Method and device for reducing uncertainty in LIBS detection between multiple batches of samples

Abstract

The present invention discloses a method and device for reducing the uncertainty of LIBS detection between multiple batches of samples, the method includes a self-correction process within the same batch of samples and a correction process between multiple batches of samples; the self-correction process within the same batch of samples is for the spectral data within the same batch of samples, which can effectively reduce the volatility of the data within the batch; the correction process between multiple batches of samples is used for the training spectral data and the predicted spectral data between multiple batches of samples, which can solve the problem that the spectral intensity of the LIBS system will change after a long time to a certain extent, and reduce the uncertainty of LIBS detection between multiple batches of samples; the two methods are used in combination to improve the robustness of the system and the prediction accuracy of the LIBS system. In addition, in view of the problem of uncertainty caused by changes in samples or environments in the actual production process, a process for judging the robustness of the model is added, so that the model has the ability of self-learning and self-supplementation, thereby reducing the uncertainty of the model.

Description

Technical Field

[0001] The present invention relates to the field of laser plasma spectrum detection technology, and in particular to a method and device for reducing uncertainty in LIBS detection between multiple batches of samples. Background Art

[0002] Laser-induced breakdown spectroscopy (LIBS) is a material composition detection technology that has developed rapidly in recent years. It has low sample preparation requirements and fast analysis speed. It can realize non-contact and long-distance online analysis. By analyzing the characteristic spectral lines of atomic ions in the plasma emission spectrum, it can achieve qualitative and quantitative analysis of the object to be tested.

[0003] Steel production is a national benchmark, and it is gradually moving towards intelligent, continuous and high-precision production. The key to high-quality production is the rapid and online detection of molten steel composition during the production process. Among the methods for analyzing steel samples, LIBS has the advantages of low sample preparation requirements, fast analysis speed, and non-contact, long-distance online analysis, and has great application potential and prospects.

[0004] However, in the process of collecting laser spectra, for spectra collected in the same time period and the same batch, due to the complex and changeable industrial field environment, the collected spectra fluctuate greatly and even affect the establishment of the training model; and the training data and prediction data used for modeling are often not collected in the same time period. The training data is collected from multiple batches in multiple time periods. The prediction data is collected at uncertain times and batches in the industrial production process. Therefore, due to changes in laser energy or environment, data from different batches will bring large errors. The existing methods to reduce data volatility are generally to collect spectral data multiple times and take the average or to eliminate the probability density of multiple spectral data. Both methods will cause the loss of original spectral data, and cannot establish models for different industrial environments well, resulting in poor model robustness and a large gap between the model prediction results and the reference results, which affects industrial production. Summary of the invention

[0005] The present invention provides a method and device for reducing the uncertainty of LIBS detection between multiple batches of samples, so as to solve the technical problems existing in the prior art that the model robustness is poor, the gap between the model prediction results and the reference results is large, and industrial production is affected.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In one aspect, the present invention provides a method for reducing the uncertainty of LIBS detection between multiple batches of samples, the method for reducing the uncertainty of LIBS detection between multiple batches of samples comprising: a self-calibration process within the same batch of samples, a calibration process between multiple batches of samples, and a calibration process for predicted data;

[0008] The self-calibration process within the same batch of samples includes:

[0009] S1, obtaining sample data of the same batch to be calibrated; wherein the sample data of the same batch to be calibrated is training spectral data or predicted spectral data, the training spectral data refers to spectral data used to train a preset calibration model, and the predicted spectral data refers to spectral data used to input the calibration model to achieve prediction of preset components of the sample;

[0010] S2, select the inner standard line of the sample data, and extract the spectral intensity value of each sample at the inner standard line; remove the spectral intensity values ​​corresponding to the samples with the same element content by probability density, and take the average value of the remaining data as the standard spectral intensity value at the inner standard line corresponding to the sample with this element content;

[0011] S3, dividing the standard spectrum intensity value by the original spectrum intensity value corresponding to the sample with the same element content to obtain the self-correction coefficient α1 of each sample; multiplying the original spectrum intensity value of the sample by its corresponding self-correction coefficient α1 to obtain the corresponding self-corrected data;

