(3) updating the dimension of each particle; using equation v i ＝w×v i +c 1 ×rand()×(p bi -x i )+c 2 ×rand()×(p g x i ) Updating the position and velocity vector of each particle; where w is the inertial weight, c 1 And c 2 For the acceleration coefficient, a rand () function is used for generating random numbers which are uniformly divided in an interval of 0-1; compare the optimal value of each particle to its current objective function and update p bi And p g Loop until iteration is complete, p g Is the result of the wavelength ultimately selected.
8. The system for improving LIBS emission spectrum detection according to claim 6 or 7, wherein the LIBS emission spectrum selection module specifically performs SNV pretreatment on the collected LIBS data to remove irrelevant noise and background, and then uses a plurality of atomic emission spectra as the LIBS emission spectrum of an element; a number of spectral data points are taken near each LIBS emission band and the bands are fused to form the LIBS emission for the element.
9. The system for improving LIBS detection according to claim 6 or 7, wherein the analytical model building module builds an elemental quantitative analytical model by using a chemometric algorithm.
10. The system for improving detection of laser-induced breakdown spectroscopy of claim 9, wherein the chemometric algorithm uses a least squares method to construct a quantitative analysis model of the elements.
CN202210434868.7A 2022-04-24 2022-04-24 Method and system for improving detection precision of laser-induced breakdown spectroscopy Pending CN114936513A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210434868.7A CN114936513A (en) 2022-04-24 2022-04-24 Method and system for improving detection precision of laser-induced breakdown spectroscopy

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210434868.7A CN114936513A (en) 2022-04-24 2022-04-24 Method and system for improving detection precision of laser-induced breakdown spectroscopy

Publications (1)

Publication Number Publication Date
CN114936513A true CN114936513A (en) 2022-08-23

Family

ID=82861541

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210434868.7A Pending CN114936513A (en) 2022-04-24 2022-04-24 Method and system for improving detection precision of laser-induced breakdown spectroscopy

Country Status (1)

Country Link
CN (1) CN114936513A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117949436A (en) * 2024-03-26 2024-04-30 宝鸡核力材料科技有限公司 Metal element component detection method and system applied to titanium alloy smelting
CN118190914A (en) * 2024-05-16 2024-06-14 长春工业大学 Method for detecting component content of additive manufacturing part

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117949436A (en) * 2024-03-26 2024-04-30 宝鸡核力材料科技有限公司 Metal element component detection method and system applied to titanium alloy smelting
CN118190914A (en) * 2024-05-16 2024-06-14 长春工业大学 Method for detecting component content of additive manufacturing part

Similar Documents

Publication Publication Date Title
Zhang et al. A novel approach for the quantitative analysis of multiple elements in steel based on laser-induced breakdown spectroscopy (LIBS) and random forest regression (RFR)
CN114936513A (en) Method and system for improving detection precision of laser-induced breakdown spectroscopy
CN104374752B (en) Rapid detection method for nutrient elements of crops based on collinear laser-induced breakdown spectroscopy
Sanaeifar et al. A data fusion approach on confocal Raman microspectroscopy and electronic nose for quantitative evaluation of pesticide residue in tea
CN103487411B (en) A kind of random forests algorithm is in conjunction with the method for Laser-induced Breakdown Spectroscopy identification type of steels
CN102262076B (en) Method for laser-induced breakdown spectroscopy element concentration determination based on spectral line combination
CN105158200B (en) A kind of modeling method for improving the Qualitative Analysis of Near Infrared Spectroscopy degree of accuracy
Dyar et al. Comparison of baseline removal methods for laser-induced breakdown spectroscopy of geological samples
CN113155809B (en) Novel spectral detection method for ore classification and real-time quantitative analysis
CN105008898A (en) System for and method of combined LIBS and IR absorption spectroscopy investigations
CN107037012B (en) Echelle spectrometer dynamic correcting method for laser induced breakdown spectroscopy acquisition
Dai et al. Quantitative determination of Al–Cu–Mg–Fe–Ni aluminum alloy using laser-induced breakdown spectroscopy combined with LASSO–LSSVM regression
CN107515202A (en) Terahertz light spectral analysis method, system and equipment
Chen et al. Quantitative analysis of soil nutrition based on FT-NIR spectroscopy integrated with BP neural deep learning
CN102830096A (en) Method for measuring element concentration and correcting error based on artificial neural network
CN112834485B (en) Non-calibration method for quantitative analysis of laser-induced breakdown spectroscopy elements
CN105717066A (en) Near-infrared spectrum recognition model based on weighting association coefficients
Chen et al. Quantitative analysis of chromium in pork by PSO-SVM chemometrics based on laser induced breakdown spectroscopy
CN111239072B (en) Method for accurately measuring temperature of combustion gas
CN108956584A (en) The quick and precisely detection method of heavy metal element chromium in a kind of mulberry fruit
Wei et al. Confocal Raman microspectroscopy combined with spectral screening algorithms for quantitative analysis of starch in rice
CN111272735B (en) Detection method of laser-induced breakdown spectroscopy
Zhang et al. Application of swarm intelligence algorithms to the characteristic wavelength selection of soil moisture content
Mubarok et al. Non-destructive method for predicting sapodilla fruit quality using near infrared spectroscopy
CN114486160B (en) Thermochemical unbalanced identification analysis method for high enthalpy flow field

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