CN104730041A - Method and apparatus for improving plastic identification precision of laser probe - Google Patents
Method and apparatus for improving plastic identification precision of laser probe Download PDFInfo
- Publication number
- CN104730041A CN104730041A CN201310710790.8A CN201310710790A CN104730041A CN 104730041 A CN104730041 A CN 104730041A CN 201310710790 A CN201310710790 A CN 201310710790A CN 104730041 A CN104730041 A CN 104730041A
- Authority
- CN
- China
- Prior art keywords
- intensity
- characteristic spectral
- line
- plastics
- spectral line
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Abstract
The invention discloses a method for improving the plastic identification precision of a laser probe. According to the method, the plasma spectrum of each plastic to be detected is acquired by using a laser probe device, then characteristic spectral lines are selected, and intensities of the characteristic spectral line are read and normalized; then the normalized intensities of three key characteristic spectral lines in M2 spectra of plastics to be identified, i.e., C-N (0,0), C-C (0,0) and O I777.41 nm, are separately multiplied by a weighting factor; and finally, an SVM classifier model obtained through training is employed, to-be-identified spectral data is used as input of the classifier model, and the identification precision of each plastic is obtained according to output. The method is simple to realize, effectively improves the classified weight of key characteristic spectral lines, enables contribution of more characteristic spectral lines to classification to be improved and eventually improves plastic identification and classification accuracy.
Description
Technical field
The invention belongs to laser accurate detection technique field, be specially a kind of laser probe plastic identification method and device thereof.
Background technology
Along with plastic products consumption figure constantly increases, waste plastic is also on the increase, and is Resource recovery and protection of the environment, and plastics classification becomes the problem needing solution badly.The electron microprobe examination device that Laser-induced Breakdown Spectroscopy (Laser-Induced breakdown Spectroscopy is called for short LIBS) technology and domestic scholars are familiar with has many similarities in principle, is therefore called " laser probe ".Laser probe plastic identification assorting process focuses the laser beam into plastic sample surface, inspire plasma, plasma emissioning light is collected through light collector and optical fiber and is coupled in spectrometer, utilizing emitted light implementation space after spectrometer of different wave length is separated and is recorded on the diverse location of ICCD target surface, final formation plasma spectrometry, includes many characteristic spectral lines in each plasma spectrometry.Due to kind, the content difference of characteristic element in different plastics, cause the intensity of individual features spectral line in its plasma spectrometry variant.Adopt the plasma spectrometry of standard plastic to carry out training classifier model, namely obtain the mapping relations of characteristic spectral line intensity in plasma spectrometry and plastics kind, utilize the discriminator that this sorter model can realize plastic refuse.The advantages such as this technology has that recognition speed is fast, sample does not need pre-service, the color that be not limited to plastics little to sample damage.
Research paper " the Laser-induced Breakdown Spectroscopy plastic identification Research on classifying method based on principal component analysis (PCA) and artificial neural network " (spectroscopy and spectral analysis, 32nd volume, 12nd phase) mention principal component analysis (PCA) (Principle Component Analysis, be called for short PCA) combine with laser probe technology with back-propagation artificial neural network (Back-propagation Artificial Neural Network) two kinds of algorithms, be 97.5% to the average accuracy of identification of 7 kinds of plastics.Research paper " Laser-induced plasmaspectroscopy for plastic identification " (Polymer Engineering and Science, November2000, Vol.40, No.11) mention linear correlation, rank correlation two kinds of statistical correlation algorithms to combine with laser probe technology 90% is reached to the average recognition accuracy of 6 kinds of plastics.
Although above experimental study achieves higher discriminator precision, its accuracy can't meet the demand of commercial Application, awaits further raising.
Summary of the invention
The invention provides a kind of method improving laser probe plastic identification precision, object is to improve the recognition accuracy of laser probe to plastic material further.
A kind of method improving laser probe plastic identification precision provided by the invention, it is characterized in that, the method comprises the steps:
1st step adopts the collection of laser probe device often to plant the plasma spectrometry of plastics to be measured, obtains M altogether
2individual plasma spectrometry;
2nd step selected characteristic spectral line, reading characteristic spectral line intensity, and normalized is done to the intensity of characteristic spectral line;
The M of the 3rd step plastics to be identified
2in individual spectrum, the normalized intensity of C-N (0,0), C-C (0,0) and O I777.41nm tri-key feature spectral lines is multiplied by f respectively
1, f
2, f
3three weight factors, thus obtain M
2group spectroscopic data, equally according to corresponding plastic type, sets a label for often organizing spectroscopic data;
4th step adopts the SVM classifier model of training and obtaining, by M to be identified
2group spectroscopic data is as the input of sorter model, and output is the prediction label often organized corresponding to spectroscopic data, the prediction label exported and physical tags is contrasted, thus obtains the recognition accuracy of often kind of plastics;
The laser probe device that described SVM classifier model uses when setting up, technological parameter are identical with the 1st step with ambiance.