[0012] The multi-batch sample calibration process includes:

[0013] S1, obtaining multiple batches of sample data to be calibrated;

[0014] S2, self-calibrating each batch of sample data to obtain multiple batches of sample data after self-calibration;

[0015] S3, taking the spectral intensity values ​​at the inner standard line of the n batches of sample data after self-calibration and the spectral intensity values ​​at the inner standard line of the first batch of sample data after self-calibration as quotients, and obtaining n-1 model correction coefficients β1, β2, β3…β n-1 ; Where n = 2, 3, 4...N, N represents the total batches of multiple batches of sample data;

[0016] S4, establishing an inter-batch calibration model using the model calibration coefficient corresponding to each batch of sample data and the spectral intensity value at the internal standard line of the batch after self-calibration; and implementing calibration between multiple batches of samples using the inter-batch calibration model;

[0017] The correction process of the predicted data includes:

[0018] S1, obtaining the prediction data to be corrected;

[0019] S2, self-correcting and multi-batch correction of the predicted data to obtain the corrected predicted data;

[0020] S3, determining whether the spectral intensity at the inner standard line of the corrected prediction data is within the intensity probability interval at the inner standard line corresponding to the training data, if yes, executing S4; otherwise, executing S5;

[0021] S4, bringing the corrected prediction data into a preset calibration model for calculation;

[0022] S5, add this batch of prediction data to the training data, collect prediction data again, and then execute S3; until the spectral intensity at the inner standard line of the prediction data is within the intensity probability interval at the inner standard line corresponding to the training data, substitute the prediction data into the model calculation.

[0023] Furthermore, the batch calibration model is established by using the model calibration coefficient corresponding to each batch of sample data and the spectral intensity value at the internal standard line after self-calibration of the batch, including:

[0024] The least squares support vector machine is used as the base learner, and the standard spectral intensity value of the nth batch of spectral data after self-correction is taken as input, and its corresponding model correction coefficient β n-1 As output;

[0025] The input and output are fed into the base learner for training to obtain the batch correction model.

[0026] On the other hand, the present invention provides a device for reducing the uncertainty of LIBS detection between multiple batches of samples, the device for reducing the uncertainty of LIBS detection between multiple batches of samples comprises: a self-correction module within the same batch of samples, a correction module between multiple batches of samples, and a correction module for predicted data;

[0027] The self-calibration module within the same batch of samples is used for:

[0028] S1, obtaining sample data of the same batch to be calibrated; wherein the sample data of the same batch to be calibrated is training spectral data or predicted spectral data, the training spectral data refers to spectral data used to train a preset calibration model, and the predicted spectral data refers to spectral data used to input the calibration model to achieve prediction of preset components of the sample;

[0029] S2, select the inner standard line of the sample data, and extract the spectral intensity value of each sample at the inner standard line; remove the spectral intensity values ​​corresponding to the samples with the same element content by probability density, and take the average value of the remaining data as the standard spectral intensity value at the inner standard line corresponding to the sample with this element content;

[0030] S3, dividing the standard spectrum intensity value by the original spectrum intensity value corresponding to the sample with the same element content to obtain the self-correction coefficient α1 of each sample; multiplying the original spectrum intensity value of the sample by its corresponding self-correction coefficient α1 to obtain the corresponding self-corrected data;

[0031] The multi-batch sample correction module is used for:

[0032] S1, obtaining multiple batches of sample data to be calibrated;

[0033] S2, self-calibrating each batch of sample data to obtain multiple batches of sample data after self-calibration;

[0034] S3, taking the spectral intensity values ​​at the inner standard line of the n batches of sample data after self-calibration and the spectral intensity values ​​at the inner standard line of the first batch of sample data after self-calibration as quotients, and obtaining n-1 model correction coefficients β1, β2, β3…β n-1 ; Where n = 2, 3, 4...N, N represents the total batches of multiple batches of sample data;

[0035] S4, establishing an inter-batch calibration model using the model calibration coefficient corresponding to each batch of sample data and the spectral intensity value at the internal standard line of the batch after self-calibration; and implementing calibration between multiple batches of samples using the inter-batch calibration model;

[0036] The correction module for the predicted data is used for:

