Method for rapidly identifying and sorting metal components of waste electronic products based on multispectral fusion

Abstract

The invention discloses a multispectral fusion waste electronic product metal component rapid identification and sorting method, and relates to the technical field of metal identification, the precision conditions of an XRF spectrometer and an LIBS spectrometer are respectively judged, and the average offset degree, the identification error coefficient and the identification error prediction value are calculated based on the collected characteristic peaks; s3, judging the precision conditions, directly skipping S2 to execute S3 when the precision conditions of the XRF spectrograph and the LIBS spectrograph are normal, and executing S2 when the precision conditions of the XRF spectrograph and the LIBS spectrograph are abnormal; performing precision correction on the spectrometers with abnormal precision conditions, and correcting the characteristic peaks collected by the spectrometers based on the characteristic peak offset correction coefficient; metal components are judged and preliminarily recognized and sorted on the basis of comparison between the characteristic peaks of the metal particles to be detected and the standard characteristic peak database, and the metal particles to be detected with the atom sequence number larger than 13 are secondarily recognized through the XRF spectrograph and sorted through the pneumatic injection sorting machine.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of metal identification, and in particular to a multi-spectrum fusion method for rapid identification and sorting of metal components of waste electronic products. BACKGROUND

[0002] With the continuous development of technology, electronic products are updated and iterated more and more quickly, thus raising the problem of how to improve the efficient and accurate recovery of metals in waste electronic products. Traditional waste electronic product recovery mainly relies on manual disassembly, magnetic separation, and eddy current sorting, which has problems such as low efficiency and low precision. In recent years, spectrometers have gradually been widely used in waste electronic products, but single spectrometers have certain defects. LIBS spectrometers are fast and have high precision in light element detection but low precision in heavy metal analysis, and XRF analysis is slow but has high precision in heavy metal analysis.

[0003] Chinese patent application with publication number CN118071731A and publication date of 2024.05.24 discloses a waste metal identification method, which includes establishing a metal segmentation data set and a thermal imaging data set; collecting a metal image to be detected, and judging the metal category through original gray value analysis and comparison, sorting the metal to be detected based on the first judgment result, heating and collecting thermal imaging images of the metal to be detected with multiple metals in the first judgment, and secondarily judging the sorted metal to be detected through comparison of the thermal imaging data set. The method can make a first judgment on the metal category through visual processing, and make a second judgment on the metal category through thermal imaging images.

[0004] In the above application, the metal category is first judged through visual processing based on the data image set and the thermal imaging set, and then the metal is secondarily judged through thermal imaging images. The method of thermal imaging has a large error, and does not consider the problem of reduced precision in the actual process, and has the problem of unstable precision.

[0005] Therefore, the present application provides a multi-spectrum fusion method for rapid identification and sorting of metal components of waste electronic products. SUMMARY

[0006] The present application aims to solve the above problems by providing a multi-spectrum fusion method for rapid identification and sorting of metal components of waste electronic products.

[0007] To achieve the above purpose, the present application specifically adopts the following technical solutions:

[0008] A multi-spectrum fusion method for rapid identification and sorting of metal components of waste electronic products, comprising:

[0009] S1, respectively, the precision condition of the XRF spectrometer and the LIBS spectrometer is judged, the characteristic peaks of the correction metal sample are collected by the XRF spectrometer and the LIBS spectrometer, the average offset degree of the characteristic peaks, the identification error coefficient of the spectrometer and the identification error prediction value of the spectrometer are calculated based on the collected characteristic peaks, and the precision condition of the spectrometer is judged, when the precision conditions of the XRF spectrometer and the LIBS spectrometer are normal, S3 is directly executed;

[0010] S2, when the precision condition of any one or both of the XRF spectrometer and the LIBS spectrometer is abnormal, the precision of the spectrometer with abnormal precision condition is corrected, the characteristic peak offset correction coefficient and the characteristic peak correction function are calculated, and the precision of the spectrometer is corrected based on the characteristic peak correction function;

[0011] S3, the characteristic peaks of the metal particles to be measured are obtained by the LIBS spectrometer, the metal components are judged and preliminarily identified and sorted based on the comparison between the characteristic peaks of the metal particles to be measured and the standard characteristic peak database, the metal components containing the metal particles to be measured with atomic number greater than 13 are identified in the preliminary identification and sorting, and the secondary identification is performed by the XRF spectrometer, and the sorting is performed by the pneumatic jet sorting machine after the identification.

