Adaptive calibration method and apparatus for material detection model based on libs

Abstract

The present invention relates to the technical field of laser spectroscopy analysis and provides an adaptive calibration method and apparatus for a material detection model based on LIBS. The method comprises: acquiring spectral data of a mixed material collected by a LIBS device, as well as an actual concentration of a target element in a finished sintered ore obtained after the mixed material undergoes reaction; obtaining an actual concentration of the target element in the mixed material on the basis of the actual concentration of the target element in the finished sintered ore; obtaining a predicted concentration of the target element in the mixed material on the basis of the spectral data and a material detection model; obtaining a calibration function on the basis of the actual concentration and the predicted concentration of the target element in the mixed material; and using the calibration function to calibrate the material detection model.

Description

Adaptive Calibration Method and Apparatus for LIBS-based Material Detection Model

[0001] Related cross-references

[0002] This application is based on and claims priority to Chinese Patent Application No. 202411720117.7, filed on November 28, 2024, the entire contents of which are incorporated herein by reference. Technical Field

[0003] This invention relates to the field of laser spectroscopy analysis technology, and in particular to an adaptive calibration method and apparatus for a LIBS-based material detection model. Background Technology

[0004] LIBS (Laser-Induced Breakdown Spectroscopy) is a powerful analytical technique used to determine the chemical composition of materials. LIBS involves irradiating the sample surface with high-power laser pulses to create a plasma, then collecting and analyzing the spectrum emitted by the plasma to determine the types and concentrations of elements in the sample. This technique has attracted widespread attention due to its speed, lack of complex sample preparation requirements, remote operation capability, and multi-element detection ability.

[0005] Taking the testing of a mixture as an example, after the LIBS equipment collects spectral data at the testing point of the mixture, it inputs the spectral data into a preset material detection model to analyze and obtain the types and concentrations of elements in the mixture. This material detection model requires manual sampling and testing to obtain the test results, and is constructed based on the test results and spectral data, which is a very large workload. After the material detection model is built, the LIBS equipment, during long-term use in the field, is affected by changes in field conditions, which can cause the test results to deviate. To address this issue, the current common practice is to reconstruct the material detection model through manual sampling and testing, or to combine manual sampling and testing with some algorithms for result calibration. Both of these solutions increase the workload significantly and cause inconvenience to on-site production. Furthermore, both of these solutions are proactive, making real-time calibration difficult. Summary of the Invention

[0006] In view of the above-mentioned deficiencies of the prior art, the present invention provides an adaptive calibration method and apparatus for a LIBS-based material detection model to solve the technical problem of deviation in the detection results of the material detection model.

[0007] To achieve the above and other related objectives, this invention provides an adaptive calibration method for a LIBS-based material detection model, comprising: acquiring spectral data of a mixture collected by a LIBS device, and the actual concentration of a target element in the finished sinter obtained after the mixture undergoes a reaction; obtaining the actual concentration of the target element in the mixture based on the actual concentration of the target element in the finished sinter; obtaining the predicted concentration of the target element in the mixture based on the spectral data and the material detection model; obtaining a calibration function based on the actual and predicted concentrations of the target element in the mixture; and calibrating the material detection model using the calibration function.

[0008] In one embodiment of the present invention, acquiring the spectral data of the mixture collected by the LIBS device, and the actual concentration of the target element in the finished sinter obtained after the mixture reacts, includes: acquiring the spectral data of the mixture collected by the LIBS device in a first time period; obtaining a second time period based on the size and speed of the conveying channel between the discharge point of the mixture and the collection point of the finished sinter, the spectral detection time from the discharge point to the LIBS device, and the first time period; and acquiring the actual concentration of the target element in the finished sinter in the second time period.

[0009] In one embodiment of the present invention, obtaining the predicted concentration of a target element in the mixture based on the spectral data and the material detection model includes: obtaining multiple first predicted concentrations of the target element in the mixture during the first time period based on the spectral data of the mixture during the first time period and the material detection model; obtaining the average value and variance of the multiple first predicted concentrations based on the multiple first predicted concentrations; removing first predicted concentrations that deviate significantly from the average value based on the average value and variance of the multiple first predicted concentrations; and calculating the average value of the removed first predicted concentrations to obtain the predicted concentration of the target element in the mixture.