[0037] S1, obtaining the prediction data to be corrected;

[0038] S2, self-correcting and multi-batch correction of the predicted data to obtain the corrected predicted data;

[0039] S3, determining whether the spectral intensity at the inner standard line of the corrected prediction data is within the intensity probability interval at the inner standard line corresponding to the training data, if yes, executing S4; otherwise, executing S5;

[0040] S4, bringing the corrected prediction data into a preset calibration model for calculation;

[0041] S5, add this batch of prediction data to the training data, collect prediction data again, and then execute S3; until the spectral intensity at the inner standard line of the prediction data is within the intensity probability interval at the inner standard line corresponding to the training data, substitute the prediction data into the model calculation.

[0042] Furthermore, the batch calibration model is established by using the model calibration coefficient corresponding to each batch of sample data and the spectral intensity value at the internal standard line after self-calibration of the batch, including:

[0043] The least squares support vector machine is used as the base learner, and the standard spectral intensity value of the nth batch of spectral data after self-correction is taken as input, and its corresponding model correction coefficient β n-1 As output;

[0044] The input and output are fed into the base learner for training to obtain the batch correction model.

[0045] On the other hand, the present invention further provides an electronic device, comprising a processor and a memory; wherein the memory stores at least one instruction, and the instruction is loaded and executed by the processor to implement the above method.

[0046] In yet another aspect, the present invention further provides a computer-readable storage medium, wherein the storage medium stores at least one instruction, and the instruction is loaded and executed by a processor to implement the above method.

[0047] The beneficial effects brought about by the technical solution provided by the present invention include at least:

[0048] 1. The self-correction within the same batch of samples of the present invention corrects the spectral data of the same batch according to the average intensity after probability density removal at the internal standard line, which can reduce the volatility of the training data of the same batch and improve the reliability of the data.

[0049] 2. The multi-batch sample correction of the present invention corrects the training data and prediction data of different batches to the same intensity level, which greatly avoids the errors caused by unstable laser energy or environmental changes when the time interval between spectral data collection between different batches is large, and greatly reduces the uncertainty of LIBS detection between multiple batches of samples.

[0050] 3. In actual production, the robustness of the model will be judged. If the predicted data does not fall within the probability range of the current model training data, the training data will be self-learned and self-supplemented to improve the robustness of the model, reduce the uncertainty of the entire system, and improve the accuracy of the prediction results. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0052] Figure 1 is a flow chart of a self-calibration method for samples from the same batch provided by an embodiment of the present invention;

[0053] Figure 2is a flow chart of a multi-batch sample calibration method provided by an embodiment of the present invention; wherein Θi represents the intensity value at the inner standard line of the i-th batch of data;

[0054] Figure 3 It is a flow chart of the prediction data correction method provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0055] In order to make the objectives, technical solutions and advantages of the present invention more clear, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0056] First embodiment

[0057] The present embodiment provides a method for reducing the uncertainty of LIBS detection between multiple batches of samples. The method can be implemented by an electronic device. The method includes a self-calibration process for samples within the same batch, a calibration process between multiple batches of samples, and a process for judging the robustness of the model. The method can reduce the volatility of data collected by the LIBS detection system, and to a certain extent solve the problem that the spectral intensity of the LIBS system will change after a long time. It can improve the robustness of the system, reduce the uncertainty of the prediction results, and significantly improve the accuracy of the detection results. The method can be used for various LIBS detection component data modeling and analysis processes.

[0058] Specifically, the implementation process of the method for reducing the uncertainty of LIBS detection between multiple batches of samples is as follows:

[0059] 1. Self-calibration process within the same batch of samples.

[0060] The self-calibration process of samples in the same batch is aimed at the spectral data in the same batch of samples, which can effectively reduce the volatility of the data within the batch, such as Figure 1 As shown, the correction process includes the following steps:

[0061] S1, obtaining the sample data of the same batch to be calibrated;

[0062] Among them, the sample data of the same batch to be corrected are training spectral data or predicted spectral data, the training spectral data refers to spectral data used to train a preset calibration model (for example, a support vector machine SVM calibration model for predicting the element content of the sample), and the predicted spectral data refers to spectral data of the sample to be tested that is used to input the calibration model to realize the prediction of the preset components of the sample.