[0012] Preferably, the S1 comprises the following steps:

[0013] S11, the single substance or alloy of the elements contained in the metal components contained in the waste electronic products to be processed is selected as the correction metal sample, the standard characteristic peaks Mt in the spectrum image of the correction metal sample are queried and recorded i ;

[0014] S12, the characteristic peak measurement value Tz of the correction metal sample detected by the XRF spectrometer and the LIBS spectrometer is obtained ij , the characteristic peak measurement value Tz ij and the standard characteristic peak Mt i are calculated, and the average offset degree Av of the characteristic peaks of the correction metal sample is obtained i ;

[0015] S13, the average offset degree Av of the characteristic peaks of the correction metal sample in the detection period is obtained i , and the identification error coefficient Jd of the XRF spectrometer and the LIBS spectrometer is calculated z ;

[0016] S14, the work log of the precision detection of the spectrometer is queried, the identification error coefficient Jd of the spectrometer in the past h periods is obtained z , and the identification precision change rate Bh of the spectrometer is calculated z ;

[0017] S15, the identification error prediction value Yc of the spectrometer is calculated zAccording to the error prediction value Yc of the spectrometer z The accuracy of the spectrometer is judged, and corresponding measures are taken.

[0018] Preferably, the S11 comprises the following steps:

[0019] S11, selecting the elemental substance or alloy contained in the metal component as a correction metal sample according to the metal component contained in the waste electronic products to be processed, querying and recording the standard characteristic peaks Mt in the correction metal sample spectrum image i ;

[0020] Wherein, i represents the number of characteristic peak types, i=1, 2, …, m, and m represents the maximum value of the characteristic peak type number.

[0021] Preferably, the S12 comprises the following steps:

[0022] S12, measuring the characteristic peak value Tz of the correction metal sample by the XRF spectrometer and the LIBS spectrometer ij , calculating the average offset degree Av of the characteristic peak of the correction metal sample i :

[0023]

[0024] j represents the sequential number of all data of this type of characteristic peak, j=1, 2, …, n, and n represents the maximum value of the sequential number.

[0025] Preferably, the S13 comprises the following steps:

[0026] S13, obtaining the average offset degree Av of the characteristic peak of the correction metal sample in the detection period i , calculating the identification error coefficient Jd of the XRF spectrometer and the LIBS spectrometer z :

[0027]

[0028] Wherein, z represents the number of detection periods, z=1, 2, …, h, and maxAv i represents the maximum value of Av i in the detection period.

[0029] Preferably, the S14 comprises the following steps:

[0030] S14, querying the work log of the spectrometer accuracy detection, obtaining the spectrometer identification error coefficient Jd of the past h periods z , calculating the identification accuracy change rate Bh of the spectrometer z :

[0031]

[0032] wherein, represents the mean value of the spectrometer identification error coefficient Jd z in the past h cycles, represents the mean value of the cycle number z in the past h cycles.

[0033] Preferably, the S15 comprises the following steps:

[0034] S15, calculating the spectrometer identification error prediction value Yc z , according to the error prediction value Yc z judging the accuracy of the spectrometer identification and taking corresponding measures:

[0035] Yc z = Jd z (1 + Bh z )

[0036] When the error prediction value Yc is greater than 1, it indicates that the current spectrometer identification error is large, the accuracy is abnormal, and the spectral characteristic peak needs to be corrected. Wherein, is the mean value of the detection identification error coefficient Jd z in the past h times, and σ represents the variance of the detection identification error coefficient Jd z in the past h times.