[0010] In one embodiment of the present invention, obtaining the actual concentration of the target element in the mixture based on the actual concentration of the target element in the finished sinter includes: obtaining the actual concentration of the target element in the mixture based on the actual concentration of the target element in the finished sinter and a preset first relationship; wherein the first relationship is obtained through the following steps: obtaining the actual concentration of the target element in the finished sinter in a third time period; obtaining the actual concentration of the target element in the mixture in a fourth time period by offline analysis, wherein the mixture in the fourth time period is reacted to obtain the finished sinter in the third time period; obtaining the first relationship based on the actual concentration of the target element in the finished sinter in the third time period and the actual concentration of the target element in the mixture in the fourth time period.

[0011] In one embodiment of the present invention, obtaining the actual concentration of the target element in the mixture based on the actual concentration of the target element in the finished sinter includes: obtaining the concentration of the target element in the mixture before reaction based on the actual concentration of the target element in the finished sinter; and obtaining the actual concentration of the target element in the mixture based on the concentration of the target element in the mixture before reaction.

[0012] In one embodiment of the present invention, obtaining the concentration of the target element in the mixture before reaction based on the actual concentration of the target element in the finished sinter includes: obtaining the concentration of the target element in the mixture before reaction based on the actual concentration of the target element in the finished sinter and a preset second relationship; wherein the second relationship is obtained through the following steps: obtaining the proportion of each ingredient in the mixture; obtaining the content of other elements that react with the target element in each ingredient; and obtaining the second relationship based on the proportion and the content.

[0013] In one embodiment of the present invention, obtaining the actual concentration of the target element in the mixture based on the concentration of the target element in the mixture before reaction includes: obtaining the actual concentration of the target element in the mixture based on the concentration of the target element in the mixture before reaction and a preset third relationship; wherein, the third relationship is obtained through the following steps: obtaining the actual concentration of the target element in the finished sinter in a third time period; obtaining the actual concentration of the target element in the mixture in a fourth time period by offline analysis, wherein the mixture in the fourth time period is reacted to obtain the finished sinter in the third time period; obtaining the concentration of the target element in the mixture before reaction in the fourth time period based on the actual concentration of the target element in the finished sinter in the third time period and the second relationship; obtaining the third relationship based on the actual concentration of the target element in the mixture in the fourth time period and the concentration of the target element in the mixture before reaction in the fourth time period.

[0014] In one embodiment of the present invention, the material detection model is obtained through the following steps: acquiring the spectral data of the mixture collected by the LIBS device and the mixture sample corresponding to the spectral data; acquiring the types and concentrations of elements in the mixture sample through offline testing; and obtaining the material detection model based on the spectral data of the mixture and the types and concentrations of elements in the corresponding sample.

[0015] In one embodiment of the present invention, a calibration function is obtained based on the actual concentration and predicted concentration of the target element in the mixture, including: obtaining the calibration function based on a first set consisting of the actual concentrations of multiple target elements in the mixture within a preset time period, and a second set consisting of multiple predicted concentrations of the target element.

[0016] To achieve the above and other related objectives, the present invention also provides an adaptive calibration device for a LIBS-based material detection model, comprising: a data acquisition module for acquiring spectral data of a mixture collected by a LIBS device, and the actual concentration of a target element in the finished sinter obtained after the mixture undergoes a reaction; a first calculation module for obtaining the actual concentration of the target element in the mixture based on the actual concentration of the target element in the finished sinter; a second calculation module for obtaining the predicted concentration of the target element in the mixture based on the spectral data and the material detection model; a third calculation module for obtaining a calibration function based on the actual and predicted concentrations of the target element in the mixture; and a calibration module for calibrating the material detection model using the calibration function.