The device realizing said method provided by the invention, is characterized in that, it comprises laser instrument, optical maser wavelength catoptron, condenser lens, electric platforms, argon gas gas blow pipe, light collector, optical fiber, spectrometer, ICCD, triggers line, data line, and data processor;
Laser instrument, optical maser wavelength catoptron and condenser lens are positioned in same light path successively; Electric platforms is positioned on the emitting light path of condenser lens, for placing plastics to be measured; Argon gas gas blow pipe is used for sending into argon gas to laser action district, and light collector is for collecting plasma spectrometry, and light collector is connected with spectrometer by optical fiber, and ICCD is connected with laser instrument by triggering line, and ICCD is connected with data processor by data line.
Because the method for identifying and classifying of existing laser probe to plastic material exists many disadvantages, method provided by the invention can improve plastic identification classify accuracy.Specifically, the inventive method has following characteristics and effect:
(1) implementation procedure of method is simple: classify for plastics, only need C-N (0 in all spectrum, 0), the normalized intensity of C-C (0,0) and O I777.41nm tri-key feature spectral lines is multiplied by three corresponding weight factors respectively.
(2) method that the present invention proposes effectively improves the classified weight of key feature spectral line, the more contribution rate of multiple features spectral line to classification is increased to some extent, finally improves plastic identification classify accuracy.
(3) the inventive method is the change to original spectral data, can combine with the sorting algorithm such as artificial neural network, support vector machine.
Accompanying drawing explanation
The structural representation of the laser probe plastic identification device that Fig. 1 provides for example of the present invention;
Wherein, 1. laser instrument; 2. optical maser wavelength catoptron; 3. condenser lens; 4. electric platforms; 5. argon gas gas blow pipe; 6. light collector; 7. optical fiber; 8. spectrometer; 9.ICCD; 10. trigger line; 13. data lines; 14. data processors.
Embodiment
Below in conjunction with specific embodiment, the specific embodiment of the present invention is described further.It should be noted that at this, the explanation for these embodiments understands the present invention for helping, but does not form limitation of the invention.In addition, if below in described each embodiment of the present invention involved technical characteristic do not form conflict each other and just can mutually combine.
The raising chosen for laser probe plastic identification classify accuracy of characteristic spectral line has material impact, and first, in the plasma spectrometry of not otherwise identical plastic, characteristic spectral line intensity needs to be proportional to its corresponding element content; Secondly, the intensity of characteristic spectral line can not be too weak, and characteristic spectral line intensity is stronger, and its weight in plastics are classified is just larger, otherwise then weight is less.Due to the main component that oxygen element and nitrogen element are multiple plastics, for not otherwise identical plastic, the content of two kinds of elements is different; In addition, aromatic series plastics and aliphatics plastics have larger difference in C-C key concentration.Therefore, C-N (0,0), C-C (0,0), O I777.41nm tri-key feature spectral lines are classified significant to plastic identification.
In employing ar gas environment elimination air, the element spectral line such as nitrogen, oxygen is to the interference of plastics plasma spectrometry, make the not difference of otherwise identical plastic on nitrogen, oxygen element content can by its plasma spectrometry in C-N (0,0), O I777.41nm two characteristic spectral lines intensity reflected.For the intensity of C-C in plastics plasma spectrometry (0,0) molecular spectrum, aromatic series plastics are greater than aliphatics plastics.But relative to the intensity of metallic element spectral line and protium spectral line, the intensity of O I777.41nm, C-N (0,0), C-C (0,0) three characteristic spectral lines is more weak, the weight in plastics classification is less.In order to increase the classified weight of above three spectral lines, improve plastic identification classify accuracy, the solution that the inventive method provides comprises the steps:
(1) SVM classifier model is trained:
(1.1) plasma spectrometry of S kind plastics is gathered.To obtain optimal spectrum signal to background ratio for standard, Optimizing Process Parameters, comprising: pulsed laser energy, defocusing amount (laser beam focus focus is relative to the distance of sample surfaces), ICCD time delay and gate-width.Adopt the mode at plasma side top blast argon gas, obtain ar gas environment, remove air to the impact of plastics plasma.Keep same process parameter and ar gas environment, gather the plasma spectrometry of various plastics, often kind of plastics gather N number of plasma spectrometry, and S kind plastics gather M altogether
1=S × N number of spectrum (value of S is plastics kind quantity common in life, general S>=11,50≤N≤150).
(1.2) selected characteristic spectral line.Qualitative analysis is carried out to the plasma spectrometry of often kind of plastics, determines the 12 kinds of elements mainly contained in plastic material.Zoom in self-absorption effect, overlapping few be criterion, select 15 spectral lines corresponding to 12 kinds of elements as characteristic spectral line, be respectively: C I247.86nm, H I656.3nm, Mg II279.55nm, F I685.7nm, ClI725.7nm, Ti II334.94nm, F I739.9nm, C-N (0,0), N I746.9nm, Ca II393.34nm, K I766.5nm, C
2(0,0), OI777.3nm, Na I589.06nm, Cl I837.59nm.
(1.3) characteristic spectral line intensity is read.Wavelength centered by standard line wavelength in America NI ST ground atom spectra database, read out ± 0.1nm within the scope of the maximum intensity of (considering that temperature fluctuation causes line wavelength minor shifts) as line strength.M altogether
1individual spectrum, reads the intensity of 15 characteristic spectral lines in each spectrum successively.