[0037] The spectrometer acts on the sample with external energy, so that the atoms / molecules of the sample are excited, and the excited sample will release characteristic signals. Through a grating, a prism or an interferometer, the composite light is separated by wavelength / energy, and the intensity of light of different wavelengths is recorded by a detector. By comparing the characteristic peak with the database, the elements / molecules are quantitatively or qualitatively analyzed.

[0038] The X-ray fluorescence spectrometer XRF irradiates the sample with X-rays, excites the inner shell electrons of the atoms, makes them transition and release characteristic X-ray fluorescence. By detecting the energy and intensity of these fluorescent light, the element composition of the sample is determined. Laser-induced breakdown spectrometer LIBS uses high-energy laser pulses to bombard the sample surface to generate plasma, and analyzes the emission spectrum to determine the element composition. LIBS has a significant advantage in the detection of light elements, while XRF has an advantage in the detection of heavy elements.

[0039] Preferably, the S2 comprises the following steps:

[0040] S21, acquiring the characteristic peak measurement value Tz ij of the calibration metal sample by the XRF spectrometer and the LIBS spectrometer, calculating the characteristic peak shift coefficient;

[0041] S22, correcting the characteristic peak shift coefficient Ze i, calculate the characteristic peak correction function, and introduce the characteristic peak correction function into the spectral instrument correction accuracy.

[0042] Preferably, the S21 comprises the following steps:

[0043] S21, collecting the characteristic peak measurement value Tz of the calibration metal sample by the XRF spectrometer and the LIBS spectrometer ij , calculate the characteristic peak offset coefficient Ze i :

[0044]

[0045] The characteristic peak offset coefficient calculation formula is as above.

[0046] Preferably, the S22 comprises the following steps:

[0047] Based on the characteristic peak offset correction coefficient Ze i , calculate the characteristic peak correction function, and introduce the characteristic peak correction function into the spectral instrument correction accuracy, and the expression of the characteristic peak correction function is as follows:

[0048] Xz ik = (1 + Ze i )CL ik

[0049] Wherein, Xz ik is the corrected characteristic peak, CL ik is the uncorrected characteristic peak k represents the number of the metal particles to be measured, k = 1, 2, …, a.

[0050] Preferably, the S3 comprises the following steps:

[0051] S31, preliminary identification of the metal particles to be measured by the LIBS spectrometer, the metal particles to be measured being generated by crushing the metal in waste electronic products, and the identification including identifying the element type and the corresponding atomic number, for the metal components identified in the metal particles to be measured, the metal atomic number is not greater than 13, directly through the pneumatic blowing sorting machine to control the high pressure gas valve to blow the gas flow, and the target particles are blown into the specified collection tank to complete the preliminary sorting, and for the metal components identified in the metal particles to be measured, the atomic number is greater than 13, and the next step of identification and sorting is entered.

[0052] S32, further identification and sorting of the metal components identified in the metal particles to be measured in the preliminary identification and sorting, secondary identification by the XRF spectrometer, and after identification, the target particles are blown into the specified collection tank by the pneumatic blowing sorting machine to control the high pressure gas valve to blow the gas flow, and the target particles are blown into the specified collection tank to complete the sorting.

[0053] The pneumatic blowing sorting machine is a high-speed automatic sorting equipment based on spectrum analysis, which realizes efficient separation between metal and non-metal, and different metals through high-precision jet of target particles by high-pressure airflow. The specific sorting rules are specifically formulated and adjusted according to the actual application requirements of metal purity or the price fluctuation of precious metals.

[0054] The beneficial effects of the present application are as follows:

[0055] 1、The present application calculates the average offset degree, identification error coefficient and identification error prediction value by the standard characteristic peak and the characteristic peak measurement value of the corrected metal sample, and the error prediction value reflects the precision of the spectrometer, which helps managers to find the low precision of the spectrometer in time.

[0056] 2、The present application corrects the precision by directly correcting the characteristic peak measurement error, so that the specific situation of the influence of environmental factors or spectrometer itself on the precision of the spectrometer is not considered, and the calculation amount is reduced; the precision of the spectrometer is corrected, the accuracy of the identification of the metal component by the spectrometer is ensured, and the situation that the wrong result is used to sort the metal particles to be tested when the precision of the spectrometer is low is avoided, so that the precious metal is wasted or the recovery cost is increased.