[0017] The beneficial effects of this invention are as follows: This invention proposes an adaptive calibration method and apparatus for a LIBS-based material detection model. This method analyzes the concentration of a target element in the finished sinter to deduce the actual concentration of that target element in the mixture. Simultaneously, the predicted concentration of the target element in the mixture can be obtained based on the material detection model. By comparing the actual and predicted concentrations of the target element, a calibration function can be obtained, thereby calibrating the material detection model. This process eliminates the need for manual sampling and analysis, avoids additional workload, and automatically completes the calibration using the analysis results of the finished sinter, without affecting on-site production. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 is a flowchart of a calibration method provided in an embodiment of the present invention;

[0020] Figure 2 is a flowchart of step S100 provided in an embodiment of the present invention;

[0021] Figure 3 is a flowchart of the first relationship acquisition process provided in an embodiment of the present invention;

[0022] Figure 4 is a flowchart of step S200 provided in an embodiment of the present invention;

[0023] Figure 5 is a flowchart of obtaining the second relationship according to an embodiment of the present invention;

[0024] Figure 6 is a flowchart of obtaining the third relationship according to an embodiment of the present invention;

[0025] Figure 7 is a flowchart of the material detection model acquisition process provided in an embodiment of the present invention;

[0026] Figure 8 is a flowchart of the calculation of the predicted concentration of the target element in the mixture according to an embodiment of the present invention;

[0027] Figure 9 is a schematic diagram of the structure of a calibration device provided in an embodiment of the present invention.

[0028] Explanation of reference numerals in the attached figures: 101, Data acquisition module; 102, First calculation module; 103, Second calculation module; 104, Third calculation module; 105, Calibration module. Detailed Implementation

[0029] The following specific embodiments illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. It should be noted that, unless otherwise specified, the following embodiments and features can be combined with each other. In addition to the specific methods, equipment, and materials used in the embodiments, based on the knowledge of the prior art and the description of the present invention by those skilled in the art, any prior art methods, equipment, and materials similar to or equivalent to those described in the embodiments of the present invention can be used to implement the present invention.

[0030] It should be understood that the terminology used in the embodiments of this invention is for describing specific implementations and not for limiting the scope of protection of this invention. Unless otherwise defined, all technical and scientific terms used in this invention have the same meaning as commonly understood by one of ordinary skill in the art.

[0031] In the following description, numerous details are explored to provide a more thorough explanation of embodiments of the invention. However, it will be apparent to those skilled in the art that embodiments of the invention may be practiced without these specific details. In some embodiments, well-known structures and devices are shown in block diagram form rather than in detail to avoid obscuring embodiments of the invention.

[0032] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functions, and operations that may be implemented in the methods and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0033] In this invention, based on the spectral data collected by the LIBS device, the element types and concentrations in the mixture can be analyzed using a material detection model. This concentration information includes the concentration of the target element, which can be referred to as the predicted concentration of the target element. Initially, the material detection model is relatively accurate, and the predicted concentrations generally match the actual concentrations. However, over time and due to changes in the detection environment, the predicted concentrations may deviate. Simultaneously, the mixture produces sintered ore after reaction. Since the reaction process is known, the concentration of the target element in the mixture can be deduced from the concentration of the target element in the sintered ore; this concentration can be considered the actual concentration of the target element. The core idea of ​​this invention is to calculate a calibration function based on the predicted and actual concentrations of the target element obtained through these two methods to calibrate the material detection model, thereby ensuring that the element types and concentrations obtained from the material detection model are sufficiently accurate.

[0034] It should be noted that although the concentration of elements in the mixture can be deduced from the test results of the finished sinter, this method is only applicable to some elements. For some elements, they are no longer present in the finished sinter after the reaction, and their concentration in the mixture cannot be calculated by reverse deduction. Furthermore, the concentration changes of each element before and after the reaction are different, so the calculation process for reverse deduction varies for different elements. Therefore, although reverse deduction can provide relatively accurate concentration information for some elements, it cannot detect the concentration of all elements in the mixture.

[0035] The method of obtaining predicted concentrations using a material detection model differs from this approach, as it can analyze the types of elements in the mixture and the concentration information of each element. Therefore, in this invention, only the predicted and actual concentrations of the target element are used to obtain a calibration function, and then the calibrated material detection model is used to obtain the types and concentrations of elements in the mixture. The following is a detailed explanation of this inventive concept.

[0036] Please refer to Figure 1. Figure 1 shows an adaptive calibration method for a LIBS-based material detection model provided by an embodiment of the present invention, including steps S100 to S500. Each step will be described in detail below.

[0037] Step S100: Obtain the spectral data of the mixture collected by the LIBS equipment, and the actual concentration of the target element in the finished sinter obtained after the mixture reacts.