(1.4) normalized of characteristic spectral line intensity.To each spectrum, the intensity of all 15 characteristic spectral lines all divided by C I247.86nm line strength in same spectrum, thus converts the actual strength of characteristic spectral line to normalized intensity, thus obtains M
1group spectroscopic data.Often organize the plastic type belonging to 15 these spectrum of normalized intensity value unique identification in spectroscopic data.
(1.5) acquisition of weight factor.Obtain 15 average normalized intensity level: I that 15 characteristic spectral lines correspond to
cI, I
hI, I
mgII..., I
clI, wherein each average normalized intensity level is M
1the mean value of individual spectrum.To 6 average normalized intensity level (I corresponding to 5 strip metal element spectral lines easy excitated in characteristic spectral line and protium spectral line
mgII, I
tiII, I
naI, I
caII, I
kI, I
hI) be averaged again, obtain the overall average normalized intensity value I of easy excitated spectral line
tot:
By I
totrespectively divided by C-N (0,0), C-C (0,0) and the average normalized intensity level corresponding to O I777.41nm tri-key feature spectral lines: I
c-N, I
c-C, I
oI, obtain three weight factor: f
1, f
2, f
3.
(1.6) C-N (0,0) is increased, the classified weight of C-C (0,0) and O I777.41nm tri-key feature spectral lines.All M
1in group spectroscopic data, the normalized intensity value of C-N (0,0), C-C (0,0) and O I777.41nm tri-key feature spectral lines is multiplied by f respectively
1, f
2, f
3three factors, make the classified weight of three spectral lines be increased to the level of metallic element spectral line and hydrogen spectral line.
(1.7) support vector machine (SVM) algorithm data process.Under Matlab software environment, run LIBSVM software toolkit, wherein the kernel function of SVM algorithm adopts radial basis function (RBF).Nuclear parameter g in the punishment parameter C of SVM algorithm and RBF kernel function, the two has material impact to the discriminator precision of SVM model, adopts genetic algorithm and cross verification to be optimized C, g parameter respectively.M
1group spectroscopic data (often organize spectroscopic data and comprise 15 normalized intensity values, corresponding to 15 characteristic spectral lines), according to the plastic type often organized corresponding to spectroscopic data, is its setting label.By M
1the label of group spectroscopic data and correspondence thereof, as the input of SVM algorithm, for training SVM, sets up SVM classifier model.
(2) SVM classifier model is to the discriminator of identical type plastic products
(2.1) for other one group of plastics to be identified, adopt identical laser probe device, under identical technological parameter and ambiance, gather the plasma spectrometry of often kind of plastics, obtain M altogether
2individual plasma spectrometry.
(2.2) according to mode selected characteristic spectral line identical in step (1.2) to (1.4), reading characteristic spectral line intensity, and normalized is done to the intensity of characteristic spectral line.
(2.3) M of plastics to be identified
2in individual spectrum, the normalized intensity of C-N (0,0), C-C (0,0) and O I777.41nm tri-key feature spectral lines is multiplied by f respectively
1, f
2, f
3three weight factors, thus obtain M
2group spectroscopic data (often organize spectroscopic data comprise 15 normalized intensities, corresponding to 15 characteristic spectral lines), equally according to corresponding plastic type, sets a label for often organizing spectroscopic data, label set-up mode with above (1.7) identical.Adopt the SVM classifier model of training and obtaining, by M to be identified
2group spectroscopic data (not comprising label) is as the input of disaggregated model, and output is the prediction label often organized corresponding to spectroscopic data.The prediction label exported and physical tags are contrasted, thus obtains the recognition accuracy of often kind of plastics.
As shown in Figure 1, it comprises laser instrument 1, optical maser wavelength catoptron 2 to the device realizing said method provided by the invention, condenser lens 3, electric platforms 4, argon gas gas blow pipe 5, light collector 6, optical fiber 7, spectrometer 8, ICCD9, triggers line 10, data line 11, and data processor 12.
Laser instrument 1, optical maser wavelength catoptron 2 and condenser lens 3 are positioned in same light path successively, electric platforms 4 is positioned on the emitting light path of condenser lens 3, argon gas gas blow pipe 5 is for sending into argon gas to laser action district, light collector 6 is for collecting plasma spectrometry, light collector 6 is connected with spectrometer 8 by optical fiber 7, ICCD is connected with laser instrument 1 by triggering line 10, and ICCD is connected with data processor 12 by data line 11.Laser instrument 1 exports pulse laser beam (wavelength 532nm), and laser beam, after catoptron 2 reflects, vertically focuses on plastic sample on the surface through after condenser lens 3 (focal distance f=150mm), inspires plasma.Plasma emissioning light is collected by light collector 6 and is coupled in optical fiber 7 (diameter 50 μm, long 2m), in optical fiber 7 transmission also lead-in light spectrometer 8.While laser instrument 1 outgoing laser beam, also pulse triggering signal is sent to ICCD, trigger the electronic switch of ICCD, plasma spectrometry is recorded by ICCD after spectrometer diffraction, light splitting, and the spectral information of each element is presented on data processor 12 after software process.Sample is placed on two-dimentional electric platforms 4, and in spectra collection process, keeps sample to do " bow " zigzag motion by its drive.