[0063] S2, select the internal standard line of the sample data, and extract the spectral intensity value of each sample at the internal standard line; after the spectral intensity values ​​corresponding to the samples with the same content of the element to be measured are eliminated by probability density, the average value of the remaining data is taken as the standard spectral intensity value at the internal standard line corresponding to the sample of this element content;

[0064] The internal standard line refers to: a spectral line that is symmetrical to the analytical line is selected from the matrix elements (the main elements in the sample) or the added elements that do not exist in the matrix, and forms an analytical line pair with the analytical line. Quantitative analysis is performed based on the relationship between the ratio of the absolute intensities of the analytical line and the internal standard line and the concentration.

[0065] S3, dividing the standard spectrum intensity value by the original spectrum intensity value corresponding to the sample with the same element content to obtain the self-correction coefficient α1 of each sample; multiplying the original spectrum intensity value of the sample by its corresponding self-correction coefficient α1 to obtain the corresponding self-corrected data;

[0066] 2. Correction process between multiple batches of samples.

[0067] The multi-batch sample calibration process is mainly used for training spectral data and predicting spectral data between multi-batch samples. It can solve the problem that the spectral intensity of the LIBS system will change over a long period of time to a certain extent, and reduce the uncertainty of LIBS detection between multi-batch samples. Figure 2 As shown, it includes the following steps:

[0068] S1, obtaining multiple batches of sample data to be calibrated;

[0069] S2, self-correcting the first batch of collected training data, the second batch of collected training data, the third batch of collected training data ... the Nth batch of collected training data, respectively, to obtain N batches of self-corrected training data;

[0070] S3, the intensity value at the inner standard line of the second batch of training data after self-correction is divided by the intensity value at the inner standard line of the first batch of data after self-correction, and the model correction coefficient β1 is obtained. The intensity value at the inner standard line of the third batch of training data is divided by the intensity value at the inner standard line of the first batch of data, and the model correction coefficient β2 is obtained. The intensity value at the inner standard line of the Nth batch of training data is divided by the intensity value at the inner standard line of the first batch of data, and the model correction coefficient β N-1 ;

[0071] S4, using β1, β2…β N-1 , and the spectral intensity values ​​after self-correction at the inner standard line of the second batch, the third batch ... the Nth batch of training data are used to establish an inter-batch correction model; and the inter-batch correction model is used to realize correction between multiple batches of samples;

[0072] Specifically, in this embodiment, the batch correction model is based on the least squares support vector machine as the base learner, taking the spectral intensity value data of the Xth batch after self-correction as input and the corresponding correction coefficient β X As output (X is 1, 2, 3...N), the input and output are brought into the base learner for training to obtain the batch-to-batch correction model of spectral intensity value-model correction coefficient β.

[0073] 3. Correction process of predicted data, such as Figure 3 As shown, it includes the following steps:

[0074] S1, obtaining the prediction data to be corrected;

[0075] S2, self-correcting and multi-batch correction of the predicted data to obtain the corrected predicted data;

[0076] S3, determining whether the spectral intensity at the inner standard line of the corrected prediction data is within the intensity probability interval at the inner standard line corresponding to the training data, if yes, executing S4; otherwise, executing S5;

[0077] S4, bringing the corrected prediction data into a preset calibration model for calculation;

[0078] S5, add this batch of prediction data to the training data, collect prediction data again, and then execute S3; until the spectral intensity at the inner standard line of the prediction data is within the intensity probability interval at the inner standard line corresponding to the training data, substitute the prediction data into the model calculation.

[0079] The corrected data can be obtained through the above correction process, and the calibration model can be established using the corrected training spectrum data. For example, the establishment of a calibration model for predicting the element content of a sample can include the following process: using the corrected sample spectrum data as a training set, inputting the sample spectrum data and the corresponding element content as a training set, optimizing the SVM regression model parameters, and establishing a calibration model between the relevant element content and the spectrum data; thereafter, inputting the spectrum data of the sample to be tested as a prediction set, and inputting it into the SVM calibration model, quantitative prediction can be performed to obtain the content of the corresponding element in the sample to be tested.