[0038] The actual concentration of target elements in finished sinter can be obtained through techniques such as X-ray fluorescence spectroscopy (XRF), atomic absorption spectroscopy, and manual sampling and analysis. These methods can accurately determine the content of multiple elements in finished sinter.

[0039] The target element mentioned here refers to a certain element that is present in both the finished sinter and the mixture. In a specific embodiment of the present invention, the mixture is, for example, composed of fuel, dolomite, and limestone, and the target element can be calcium. The embodiments described below also use calcium as an example for detailed description. It is understood that when the materials in the mixture are of other types, the target element can also be other elements.

[0040] The aforementioned spectral data and the actual concentrations of the target elements in the finished sinter are typically collected in real time on-site and stored in a database. Therefore, in step S100, these data can be retrieved from historical data for subsequent analysis. In a specific embodiment of the present invention, for example, a calibration function can be calculated once using M days as a window, based on the spectral data and the actual concentrations of the target elements in the finished sinter during these M days. The calibration function is then calculated again in the same manner for the next window, and so on.

[0041] Please refer to Figure 2. In a specific embodiment of the present invention, step S100 includes: S110, acquiring the spectral data of the mixture collected by the LIBS device in the first time period; S120, obtaining the second time period based on the conveying channel size and speed between the feeding point of the mixture and the collection point of the finished sinter, the spectral detection time from the feeding point to the LIBS device, and the first time period; S130, acquiring the actual concentration of the target element in the finished sinter in the second time period. The reason for calculating the second time period based on the first time period is to correlate the spectral data with the actual concentration of the target element in the finished sinter. That is, the mixture corresponding to the spectral data of the first time period, after reaction, yields the finished sinter corresponding to the second time period. In this way, the spectral data and concentration data of the two have a corresponding relationship.

[0042] In a specific embodiment of the present invention, the start time of the first time period is t1 and the end time is t2, and the start time of the second time period is t3 and the end time is t4. Their relationship can be calculated as follows: (1) First, obtain the spectral detection time Δt1 from the feeding point to the LIBS device. Then, the spectral data in the [t1, t2] time period corresponds to the mixture in the [t1-Δt1, t2-Δt1] time period; (2) The conveying channel between the mixture and the finished sintered ore includes the sintering machine, the ring cooler and other conveyor belts. The speed of the sintering machine trolley is denoted as v1 and the length is denoted as l1. The speed of the ring cooler is denoted as v2 and the length is denoted as l2. The conveying time of other conveyor belts is denoted as Δt2. This value can be calculated by the length and speed of other conveyor belts. Based on these parameters, the following relationship can be obtained:

[0043] Based on the above relationship, the relationship between the first time period [t1, t2] and the second time period [t3, t4] is established. When calculating the calibration function, the estimated concentration of the target element in the mixture calculated based on the spectral data of the first time period and the actual concentration of the target element in the mixture calculated based on the actual concentration of the target element in the finished sinter of the second time period should be taken as a data pair, and the calibration function should be obtained based on one or more data pairs.

[0044] Step S200: Based on the actual concentration of the target element in the finished sinter, obtain the actual concentration of the target element in the mixture.

[0045] In this step, the reaction process of the mixture is defined, therefore there is a certain mapping relationship between the actual concentration of the target element in the finished sinter and the actual concentration of the target element in the mixture. Therefore, the actual concentration of the target element in the mixture can be calculated based on the actual concentration of the target element in the finished sinter and this mapping relationship.

[0046] Please refer to Figure 3. In a specific embodiment of the present invention, step S200 includes: obtaining the actual concentration of the target element in the mixture based on the actual concentration of the target element in the finished sinter and a preset first relationship. The first relationship is obtained through the following steps:

[0047] S201. Obtain the actual concentration of the target element in the finished sinter during the third time period. S202. Obtain the actual concentration of the target element in the mixture during the fourth time period through offline analysis. Wherein, the mixture in the fourth time period, after reaction, yields the finished sinter in the third time period. S203. Based on the actual concentrations of the target element in the finished sinter during the third time period and the actual concentrations of the target element in the mixture during the fourth time period, obtain the first relationship.