Example:
Based on support vector machine (SVM) algorithm, laser probe technology is adopted to carry out discriminator to 11 kinds of plastics.As shown in Figure 1, the laser instrument of employing is Q-switch Nd:YAG pulsed laser (optical maser wavelength 532nm, repetition frequency 10Hz, pulsewidth 5ns) to device.Laser beam focuses on plastic sample surface after catoptron and plano-convex lens (focal distance f=150mm).Institute's activated plasma utilizing emitted light is collected by light collector and is coupled in the optical fiber of diameter 50 μm, long 2m, through Optical Fiber Transmission to spectrometer (Andor ME5000, be equipped with Andor DH334T ICCD, wavelength coverage 230 ~ 850nm, resolution lambda/Δ λ=5000) in carry out spectrum analysis.Sample is placed on the two-dimentional electric platforms of conputer controlled, keeps sample to do " bow " zigzag motion in spectra collection process, guarantees that laser pulse is all applied to a reposition of sample surfaces.Be blown into the flow velocity of 15L/min the impact that argon gas removes air plasma in plasma side.Choose 11 kinds of plastic standard product for setting up svm classifier model: teflon (PTFE), polystyrene (PS), polycarbonate (PC), polyurethane (PU), engineering plastics (ABS), Polyvinylchloride (PVC), organic glass (PMMA), nylon-6 (PA-6), polypropylene (PP), tygon (PE) and polyoxymethylene (POM).Table 1 lists molecular formula and the color of above-mentioned 11 kinds of plastics.
1. set up SVM classifier model:
(1.1) plastics plasma spectrometry is gathered.First, adopt standard mercury light source to calibrate spectrometer, after calibration, wavelength error is less than 0.05nm (the standard line wavelength relative in America NI ST ground atom spectra database).Secondly, to obtain optimal spectrum signal to background ratio for standard, technological parameter is optimized: be set to 49mJ after laser single-pulse energy optimization (when laser energy is 49mJ, the relative standard deviation of adding up 300 pulsed laser energies acquisitions is 1.977), consider that catoptron and condenser lens are to the reflection of laser, the final laser energy arriving sample surfaces is 44mJ.Laser beam foucing is positioned at 1mm place below sample surfaces, and rhegmalypt diameter is about 100 μm, and energy density is about 140J/cm
2.ICCD gate-width after optimization and time delay are respectively 1 μ s and 1.1 μ s.Again, under keeping technological parameter and the strictly constant condition of ar gas environment, often kind of plastics gather 50 spectrum, and 11 kinds of plastics obtain 550 spectrum altogether.
(1.2) the choosing of characteristic spectral line.Through the qualitative analysis to plastics plasma spectrometry, main containing 11 kinds of elements in 11 kinds of plastics, be C, H, O, N, Mg, Ti, Ca, K, Na, Cl and F respectively.Zoom in self-absorption effect, overlapping few be criterion, select 13 spectral lines corresponding to these elements and other 2 molecular band as characteristic spectral line (shown in table 2).
(1.3) characteristic spectral line intensity is read.Wavelength centered by standard line wavelength in America NI ST ground atom spectra database, read out ± 0.1nm within the scope of the maximum intensity of (considering that temperature fluctuation causes line wavelength minor shifts) as line strength.Totally 550 spectrum, read out the intensity of 15 characteristic spectral lines in each spectrum, the intensity of all 15 characteristic spectral lines does normalized divided by C I247.86nm line strength in same spectrum, thus obtain the 550 groups of spectroscopic datas corresponding to 11 kinds of plastics, often organize spectroscopic data (corresponding to a spectrum) to comprise 14 normalized intensities (the normalized intensity value of C I247.86nm spectral line is 1, can remove), the plastics kind belonging to this group 14 these spectrum of normalized intensity unique identification.
(1.4) weight factor obtains.Obtain 14 average normalized intensity levels (see table 3) that 15 characteristic spectral lines correspond to, wherein every average normalized intensity level corresponding to bar characteristic spectral line is the mean value of 550 spectrum.To 6 average normalized intensity level (I corresponding to 5 strip metal element spectral lines and protium spectral line
mgII, I
tiII, I
naI, I
caII, I
kI, I
hI) be averaged again, obtain the overall average normalized intensity value I of easy excitated spectral line
tot:
By I
totrespectively divided by C-N (0,0), C-C (0,0) and the average normalized intensity level corresponding to O I777.41nm tri-spectral lines: I
c-N=0.4718, I
c-C=0.3715, I
oI=0.4716, obtain three weight factor: f
1=45.2, f
2=5.96, f
3=4.755.
(1.5) C-N (0,0) is increased, the classified weight of C-C (0,0) and O I777.41nm tri-key feature spectral lines.By C-N (0,0), C-C (0,0) in 550 groups of spectroscopic datas, the normalized intensity of O I777.41nm tri-spectral lines is multiplied by three factor f all respectively
1=45.2, f
2=5.96, f
3=4.755.As shown in table 4, A row and B row represent respectively, before and after the normalized intensity adjustment of spectral line, and each normalization characteristic spectral line classified weight that application principal component analysis (PCA) (PCA) obtains.Obviously, after adjustment, the classified weight of C-N (0,0), C-C (0,0) and O I777.41nm tri-key feature spectral lines increases to some extent.