[0080] In summary, the data correction method of this embodiment can not only use the self-correction method to reduce the uncertainty of training data in the same batch, reduce the fluctuation of spectral data, and reduce the impact of the environment on the spectrum; and for training data of multiple batches and multiple time periods and prediction data used in industrial production processes, a spectral data correction method between multiple batches of samples is used to establish an inter-batch correction model for inter-batch data correction, thereby correcting multiple batches of training data and prediction data to the same level, avoiding errors caused by large collection time intervals, unstable laser energy or environmental changes in spectral data between different batches, greatly reducing the uncertainty of LIBS detection between multiple batches of samples. In addition, in view of the uncertainty caused by changes in samples or the environment in the actual production process, the process of judging the robustness of the model is added, so that the model has the ability of self-learning and self-supplementation, thereby reducing the uncertainty of the model.

[0081] Second embodiment

[0082] This embodiment provides a device for reducing the uncertainty of LIBS detection between multiple batches of samples, including: a self-correction module within the same batch of samples, a correction module between multiple batches of samples, and a correction module for predicted data;

[0083] The self-calibration module within the same batch of samples is used to perform the following steps:

[0084] S1, obtaining sample data of the same batch to be calibrated; wherein the sample data of the same batch to be calibrated is training spectral data or predicted spectral data, the training spectral data refers to spectral data used to train a preset calibration model, and the predicted spectral data refers to spectral data used to input the calibration model to achieve prediction of preset components of the sample;

[0085] S2, select the internal standard line of the sample data, and extract the spectral intensity value of each sample at the internal standard line; after the spectral intensity values ​​corresponding to the samples with the same content of the element to be measured are eliminated by probability density, the average value of the remaining data is taken as the standard spectral intensity value at the internal standard line corresponding to the sample of this element content;

[0086] S3, dividing the standard spectrum intensity value by the original spectrum intensity value corresponding to the sample with the same element content to obtain the self-correction coefficient α1 of each sample; multiplying the original spectrum intensity value of the sample by its corresponding self-correction coefficient α1 to obtain the corresponding self-corrected data;

[0087] The multi-batch sample correction module is used to perform the following steps:

[0088] S1, obtaining multiple batches of sample data to be calibrated;

[0089] S2, self-calibrating each batch of sample data to obtain multiple batches of sample data after self-calibration;

[0090] S3, taking the spectral intensity value at the inner standard line of the nth batch of sample data after self-correction and the spectral intensity value at the inner standard line of the first batch of sample data after self-correction as the quotient, and obtaining the model correction coefficient β corresponding to the nth batch of sample data n-1 ; Where n = 2, 3, 4...N, N represents the total batches of multiple batches of sample data;

[0091] S4, establishing an inter-batch calibration model using the model calibration coefficient corresponding to each batch of sample data and the spectral intensity value after self-calibration at the internal standard line; and implementing calibration between multiple batches of samples using the inter-batch calibration model;

[0092] The correction module for the predicted data is used to perform the following steps:

[0093] S1, obtaining the prediction data to be corrected;

[0094] S2, self-correcting and multi-batch correction of the predicted data to obtain the corrected predicted data;

[0095] S3, determining whether the spectral intensity at the inner standard line of the corrected prediction data is within the intensity probability interval at the inner standard line corresponding to the training data, if yes, executing S4; otherwise, executing S5;

[0096] S4, bringing the corrected prediction data into a preset calibration model for calculation;

[0097] S5, add this batch of prediction data to the training data, collect prediction data again, and then execute S3; until the spectral intensity at the inner standard line of the prediction data is within the intensity probability interval at the inner standard line corresponding to the training data, substitute the prediction data into the model calculation.

[0098] The device for reducing the uncertainty of LIBS detection between multiple batches of samples of this embodiment corresponds to the method for reducing the uncertainty of LIBS detection between multiple batches of samples of the first embodiment mentioned above; wherein, the functions implemented by each functional module in the data correction device of this embodiment correspond one-to-one to each process step in the method for reducing the uncertainty of LIBS detection between multiple batches of samples of the first embodiment mentioned above; therefore, they will not be repeated here.