[0048] In this step, the primary relationship reflects the concentration correspondence of the target element in the mixture and the finished sinter before and after the reaction. This primary relationship is fixed and can be obtained in an experimental environment according to steps S201-S203, and the obtained primary relationship is saved as a preset known quantity. Considering that some parameters of the on-site equipment may affect the primary relationship during actual production, the data in steps S201 and S202 are best obtained from the actual production scenario, so that the constructed primary relationship is more accurate.

[0049] Please refer to Figure 4. In a specific embodiment of the present invention, step S200 includes: S210, obtaining the concentration of the target element in the mixture before reaction based on the actual concentration of the target element in the finished sinter; S220, obtaining the actual concentration of the target element in the mixture based on the concentration of the target element in the mixture before reaction. In the previous embodiment, a first relationship between the actual concentration of the target element in the finished sinter and the actual concentration of the target element in the mixture was directly established. This embodiment is different; a mapping relationship between the actual concentration of the target element in the finished sinter and the actual concentration of the target element in the mixture is established through two steps, S210 and S220. Steps S210 and S220 will be described in detail below.

[0050] Please refer to Figure 5. Step S210 includes: obtaining the concentration of the target element in the mixture before reaction based on the actual concentration of the target element in the finished sinter and the preset second relationship. The second relationship is obtained through steps S211 to S213.

[0051] Step S211: Obtain the proportion of each ingredient in the mixture.

[0052] The proportions of each material in the mixture are measured periodically by specialized equipment. Therefore, the measurement results from this equipment can be directly obtained in this step, thus yielding the proportion data. For example, in a real production scenario, the proportions of fuel, dolomite, and limestone in the mixture are P1, P2, and P3.

[0053] Step S212: Obtain the content of other elements that react with the target element in each ingredient.

[0054] Taking calcium as the target element as an example, other elements that react with the target element include carbon and sulfur. Therefore, this step requires obtaining the carbon and sulfur content of each ingredient. Using the above scenario as an example, let's decompose the carbon and sulfur content in the fuel into C1 and S1, the carbon content in dolomite into C2, and the carbon content in limestone into C3.

[0055] Step S213: Based on the ratio and content, obtain the second relationship.

[0056] This step mainly involves constructing the second relationship by combining the reaction relationship of the mixture. Taking the above scenario as an example, we can obtain the following relationship: Ca1=(1-(P1*(C1+S1)+P2*C2+P3*C3))*Ca2, where Ca1 is the concentration of the target element in the mixture before the reaction, and Ca2 is the actual concentration of the target element in the finished sinter. This relationship is the second relationship.

[0057] Please refer to Figure 6. In a specific embodiment of the present invention, step S220 includes: obtaining the actual concentration of the target element in the mixture based on the concentration of the target element before reaction and a preset third relationship. The third relationship is obtained through the following steps:

[0058] S221. Obtain the actual concentration of the target element in the finished sinter during the third time period; S222. Obtain the actual concentration of the target element in the mixture during the fourth time period through offline analysis, wherein the mixture in the fourth time period is reacted to obtain the finished sinter during the third time period; S223. Based on the actual concentration of the target element in the finished sinter during the third time period and the second relationship, obtain the concentration of the target element in the mixture before the reaction in the fourth time period; S224. Based on the actual concentration of the target element in the mixture during the fourth time period and the concentration of the target element in the mixture before the reaction in the fourth time period, obtain the third relationship.

[0059] Theoretically, the Ca1 calculated from Ca2 in step S213 is actually equal to the actual concentration of the target element in the mixture. However, considering factors such as the accuracy of the scale, the feeding speed, the level of testing, and the actual proportion in actual production, there is a certain difference between the calculated Ca1 and the actual concentration of the target element in the mixture. Therefore, a third relationship can be obtained by fitting historical data.

[0060] The third and fourth time periods mentioned in the above steps are to distinguish them from the first and second time periods. Essentially, they are both for the purpose of mapping the data. The mapping relationship between the third and fourth time periods can be calculated based on the size and speed of the conveying channel between the feeding point of the mixture and the collection point of the finished sintered ore.

[0061] In the above steps, two calculation methods for step S200 are proposed, whereby the actual concentration of calcium in the finished sinter in the third time period is Ca2, and the actual concentration of calcium in the mixture in the fourth time period is Ca0.