(1.6) support vector machine (SVM) algorithm is adopted.Under Matlab software environment, run LIBSVM software toolkit, wherein the kernel function of SVM algorithm adopts radial basis function (RBF).Nuclear parameter g in the punishment parameter C of SVM algorithm and RBF kernel function, the two has material impact to the discriminator precision of SVM model, adopt genetic algorithm and cross verification to be optimized C, g parameter respectively, after optimizing, C, g parameter is respectively: 20.12,0.012.550 groups of spectroscopic datas, according to its corresponding plastic type, set a label for often organizing spectroscopic data.Using all spectroscopic datas and corresponding label thereof as the input of SVM, for training SVM classifier, set up SVM classifier model.
2.SVM sorter is to the discriminator of waste plastic
The SVM classifier model that plastic standard product are trained is used to the 11 kinds of plastic products identifying identical type, and this retable material products is completely from waste plastics in daily life.Under same process parameter and ambiance, often kind of plastic products gather 50 spectrum, read out the intensity of characteristic spectral line in each spectrum, and do normalized divided by C I247.86nm line strength.C-N (0,0), C-C (0,0) in each spectrum, the actual normalized intensity of O I777.41nm tri-spectral lines is multiplied by three factors 45.2,5.96 and 4.755 respectively.Based on 550 groups of spectroscopic datas after " intensity adjustment ", SVM classifier model is used for identify 11 kinds of plastic products.Table 5 shows the recognition accuracy of SVM model to 11 kinds of plastic products, and A row are the recognition accuracies of not using in this inventive method situation, and B row are the recognition accuracies of using in this inventive method situation.Obviously, by increasing C-N (0,0), C-C (0,0), the normalized intensity of O I777.41nm tri-spectral lines, the discriminator accuracy of 11 kinds of plastics all reaches 100%.Therefore the method can improve laser probe plastic identification classify accuracy by adjustment characteristic spectral line intensity.
More than show and describe ultimate principle of the present invention and principal character and advantage of the present invention.The scientific research personnel of the industry should understand; the present invention is not restricted to the described embodiments; what describe in above-described embodiment and instructions just illustrates principle of the present invention; without departing from the spirit and scope of the present invention; the present invention also has various changes and modifications, and these changes and improvements all fall in the claimed scope of the invention.Application claims protection domain is defined by appending claims and equivalent thereof.
Table 1
Table 2
Table 3
Table 4.
Table 5.
Claims (8)
1. improve a method for laser probe plastic identification precision, it is characterized in that, the method comprises the steps:
1st step adopts the collection of laser probe device often to plant the plasma spectrometry of plastics to be measured, obtains M altogether
2individual plasma spectrometry;
2nd step selected characteristic spectral line, reading characteristic spectral line intensity, and normalized is done to the intensity of characteristic spectral line;
The M of the 3rd step plastics to be identified
2in individual spectrum, the normalized intensity of C-N (0,0), C-C (0,0) and O I777.41nm tri-key feature spectral lines is multiplied by f respectively
1, f
2, f
3three weight factors, thus obtain M
2group spectroscopic data, equally according to corresponding plastic type, sets a label for often organizing spectroscopic data;
4th step adopts the SVM classifier model of training and obtaining, by M to be identified
2group spectroscopic data is as the input of sorter model, and output is the prediction label often organized corresponding to spectroscopic data, the prediction label exported and physical tags is contrasted, thus obtains the recognition accuracy of often kind of plastics;
The laser probe device that described SVM classifier model uses when setting up, technological parameter are identical with the 1st step with ambiance.
2. the method for raising laser probe plastic identification precision according to claim 1, it is characterized in that, in 2nd step, the process of described selected characteristic spectral line is: carry out qualitative analysis to the plasma spectrometry of often kind of plastics to be measured, determine the 12 kinds of elements mainly contained in plastic material, zoom in self-absorption effect, overlap is criterion less, select 15 spectral lines corresponding to 12 kinds of elements as characteristic spectral line, be respectively: C I247.86nm, H I656.3nm, Mg II279.55nm, F I685.7nm, Cl I725.7nm, Ti II334.94nm, F I739.9nm, C-N (0, 0), N I746.9nm, Ca II393.34nm, K I766.5nm, C
2(0,0), O I777.3nm, Na I589.06nm, Cl I837.59nm.
3. the method for raising laser probe plastic identification precision according to claim 1, it is characterized in that, in 2nd step, the process of described reading characteristic spectral line intensity is: wavelength centered by standard line wavelength, read out ± 0.1nm within the scope of maximum intensity as line strength; M altogether
2individual spectrum, reads the intensity of 15 characteristic spectral lines in each spectrum successively.
4. the method for raising laser probe plastic identification precision according to claim 1, it is characterized in that, in 2nd step, the process of the normalized of described characteristic spectral line intensity is: to each spectrum, the intensity of all 15 characteristic spectral lines is all divided by C I247.86nm line strength in same spectrum, thus convert the actual strength of characteristic spectral line to normalized intensity, thus obtain M
2group spectroscopic data; Often organize the plastic type belonging to 15 these spectrum of normalized intensity value unique identification in spectroscopic data.