[0057] 3、The present application collects the characteristic peak of the metal particles to be tested by the LIBS spectrometer, compares the standard characteristic peak database, and preliminarily identifies and sorts the metal particles to be tested, which improves the detection efficiency while rapidly and accurately sorting the metal particles to be tested with atomic number not greater than 13; the metal particles to be tested with atomic number greater than 13 are identified and sorted by the XRF spectrometer for secondary identification and sorting, which makes up for the defect that the LIBS spectrometer has low identification and sorting precision for the metal particles to be tested with atomic number greater than 13. BRIEF DESCRIPTION OF DRAWINGS

[0058] Figure 1 is a flowchart of a multi-spectrum fusion waste electronic product metal component rapid identification and sorting method of the present application. DETAILED DESCRIPTION

[0059] In order to make the purpose, technical scheme and advantages of the embodiments of the present application more clear, the technical scheme in the embodiments of the present application will be described clearly and completely with reference to the drawings in the embodiments of the present application.

[0060] A multi-spectrum fusion waste electronic product metal component rapid identification and sorting method according to a preferred embodiment of the present application will be described in detail as follows.

[0061] The application discloses a kind of multispectral fusion's waste electronic product metal component rapid identification and sorting method, comprising the following steps:

[0062] S1, the precision condition of XRF spectrometer and LIBS spectrometer is judged respectively, the characteristic peak of calibration metal sample is collected by XRF spectrometer and LIBS spectrometer, the average deviation of characteristic peak is calculated based on the collected characteristic peak, the identification error coefficient of spectrometer and the identification error prediction value of spectrometer, the precision condition of spectrometer is judged, when the precision condition of XRF spectrometer and LIBS spectrometer is normal, directly execute S3;

[0063] S1 includes the following steps:

[0064] S11, the single substance or alloy of the metal component contained element in metal component contained in waste electronic product to be processed is selected as calibration metal sample, the standard characteristic peak Mt in calibration metal sample spectral image is inquired and recorded i ;

[0065] Wherein, i indicates the number of characteristic peak type, i=1, 2, …, m, m indicates the maximum value of characteristic peak type number.

[0066] S12, the characteristic peak measurement value Tz of XRF spectrometer and LIBS spectrometer to calibration metal sample detection is obtained ij , based on characteristic peak measurement value Tz ij And standard characteristic peak Mt i Calculated to obtain the average deviation Av of calibration metal sample characteristic peak i :

[0067]

[0068] Wherein, j indicates the order number of all data of this kind of characteristic peak, j=1, 2, …, n, n indicates the maximum value of order number.

[0069] S13, the average deviation Av of calibration metal sample characteristic peak in detection period is obtained i , the identification error coefficient Jd of XRF spectrometer and LIBS spectrometer is calculated z :

[0070]

[0071] Wherein, z indicates the number of detection period, z=1, 2, …, h, maxAv i It indicates the maximum value of Av i In detection period.

[0072] S14, the working log of spectrometer precision detection is inquired, the spectrometer identification error coefficient Jd of past h cycles is obtainedz , calculate the change rate Bh of the spectrometer recognition accuracy z :

[0073]

[0074] wherein, represents the average of the spectrometer recognition error coefficient Jd in the past h cycles z , represents the average of the cycle number z in the past h cycles.

[0075] S15, based on the recognition accuracy change rate Bh z and the recognition error coefficient Jd z calculate the spectrometer recognition error prediction value Yc z , according to the error prediction value Yc z judge the accuracy of the spectrometer recognition and take corresponding measures:

[0076] Yc z = Jd z (1 + Bh z )

[0077] When the error prediction value , it indicates that the current spectrometer recognition error is large and the accuracy is abnormal, and the spectral feature peak needs to be corrected. Among them, is the average of the recognition error coefficient Jd z in the past h times of detection, and σ represents the variance of the recognition error coefficient Jd z in the past h times of detection.