[0099] Third embodiment

[0100] This embodiment provides an electronic device, which includes a processor and a memory; wherein the memory stores at least one instruction, and the instruction is loaded and executed by the processor to implement the method of the first embodiment.

[0101] The electronic device may have relatively large differences due to different configurations or performances, and may include one or more processors (central processing units, CPU) and one or more memories, wherein the memory stores at least one instruction, and the instruction is loaded by the processor to execute the above method.

[0102] Fourth embodiment

[0103] This embodiment provides a computer-readable storage medium, which stores at least one instruction, and the instruction is loaded and executed by a processor to implement the method of the first embodiment. The computer-readable storage medium may be a ROM, a random access memory, a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, etc. The instructions stored therein may be loaded by a processor in a terminal to execute the above method.

[0104] In addition, it should be noted that the present invention can be provided as a method, an apparatus or a computer program product. Therefore, the embodiments of the present invention can take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware. Moreover, the embodiments of the present invention can take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program code.

[0105] The embodiments of the present invention are described with reference to the flowcharts and / or block diagrams of the methods, terminal devices (systems), and computer program products according to the embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of the processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, an embedded processor, or other programmable data processing terminal device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing terminal device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0106] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing terminal device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 These computer program instructions can also be loaded onto a computer or other programmable data processing terminal device, so that a series of operation steps are executed on the computer or other programmable terminal device to produce a computer-implemented process, so that the instructions executed on the computer or other programmable terminal device provide for implementing the process in the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0107] It should also be noted that, in this article, the terms "include", "comprises" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or terminal device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or terminal device. In the absence of further restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, article or terminal device including the elements.

[0108] Finally, it should be noted that the above is a preferred embodiment of the present invention. It should be pointed out that although the preferred embodiment of the present invention has been described, for those skilled in the art, once the basic creative concept of the present invention is known, several improvements and modifications can be made without departing from the principles of the present invention, and these improvements and modifications should also be regarded as the protection scope of the present invention. Therefore, the attached claims are intended to be interpreted as including the preferred embodiment and all changes and modifications that fall within the scope of the embodiments of the present invention.

Claims

1. A method for reducing the uncertainty of LIBS detection between multiple batches of samples, characterized in that: include: The self-correction process within the same batch of samples, the correction process between multiple batches of samples, and the correction process for predicted data; The self-calibration process within the same batch of samples includes: S1, obtaining sample data of the same batch to be calibrated; wherein the sample data of the same batch to be calibrated is training spectral data or predicted spectral data, the training spectral data refers to spectral data used to train a preset calibration model, and the predicted spectral data refers to spectral data used to input the calibration model to achieve prediction of preset components of the sample; S2, select the inner standard line of the sample data, and extract the spectral intensity value of each sample at the inner standard line; remove the spectral intensity values ​​corresponding to the samples with the same element content by probability density, and take the average value of the remaining data as the standard spectral intensity value at the inner standard line corresponding to the sample with this element content; S3, dividing the standard spectrum intensity value by the original spectrum intensity value corresponding to the sample with the same element content to obtain the self-correction coefficient α1 of each sample; multiplying the original spectrum intensity value of the sample by its corresponding self-correction coefficient α1 to obtain the corresponding self-corrected data; The multi-batch sample calibration process includes: S1, obtaining multiple batches of sample data to be calibrated; S2, self-calibrating each batch of sample data to obtain multiple batches of sample data after self-calibration; S3, taking the spectral intensity values ​​at the inner standard line of the n batches of sample data after self-calibration and the spectral intensity values ​​at the inner standard line of the first batch of sample data after self-calibration as quotients, and obtaining n-1 model correction coefficients β1, β2, β3…β n-1 ; Where n = 2, 3, 4...N, N represents the total batches of multiple batches of sample data; S4, establishing an inter-batch calibration model using the model calibration coefficient corresponding to each batch of sample data and the spectral intensity value at the internal standard line of the batch after self-calibration; and implementing calibration between multiple batches of samples using the inter-batch calibration model; The correction process of the predicted data includes: S1, obtaining the prediction data to be corrected; S2, self-correcting and multi-batch correction of the predicted data to obtain the corrected predicted data; S3, determining whether the spectral intensity at the inner standard line of the corrected prediction data is within the intensity probability interval at the inner standard line corresponding to the training data, if yes, executing S4; otherwise, executing S5; S4, bringing the corrected prediction data into a preset calibration model for calculation; S5, add this batch of prediction data to the training data, collect prediction data again, and then execute S3; until the spectral intensity at the inner standard line of the prediction data is within the intensity probability interval at the inner standard line corresponding to the training data, substitute the prediction data into the model calculation.