5., according to the method for described raising laser probe plastic identification precision arbitrary in Claims 1-4, it is characterized in that, in the 4th step, described SVM classifier modeling process is:
(a1), under ar gas environment, adopt same process parameter, utilize same laser probe device to gather the plasma spectrometry of multiple plastics, often kind of at least N number of plasma spectrometry of plastics collection, obtains M
1group spectroscopic data;
(a2) selected characteristic spectral line, reading characteristic spectral line intensity, and normalized is done to the intensity of characteristic spectral line;
(a5) 15 average normalized intensity levels that 15 characteristic spectral lines correspond to are obtained, again 6 average normalized intensity levels corresponding to 5 strip metal element spectral lines easy excitated in characteristic spectral line and protium spectral line are averaged again, obtain the overall average normalized intensity value I of easy excitated spectral line
tot; By I
totrespectively divided by C-N (0,0), C-C (0,0) and the average normalized intensity level corresponding to O I777.41nm tri-key feature spectral lines, obtain three weight factor: f
1, f
2, f
3.
(a6) C-N (0,0) is increased, the classified weight of C-C (0,0) and O I777.41nm tri-key feature spectral lines; All M
1in group spectroscopic data, the normalized intensity value of C-N (0,0), C-C (0,0) and O I777.41nm tri-key feature spectral lines is multiplied by f respectively
1, f
2, f
3three factors, make the classified weight of three spectral lines be increased to the level of metallic element spectral line and hydrogen spectral line;
(a7) by M
1the label of group spectroscopic data and correspondence thereof, as the input of SVM algorithm, for training SVM, sets up SVM classifier model.
6. the method for raising laser probe plastic identification precision according to claim 5, is characterized in that, the nuclear parameter g in the punishment parameter C of algorithm of support vector machine and RBF kernel function is respectively 20.12,0.012.
7. one kind realizes the device of method described in claim 1, it is characterized in that, it comprises laser instrument (1), optical maser wavelength catoptron (2), condenser lens (3), electric platforms (4), argon gas gas blow pipe (5), light collector (6), optical fiber (7), spectrometer (8), ICCD (9), trigger line (10), data line (11), and data processor (12);
Laser instrument (1), optical maser wavelength catoptron (2) and condenser lens (3) are positioned in same light path successively; Electric platforms (4) is positioned on the emitting light path of condenser lens (3), for placing plastics to be measured; Argon gas gas blow pipe (5) is for sending into argon gas to laser action district, light collector (6) is for collecting plasma spectrometry, light collector (6) is connected with spectrometer (8) by optical fiber (7), ICCD is connected with laser instrument (1) by triggering line (10), and ICCD (9) is connected with data processor (12) by data line (11).
8. device according to claim 7, it is characterized in that, in this device course of work, laser instrument (1) exports pulse laser beam, laser beam is after catoptron (2) reflection, vertically focus on plastic sample on the surface through after condenser lens (3), inspire plasma; Plasma emissioning light is collected by light collector (6) and is coupled in optical fiber (7), in optical fiber (7) transmission also lead-in light spectrometer (8); While laser instrument (1) outgoing laser beam, also pulse triggering signal is sent to ICCD, trigger the electronic switch of ICCD, plasma spectrometry is recorded by ICCD after spectrometer diffraction, light splitting, and the spectral information of each element is presented on data processor (12) after treatment; Two dimension electric platforms (4) for placing sample, and drives plastics to be measured to keep doing " bow " zigzag motion in spectra collection process.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310710790.8A CN104730041B (en) | 2013-12-20 | 2013-12-20 | Method and apparatus for improving plastic identification precision of laser probe |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310710790.8A CN104730041B (en) | 2013-12-20 | 2013-12-20 | Method and apparatus for improving plastic identification precision of laser probe |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104730041A true CN104730041A (en) | 2015-06-24 |
| CN104730041B CN104730041B (en) | 2017-03-22 |
Family
ID=53454137
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310710790.8A Active CN104730041B (en) | 2013-12-20 | 2013-12-20 | Method and apparatus for improving plastic identification precision of laser probe |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104730041B (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105527274A (en) * | 2016-01-29 | 2016-04-27 | 华中科技大学 | Efficient multipath laser probe analysis system and method |
| CN105806827A (en) * | 2016-03-11 | 2016-07-27 | 华中科技大学 | Method for identifying plastics by virtue of laser probe based on non-metallic element |