[0062] The first relationship is the relationship between Ca2 and Ca0, which can be represented by the function F(). Then we can get: Ca0=F(Ca2). By substituting Ca2 into this formula, we can get the corresponding Ca0.

[0063] The second and third relations are decompositions of the first relation and introduce an intermediate value Ca1. For example, in the above embodiment, the second relation is expressed by the formula: Ca1=(1-(P1*(C1+S1)+P2*C2+P3*C3))*Ca2, and the third relation is expressed by the function G(): Ca0=G(Ca1). Combining them, we can get: Ca0=G((1-(P1*(C1+S1)+P2*C2+P3*C3))*Ca2). Substituting Ca2 into this formula, we can obtain the corresponding Ca0.

[0064] Understandably, the second relation will differ for different target elements and reaction relationships; the third relation is deterministic for a fixed scenario; and the first relation is deterministic for a fixed scenario and the same reaction process. These relations only need to be calculated and saved once, and can be directly referenced later.

[0065] Step S300: Based on the spectral data and the material detection model, obtain the predicted concentration of the target element in the mixture.

[0066] The purpose of calculating the predicted concentration of target elements in the mixture based on the material detection model is to facilitate the construction of calibration functions.

[0067] Please refer to Figure 7. In a specific embodiment of the present invention, the material detection model is obtained through the following steps: S311, acquiring the spectral data of the mixture collected by the LIBS device and the corresponding mixture sample; S312, acquiring the types and concentrations of elements in the mixture sample through offline analysis; S313, obtaining the material detection model based on the spectral data of the mixture and the types and concentrations of elements in the corresponding sample. The material detection model can calculate the types and concentrations of all elements based on the spectral data, but in this step, only the concentration of the target element is used, and it is used as the predicted concentration of the target element in the mixture.

[0068] In step S200, the actual concentration of the target element in the mixture needs to be obtained based on the actual concentration of the target element in the finished sinter during the second time period. In step S300, the predicted concentration of the target element in the mixture needs to be obtained based on the spectral data from the first time period and the material detection model. Since the time interval between the first and second time periods can be long or short, multiple actual concentrations of the target element in the finished sinter may be collected during the second time period, and the LIBS device may collect multiple spectral data during the first time period. Therefore, it is necessary to process multiple data into one data.

[0069] Please refer to Figure 8. In a specific embodiment of the present invention, step S300 is taken as an example. Step S300 includes: S321 obtaining multiple first predicted concentrations of the target element in the mixture during the first time period based on the spectral data of the mixture in the first time period and the material detection model, denoted as (a1, a2, ..., a...). n S322. Based on multiple first predicted concentrations, obtain the average value of the multiple first predicted concentrations. And variance σ, where the mean and variance can be calculated according to specific formulas; S323, based on the mean and variance of multiple first predicted concentrations, remove the first predicted concentrations that deviate significantly from the mean, for example, remove (a1, a2, ..., a n (smaller than) And greater than The value; S324, calculate the average value of the first predicted concentration after elimination to obtain the predicted concentration of the target element in the mixture in the first time period. This result is a numerical value.

[0070] Step S400: Obtain the calibration function based on the actual and predicted concentrations of the target elements in the mixture.

[0071] In a specific embodiment of the present invention, step S400 includes: obtaining a calibration function based on a first set consisting of the actual concentrations of target elements in multiple mixtures within a preset time period, and a second set consisting of multiple predicted concentrations of target elements. In a practical scenario, the preset time period is, for example, M days. Based on the spectral data of the first time period, the predicted concentration of the target element in the mixture during the first time period is denoted as A. i Based on the actual concentration of the target element in the finished sinter during the second time period, the actual concentration of the target element in the mixture during the first time period is obtained and denoted as B. i The subscript i indicates the correspondence between the two. Based on M days of historical data, the first set (A1, A2, ..., A...) can be obtained. m ) and the second set (B1, B2, ..., B m ), where m is the total number of data pairs calculated within M days.