[0078] The error prediction value reflects the spectrometer accuracy from this accuracy detection to the next accuracy detection, which can better represent the accuracy of the spectrometer in subsequent operation than the current spectrometer accuracy, avoiding the situation that the spectrometer accuracy is insufficient from the current detection to the next detection when the current detection accuracy is qualified.

[0079] The accuracy of the spectrometer is judged to ensure the accuracy of the spectrometer when collecting metal feature peaks, and to avoid the situation that the metal component recognition and separation efficiency is low when the spectrometer accuracy is abnormal.

[0080] The spectrometer acts on the sample with external energy, so that the atoms / molecules of the sample are excited. The excited sample will release characteristic signals. The composite light is separated by grating, prism or interferometer according to wavelength / energy, and the intensity of different wavelengths of light is recorded by detector. Compare the characteristic peak with the database, and quantitatively or qualitatively analyze elements / molecules.

[0081] X-ray fluorescence spectrometer XRF irradiates the sample by X-ray, excites the inner layer electron of atom, makes it jump and releases characteristic X-ray fluorescence. By detecting the energy and intensity of these fluorescent, the element composition of the sample is determined. Laser-induced breakdown spectrometer LIBS uses high-energy laser pulse to hit the sample surface, generates plasma, analyzes its emission spectrum to determine the element composition. LIBS has obvious advantages in the detection of light elements, while XRF is superior in the detection of heavy elements.

[0082] The average deviation, identification error coefficient and identification error prediction value are calculated by the standard characteristic peak and the measured value of the characteristic peak of the calibration metal sample. The error prediction value reflects the accuracy of the spectrometer, which helps managers to find the low accuracy of the spectrometer in time.

[0083] S2, when the accuracy of any one or both of the XRF spectrometer and the LIBS spectrometer is abnormal, the accuracy of the spectrometer with abnormal accuracy is corrected, the characteristic peak correction coefficient and the characteristic peak correction function are calculated, and the accuracy of the spectrometer is corrected based on the characteristic peak correction function;

[0084] S2 includes the following steps:

[0085] S21, the characteristic peak measurement value Tz of the calibration metal sample is collected by the XRF spectrometer and the LIBS spectrometer ij , the characteristic peak deviation coefficient Ze i is calculated:

[0086]

[0087] S22, according to the characteristic peak correction coefficient Ze i , the characteristic peak correction function is calculated, the characteristic peak correction function is introduced into the calibration accuracy of the spectrometer, and the expression of the characteristic peak correction function is as follows:

[0088] Xz ik =(1+Ze i )CL ik

[0089] Wherein, Xz ik is the corrected characteristic peak, CL ik is the uncorrected characteristic peak k represents the number of metal particles to be tested, k=1, 2, …, a.

[0090] The precision is corrected by directly correcting the feature peak measurement error, so that the specific conditions of the influence of environmental factors or spectrometer itself factors on the precision of the spectrometer are not considered, and the operation amount is reduced; by correcting the precision of the abnormal spectrometer, the accuracy of the metal component identification of the spectrometer is ensured, and the situation that the wrong result is used for sorting the metal particles to be tested without making correction when the precision of the spectrometer is abnormal, resulting in waste of precious metals or high recovery cost, is avoided.

[0091] S3, obtaining the characteristic peak of the metal particles to be tested by the LIBS spectrometer, judging the metal component and preliminarily identifying and sorting based on the comparison between the characteristic peak of the metal particles to be tested and the standard characteristic peak database, and performing secondary identification on the metal particles to be tested containing metal components with atomic numbers greater than 13 in the preliminary identification and sorting through the XRF spectrometer, and sorting through the pneumatic blowing sorting machine after identification.

[0092] S3 includes the following steps:

[0093] S31, preliminarily identifying the metal particles to be tested by the LIBS spectrometer, the metal particles to be tested being generated by crushing the metal in waste electronic products, the identification including identifying the element type and the corresponding atomic number, and for the metal particles to be tested, the metal components identified by the metal particles to be tested are directly blown into the specified collection tank by the pneumatic blowing sorting machine to complete the preliminary sorting, and for the metal particles to be tested containing metal components with atomic numbers greater than 13, the next step of identification and sorting is entered.