| CN106404748A (en) * | 2016-09-05 | 2017-02-15 | 华中科技大学 | Multispectral combined laser induced breakdown spectroscopy cereal crop producing area identification method |
| CN108802010A (en) * | 2018-07-25 | 2018-11-13 | 湖北工程学院 | A kind of spectrum continuous background removal device, system and method |
| CN109063773A (en) * | 2018-08-03 | 2018-12-21 | 华中科技大学 | A method of laser microprobe nicety of grading is improved using characteristics of image |
| CN110220871A (en) * | 2019-06-19 | 2019-09-10 | 中国科学院沈阳自动化研究所 | A kind of microcell LIBS plasma spectrometry collection system |
| CN110334936A (en) * | 2019-06-28 | 2019-10-15 | 阿里巴巴集团控股有限公司 | A kind of construction method, device and the equipment of credit qualification Rating Model |
| CN110751048A (en) * | 2019-09-20 | 2020-02-04 | 华中科技大学 | Laser probe classification method and device based on image characteristic automatic spectral line selection |
| CN112782151A (en) * | 2021-02-22 | 2021-05-11 | 湖北工程学院 | Data processing method for improving classification accuracy of laser-induced breakdown spectroscopy |
| CN115615881A (en) * | 2022-10-13 | 2023-01-17 | 中国水利水电科学研究院 | Small-particle-size micro-plastic detection method and system, electronic equipment and medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2003027650A2 (en) * | 2001-09-26 | 2003-04-03 | Ben-Gurion University Of The Negev | Method for detection and discrimination of polycyclic aromatic hydrocarbons (pahs) and monoaromatics based on laser-induced breakdown spectroscopy (libs) |
| CN101750404A (en) * | 2008-12-12 | 2010-06-23 | 中国科学院沈阳自动化研究所 | Method for correcting plasma emission spectral line self-absorption effect |
| CN101949852A (en) * | 2010-07-30 | 2011-01-19 | 清华大学 | Spectral standardization-based coal quality on-line detection method |
| CN102262076A (en) * | 2011-07-26 | 2011-11-30 | 清华大学 | Laser-induced breakdown spectroscopy element concentration determination method based on spectral line combination |
-
2013
- 2013-12-20 CN CN201310710790.8A patent/CN104730041B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2003027650A2 (en) * | 2001-09-26 | 2003-04-03 | Ben-Gurion University Of The Negev | Method for detection and discrimination of polycyclic aromatic hydrocarbons (pahs) and monoaromatics based on laser-induced breakdown spectroscopy (libs) |
| WO2003027650A3 (en) * | 2001-09-26 | 2003-11-06 | Univ Ben Gurion | Method for detection and discrimination of polycyclic aromatic hydrocarbons (pahs) and monoaromatics based on laser-induced breakdown spectroscopy (libs) |
| CN101750404A (en) * | 2008-12-12 | 2010-06-23 | 中国科学院沈阳自动化研究所 | Method for correcting plasma emission spectral line self-absorption effect |
| CN101949852A (en) * | 2010-07-30 | 2011-01-19 | 清华大学 | Spectral standardization-based coal quality on-line detection method |
| CN102262076A (en) * | 2011-07-26 | 2011-11-30 | 清华大学 | Laser-induced breakdown spectroscopy element concentration determination method based on spectral line combination |
Non-Patent Citations (2)
| Title |
|---|
| MYRIAM BOUERI, ET. AL: "Identification of Polymer Materials Using Laser-Induced Breakdown Spectroscopy Combined with Artificial Neural Networks", 《APPLIED SPECTROSCOPY》 * |
| 于洋等: "支持向量机算法在激光诱导击穿光谱技术塑料识别中的应用研究", 《物理学报》 * |
Cited By (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105527274B (en) * | 2016-01-29 | 2018-01-02 | 华中科技大学 | A kind of efficient multi-path laser probe analysis System and method for |
| CN105527274A (en) * | 2016-01-29 | 2016-04-27 | 华中科技大学 | Efficient multipath laser probe analysis system and method |
| CN105806827A (en) * | 2016-03-11 | 2016-07-27 | 华中科技大学 | Method for identifying plastics by virtue of laser probe based on non-metallic element |
| CN106404748A (en) * | 2016-09-05 | 2017-02-15 | 华中科技大学 | Multispectral combined laser induced breakdown spectroscopy cereal crop producing area identification method |
| CN106404748B (en) * | 2016-09-05 | 2019-03-05 | 华中科技大学 | A kind of multiline combination laser induced breakdown spectroscopy cereal crops Production area recognition method |
| CN108802010B (en) * | 2018-07-25 | 2019-11-05 | 湖北工程学院 | A kind of spectrum continuous background removal device, system and method |
| CN108802010A (en) * | 2018-07-25 | 2018-11-13 | 湖北工程学院 | A kind of spectrum continuous background removal device, system and method |
| CN109063773A (en) * | 2018-08-03 | 2018-12-21 | 华中科技大学 | A method of laser microprobe nicety of grading is improved using characteristics of image |
| CN110220871A (en) * | 2019-06-19 | 2019-09-10 | 中国科学院沈阳自动化研究所 | A kind of microcell LIBS plasma spectrometry collection system |
| CN110334936A (en) * | 2019-06-28 | 2019-10-15 | 阿里巴巴集团控股有限公司 | A kind of construction method, device and the equipment of credit qualification Rating Model |
| CN110334936B (en) * | 2019-06-28 | 2023-09-29 | 创新先进技术有限公司 | Method, device and equipment for constructing credit qualification scoring model |
| CN110751048A (en) * | 2019-09-20 | 2020-02-04 | 华中科技大学 | Laser probe classification method and device based on image characteristic automatic spectral line selection |
| CN110751048B (en) * | 2019-09-20 | 2022-06-14 | 华中科技大学 | Laser probe classification method and device based on image characteristic automatic spectral line selection |