[0072] Then, based on (A1, A2, ..., A...), we can... m ) and (B1,B2,…,B m The calibration function J(θ) is obtained. When the actual concentration of the target element in the finished sinter sample is obtained by sampling and offline testing, there will be some outliers due to the representativeness of the sample. Therefore, the calibration function J(θ) can be solved by robust regression.

[0073] Step S500: Calibrate the material detection model using a calibration function.

[0074] Substituting the calibration function J(θ) obtained above into the material detection model completes one model calibration. The calibrated model will provide more accurate information on element types and concentrations when processing spectral data.

[0075] It should be noted that the steps of the various methods described above are only for clarity. In practice, they can be combined into one step or some steps can be split into multiple steps. As long as they contain the same logical relationship, they are all within the scope of protection of this patent. Adding insignificant modifications or introducing insignificant designs to the algorithm or process, but without changing the core design of the algorithm and process, are also within the scope of protection of this patent.

[0076] Based on practical scenarios, the calibration steps can be divided into the following three steps: (1) For a new scenario, first construct the first relationship (or construct the second and third relationships) and save it; (2) Each time calibration is required, obtain the historical data of the previous M days, which includes the spectral data collected by the LIBS device and the concentration data of the target element in the finished sinter; (3) Establish the correspondence between the spectral data and the concentration data to obtain the first set (A1, A2, ..., A m) and the second set (B1, B2, ..., B m Then, based on these two sets, the calibration function J(θ) is obtained, and the material detection model is calibrated. When subsequent calibration is needed, steps (2) and (3) can be repeated. These two steps can be completed automatically without manual intervention. Therefore, calibration can be performed automatically during the subsequent use of the equipment.

[0077] Please refer to Figure 9. Figure 9 shows an adaptive calibration device for a LIBS-based material detection model according to an embodiment of the present invention, including a data acquisition module 101, a first calculation module 102, a second calculation module 103, a third calculation module 104, and a calibration module 105. The data acquisition module 101 acquires the spectral data of the mixture collected by the LIBS device, and the actual concentration of the target element in the finished sinter obtained after the mixture reacts. The first calculation module 102 calculates the actual concentration of the target element in the mixture based on the actual concentration of the target element in the finished sinter. The second calculation module 103 calculates the predicted concentration of the target element in the mixture based on the spectral data and the material detection model. The third calculation module 104 calculates a calibration function based on the actual and predicted concentrations of the target element in the mixture. The calibration module 105 calibrates the material detection model using the calibration function.

[0078] It should be noted that the adaptive calibration device in this embodiment corresponds to the adaptive calibration method described above, and the functional modules in the adaptive calibration device may correspond to the corresponding steps in the adaptive calibration method. The adaptive calibration device in this embodiment can be implemented in conjunction with the adaptive calibration method; that is, where there is no conflict, the relevant technical details mentioned in the adaptive calibration method of the above embodiment can also be applied to the adaptive calibration device in this embodiment.

[0079] The above embodiments are merely illustrative of the principles and effects of the present invention and are not intended to limit the invention. Any person skilled in the art can modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or alterations made by those skilled in the art without departing from the spirit and technical concept disclosed in the present invention should still be covered by the claims of the present invention.

Claims

1. An adaptive calibration method for a LIBS-based material detection model, characterized in that, include: Acquire the spectral data of the mixture collected by the LIBS device, and the actual concentration of the target element in the finished sinter obtained after the mixture reacts; The actual concentration of the target element in the mixture is obtained based on the actual concentration of the target element in the finished sinter. Based on the spectral data and the material detection model, the predicted concentration of the target element in the mixture is obtained; The calibration function is obtained based on the actual and predicted concentrations of the target elements in the mixture; The material detection model is calibrated using the calibration function.

2. The adaptive calibration method for the LIBS-based material detection model according to claim 1, characterized in that, Acquiring spectral data of the mixture collected by the LIBS device, and the actual concentration of the target element in the finished sinter obtained after the mixture reacts, including: Acquire the spectral data of the mixture collected by the LIBS device in the first time period; The second time period is obtained based on the size and speed of the conveying channel between the feeding point of the mixture and the collection point of the finished sinter, the spectral detection time from the feeding point to the LIBS device, and the first time period; The actual concentration of the target element in the finished sintered ore during the second time period is obtained.