2. The method for reducing LIBS detection uncertainty between multiple batches of samples according to claim 1, characterized in that: The batch calibration model is established by using the model calibration coefficient corresponding to each batch of sample data and the spectral intensity value at the internal standard line of the batch after self-calibration, including: The least squares support vector machine is used as the base learner, and the standard spectral intensity value of the nth batch of spectral data after self-correction is taken as input, and its corresponding model correction coefficient β n-1 As output; The input and output are fed into the base learner for training to obtain the batch correction model.

3. A device for reducing the uncertainty of LIBS detection between multiple batches of samples, characterized in that: include: A self-correction module within the same batch of samples, a correction module between multiple batches of samples, and a correction module for predicted data; The self-calibration module within the same batch of samples is used for: S1, obtaining sample data of the same batch to be calibrated; wherein the sample data of the same batch to be calibrated is training spectral data or predicted spectral data, the training spectral data refers to spectral data used to train a preset calibration model, and the predicted spectral data refers to spectral data used to input the calibration model to achieve prediction of preset components of the sample; S2, select the inner standard line of the sample data, and extract the spectral intensity value of each sample at the inner standard line; remove the spectral intensity values ​​corresponding to the samples with the same element content by probability density, and take the average value of the remaining data as the standard spectral intensity value at the inner standard line corresponding to the sample with this element content; S3, dividing the standard spectrum intensity value by the original spectrum intensity value corresponding to the sample with the same element content to obtain the self-correction coefficient α1 of each sample; multiplying the original spectrum intensity value of the sample by its corresponding self-correction coefficient α1 to obtain the corresponding self-corrected data; The multi-batch sample correction module is used for: S1, obtaining multiple batches of sample data to be calibrated; S2, self-calibrating each batch of sample data to obtain multiple batches of sample data after self-calibration; S3, taking the spectral intensity values ​​at the inner standard line of the n batches of sample data after self-calibration and the spectral intensity values ​​at the inner standard line of the first batch of sample data after self-calibration as quotients, and obtaining n-1 model correction coefficients β1, β2, β3…β n-1 ; Where n = 2, 3, 4...N, N represents the total batches of multiple batches of sample data; S4, establishing an inter-batch calibration model using the model calibration coefficient corresponding to each batch of sample data and the spectral intensity value at the internal standard line of the batch after self-calibration; and implementing calibration between multiple batches of samples using the inter-batch calibration model; The correction module for the predicted data is used for: S1, obtaining the prediction data to be corrected; S2, self-correcting and multi-batch correction of the predicted data to obtain the corrected predicted data; S3, determining whether the spectral intensity at the inner standard line of the corrected prediction data is within the intensity probability interval at the inner standard line corresponding to the training data, if yes, executing S4; otherwise, executing S5; S4, bringing the corrected prediction data into a preset calibration model for calculation; S5, add this batch of prediction data to the training data, collect prediction data again, and then execute S3; until the spectral intensity at the inner standard line of the prediction data is within the intensity probability interval at the inner standard line corresponding to the training data, substitute the prediction data into the model calculation.

4. The device for reducing LIBS detection uncertainty between multiple batches of samples as claimed in claim 3, characterized in that: The batch calibration model is established by using the model calibration coefficient corresponding to each batch of sample data and the spectral intensity value at the internal standard line of the batch after self-calibration, including: The least squares support vector machine is used as the base learner, and the standard spectral intensity value of the nth batch of spectral data after self-correction is taken as input, and its corresponding model correction coefficient β n-1 As output; The input and output are fed into the base learner for training to obtain the batch correction model.

Images