[0094] S32, further identifying and sorting the metal particles to be tested containing metal components with atomic numbers greater than 13 identified in the preliminary identification and sorting, performing secondary identification through the XRF spectrometer, and after identification, blowing the target particles into the specified collection tank through the pneumatic blowing sorting machine to complete the sorting.

[0095] The pneumatic blowing sorting machine is a high-speed automatic sorting equipment based on spectral analysis, which realizes efficient separation between metal and non-metal and different metals by accurately spraying target particles through high-pressure airflow. The specific sorting rules are specifically formulated and adjusted according to the actual application requirements for metal purity or precious metal price fluctuations.

[0096] The characteristic peak of the metal particles to be measured is collected by the LIBS spectrometer, the standard characteristic peak database is compared, the metal particles to be measured are preliminarily identified and sorted, the metal particles to be measured with atomic number greater than 13 are uniformly transported to the next step of identification and detection while the metal particles to be measured with atomic number not greater than 13 are rapidly and accurately sorted, and the detection efficiency is improved; the metal particles to be measured with atomic number greater than 13 identified in the preliminary identification and sorting are subjected to secondary identification and sorting by the XRF spectrometer, and the defect that the LIBS spectrometer has low identification and sorting accuracy for the metal particles to be measured with atomic number greater than 13 is made up.

[0097] The above description of disclosed embodiments enables one of ordinary skill in the art to make or use the application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein can be applied to other embodiments without departing from the spirit or scope of the application. Thus, the present application is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A multi-spectral fusion method for rapid identification and sorting of metal components of waste electronic products, characterized in that, Comprising the following steps: S1, judging the precision condition of the XRF spectrometer and the LIBS spectrometer respectively, collecting the characteristic peaks of the calibration metal sample through the XRF spectrometer and the LIBS spectrometer, calculating the average offset degree of the characteristic peaks, the identification error coefficient of the spectrometer and the identification error prediction value of the spectrometer based on the collected characteristic peaks, judging the precision condition of the spectrometer, when the precision conditions of the XRF spectrometer and the LIBS spectrometer are both normal, directly executing S3; S2, when the precision condition of any one or both of the XRF spectrometer and the LIBS spectrometer is abnormal, correcting the precision of the spectrometer with abnormal precision condition, calculating the characteristic peak offset correction coefficient and the characteristic peak correction function, and correcting the precision of the spectrometer based on the characteristic peak correction function; S3, acquiring the characteristic peaks of the metal particles to be measured through the LIBS spectrometer, judging the metal composition and preliminarily identifying and sorting based on the comparison between the characteristic peaks of the metal particles to be measured and the standard characteristic peak database, and performing secondary identification through the XRF spectrometer for the metal particles to be measured with atomic number greater than 13 identified in the preliminary identification and sorting, and sorting through the pneumatic jet sorting machine after identification.

2. The method according to claim 1, wherein The S1 comprises the following steps: S11, selecting a correction metal sample according to the metal components contained in the waste electronic products to be treated and recording standard characteristic peaks Mt in the correction metal sample spectrum image i ; S12. Acquire the characteristic peak measurement value Tz of the calibrated metal sample using an XRF spectrometer and a LIBS spectrometer. ij Based on the standard characteristic peak Mt i and characteristic peak measurement value Tz ij Calculate and obtain the average shift Av of the characteristic peaks of the corrected metal sample i ; S13, obtaining the average offset degree Av of the characteristic peak of the correction metal sample in the detection period i , calculating the identification error coefficient Jd of the XRF spectrometer and the LIBS spectrometer based on the average offset degree Av of the characteristic peak i z ;​ S14, query the working log of the accuracy detection of the spectrometer, obtain the identification error coefficient Jd of the past h periods z , calculate the change rate Bh of the identification accuracy of the spectrometer z ; S15, based on the rate of change of the accuracy of the spectrometer Bh z and the identification error coefficient Jd z Calculate the spectrometer error prediction value Yc z , according to the spectrometer error prediction value Yc z Determine the accuracy of the spectrometer identification and take corresponding measures.