| CN112782151A (en) * | 2021-02-22 | 2021-05-11 | 湖北工程学院 | Data processing method for improving classification accuracy of laser-induced breakdown spectroscopy |
| CN112782151B (en) * | 2021-02-22 | 2023-01-13 | 湖北工程学院 | Data processing method for improving classification accuracy of laser-induced breakdown spectroscopy |
| CN115615881A (en) * | 2022-10-13 | 2023-01-17 | 中国水利水电科学研究院 | Small-particle-size micro-plastic detection method and system, electronic equipment and medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104730041B (en) | 2017-03-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104730041A (en) | Method and apparatus for improving plastic identification precision of laser probe | |
| Sheng et al. | Classification of iron ores by laser-induced breakdown spectroscopy (LIBS) combined with random forest (RF) | |
| Junjuri et al. | Femtosecond laser-induced breakdown spectroscopy studies for the identification of plastics | |
| EP1992939A1 (en) | A kernel-based method and apparatus for classifying materials or chemicals and for quantifying the properties of materials or chemicals in mixtures using spectroscopic data. | |
| US9630348B2 (en) | Detection in thermoplastics | |
| CN102982339B (en) | A kind of method that EO-1 hyperion characteristic variable is chosen | |
| Wang et al. | Rapid quantitative analysis of the acidity of iron ore by the laser-induced breakdown spectroscopy (LIBS) technique coupled with variable importance measures-random forests (VIM-RF) | |
| CN104849231A (en) | Plastic material on-line recognition method and device | |
| CN105718749A (en) | Coal quality characteristic analysis method based on large database identification | |
| Liu et al. | Rapid classification of plastic bottles by laser-induced breakdown spectroscopy (LIBS) coupled with partial least squares discrimination analysis based on spectral windows (SW-PLS-DA) | |
| CN107220625A (en) | A kind of LIBS spectrum automatic recognition classification methods to sample | |
| CN103278467A (en) | Rapid nondestructive high-accuracy method with for identifying abundance degree of nitrogen element in plant leaf | |
| Lei et al. | Geographic origin identification of coal using near-infrared spectroscopy combined with improved random forest method | |
| Nie et al. | Identification of different colored plastics by laser-induced breakdown spectroscopy combined with neighborhood component analysis and support vector machine | |
| Ansari et al. | Comparison between support vector machine and random forest for audio classification | |
| Ma et al. | Small-sample stacking model for qualitative analysis of aluminum alloys based on femtosecond laser-induced breakdown spectroscopy | |
| Zhao et al. | Laser-induced breakdown spectroscopy for the discrimination of explosives based on the relieff algorithm and support vector machines | |
| CN205538680U (en) | Device of laser probe discernment plastics | |
| CN105717093A (en) | Cement characteristic analysis method based on large database recognition | |
| CN105717094A (en) | Metal element analysis method based on large database recognition | |
| CN105806827A (en) | Method for identifying plastics by virtue of laser probe based on non-metallic element | |
| Xu et al. | Spectral preprocessing combined with feature selection improve model robustness for plastics samples classification by LIBS | |
| Andrianomena | Probabilistic learning for pulsar classification | |
| CN113567417A (en) | Method for identifying peanut oil production place based on Raman spectrum fingerprint analysis technology | |
| Shao et al. | Classification method of coal and gangue using terahertz time-domain spectroscopy, cluster analysis and principal component analysis |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CB03 | Change of inventor or designer information | ||
| CB03 | Change of inventor or designer information |
Inventor after: Li Xiangyou Inventor after: Ren Zhao Inventor after: Yu Yang Inventor after: Guo Lianbo Inventor after: Zeng Xiaoyan Inventor after: Hao Zhongqi Inventor after: Zheng Zhong Inventor after: Li Kuohu Inventor after: Shen Meng Inventor after: Zeng Qingdong Inventor before: Li Xiangyou Inventor before: Zeng Qingdong Inventor before: Ren Zhao Inventor before: Yu Yang Inventor before: Guo Lianbo Inventor before: Zeng Xiaoyan Inventor before: Lu Yongfeng Inventor before: Hao Zhongqi Inventor before: Zheng Zhong Inventor before: Li Kuohu Inventor before: Shen Meng |
|
| CB03 | Change of inventor or designer information | ||
| CB03 | Change of inventor or designer information |
Inventor after: Li Xiangyou Inventor after: Ren Zhao Inventor after: Yu Yang Inventor after: Guo Lianbo Inventor after: Zeng Xiaoyan Inventor after: Hao Zhongqi Inventor after: Zheng Zhong Inventor after: Li Kuohu Inventor after: Shen Meng Inventor after: Zeng Qingdong Inventor before: Li Xiangyou Inventor before: Zeng Qingdong Inventor before: Ren Zhao Inventor before: Yu Yang Inventor before: Guo Lianbo Inventor before: Zeng Xiaoyan Inventor before: Lu Yongfeng Inventor before: Hao Zhongqi Inventor before: Zheng Zhong Inventor before: Li Kuohu Inventor before: Shen Meng |