3. The method according to claim 2, wherein, The S11 comprises the following steps: S11, selecting a single element or an alloy containing elements in the metal components contained in the waste electronic products to be treated as a correction metal sample according to the metal components, querying and recording the standard characteristic peaks Mt in the correction metal sample spectral image i ; Wherein, i represents the number of characteristic peak types, i=1, 2, …, m, and m represents the maximum value of the characteristic peak type number.

4. The method according to claim 2, wherein, The S12 comprises the following steps: S12, acquire the characteristic peak measurement value Tz of the correction metal sample by the XRF spectrometer and the LIBS spectrometer ij , calculate the characteristic peak average offset degree Av of the correction metal sample based on the standard characteristic peak Mt i and the characteristic peak measurement value Tz ij i :​ J represents the sequential number of all data of the characteristic peak, j=1, 2, …, n, and n represents the maximum value of the sequential number.

5. The method according to claim 2, wherein, The S13 comprises the following steps: S13, obtaining the average offset degree Av of the characteristic peak of the correction metal sample in the detection period i , based on the average offset degree Av of the characteristic peak i , calculating the identification error coefficient Jd of the XRF spectrometer and the LIBS spectrometer z : wherein z denotes the number of the detection period, z = 1, 2,..., h, maxAv i denotes the maximum value of Av i in the detection period.

6. The method according to claim 2, wherein, The S14 comprises the following steps: S14, query the working log of the accuracy detection of the spectrometer, obtain the spectrometer identification error coefficient Jd in the past h periods z , calculate the change rate Bh of the identification accuracy of the spectrometer z : wherein represents the mean value of the spectrometer identification error coefficient Jd over the past h periods z represents the mean value of the period number z over the past h periods represents the mean value of the period number z over the past h periods 7. The method according to claim 2, wherein the method is characterized by, The S15 comprises the following steps: S15, based on the rate of change of the accuracy of the spectrometer Bh z and the identification error coefficient Jd z Calculate the predicted value of the spectrometer identification error Yc z , according to the error prediction value Yc z Determine the accuracy of the spectrometer identification and take corresponding measures: Yc z = Jd z (1 + Bh z ) When the error prediction value indicates that the current spectrometer recognition error is large, the precision is abnormal, and the spectral characteristic peak needs to be corrected; is the mean value of the past h detection recognition error coefficients Jd z , and σ represents the variance of the past h detection recognition error coefficients Jd z .

8. The method according to claim 1, wherein, The S2 comprises the following steps: S21, obtaining the characteristic peak measurement value Tz of the correction metal sample detected by the XRF spectrometer and the LIBS spectrometer ij , calculating the characteristic peak offset coefficient Ze i ; The characteristic peak offset coefficient calculation formula is as above; S22, correct the coefficient Ze based on the feature peak offset i , calculate the feature peak correction function, and import the feature peak correction function into the spectrometer correction accuracy: Xz ik = (1 + Ze i )CL ik wherein Xz ik is the corrected characteristic peak, CL ik k represents the number of the metal particles to be measured, k = 1, 2, …, a.

9. The method according to claim 1, wherein The S3 comprises the following steps: S31, preliminarily identifying and sorting the metal particles to be measured through the LIBS spectrometer, directly blowing the target particles into the specified collection tank through the pneumatic jet sorting machine to complete the sorting for the metal particles to be measured with metal atomic number not greater than 13 identified, and entering the next step of screening for the metal particles to be measured with atomic number greater than 13 identified; S32, further identifying and sorting the metal particles to be measured with atomic number greater than 13 identified in the preliminary identification and sorting, performing secondary identification through the XRF spectrometer, and blowing the target particles into the specified collection tank through the pneumatic jet sorting machine to complete the sorting after identification.

Images