3. The method of adaptive calibration of LIBS-based material detection models according to claim 2, wherein, Based on the spectral data and the material detection model, the predicted concentration of the target element in the mixture is obtained, including: Based on the spectral data of the mixture during the first time period and the material detection model, multiple first predicted concentrations of the target elements in the mixture during the first time period are obtained. Based on the plurality of first predicted concentrations, the average value and variance of the plurality of first predicted concentrations are obtained; Based on the average and variance of the plurality of first predicted concentrations, the first predicted concentrations that deviate significantly from the average are removed. The first predicted concentrations that deviate significantly from the average are defined as those whose absolute value of the difference from the average is greater than three times the variance. The average value of the first predicted concentration after elimination is used to obtain the predicted concentration of the target element in the mixture.

4. The adaptive calibration method for a LIBS-based material detection model according to any one of claims 1 to 3, characterized in that, The actual concentration of the target element in the mixture is obtained based on the actual concentration of the target element in the finished sinter, including: The actual concentration of the target element in the mixture is obtained based on the actual concentration of the target element in the finished sinter and the preset first relationship; The first relationship is obtained through the following steps: Obtain the actual concentration of the target element in the finished sinter during the third time period; The actual concentration of the target element in the mixture during the fourth time period was obtained by offline testing, and the mixture during the fourth time period was reacted to obtain the finished sintered ore during the third time period. The first relationship is obtained based on the actual concentration of the target element in the finished sinter during the third time period and the actual concentration of the target element in the mixture during the fourth time period.

5. The adaptive calibration method for a LIBS-based material detection model according to any one of claims 1 to 3, characterized in that, The actual concentration of the target element in the mixture is obtained based on the actual concentration of the target element in the finished sinter, including: Based on the actual concentration of the target element in the finished sinter, the concentration of the target element in the mixture before reaction is obtained; The actual concentration of the target element in the mixture is obtained based on the concentration of the target element before it reacts.

6. The adaptive calibration method for the LIBS-based material detection model according to claim 5, characterized in that, Based on the actual concentration of the target element in the finished sinter, the concentration of the target element in the mixture before reaction is obtained, including: Based on the actual concentration of the target element in the finished sinter and the preset second relationship, the concentration of the target element in the mixture before reaction is obtained; The second relationship is obtained through the following steps: Obtain the proportion of each ingredient in the mixture; Obtain the content of other elements that react with the target element in each of the ingredients; The second relationship is obtained based on the ratio and the content.

7. The adaptive calibration method for the LIBS-based material detection model according to claim 6, characterized in that, The actual concentration of the target element in the mixture is obtained based on the concentration of the target element before reaction, including: Based on the concentration of the target element in the mixture before reaction and the preset third relationship, the actual concentration of the target element in the mixture is obtained; The third relationship is obtained through the following steps: Obtain the actual concentration of the target element in the finished sinter during the third time period; The actual concentration of the target element in the mixture during the fourth time period was obtained by offline testing, and the mixture during the fourth time period was reacted to obtain the finished sintered ore during the third time period. Based on the actual concentration of the target element in the finished sinter during the third time period and the second relationship, the concentration of the target element in the mixture before unreacted reaction during the fourth time period is obtained; The third relationship is obtained based on the actual concentration of the target element in the mixture during the fourth time period and the concentration of the target element in the mixture before the reaction during the fourth time period.

8. The adaptive calibration method for a LIBS-based material detection model according to any one of claims 1 to 7, characterized in that, The material detection model is obtained through the following steps: Acquire the spectral data of the mixture collected by the LIBS device and the mixture sample corresponding to the spectral data; The types and concentrations of elements in the mixture sample were obtained through offline testing. The material detection model is obtained based on the spectral data of the mixture and the types and concentrations of elements in the corresponding samples.

9. The adaptive calibration method for a LIBS-based material detection model according to any one of claims 1 to 7, characterized in that, Based on the actual and predicted concentrations of the target elements in the mixture, a calibration function is obtained, including: The calibration function is obtained based on a first set consisting of the actual concentrations of the target elements in multiple mixtures within a preset time period, and a second set consisting of multiple predicted concentrations of the target elements.

10. An adaptive calibration device for a LIBS-based material detection model, characterized in that, include: The data acquisition module is used to acquire the spectral data of the mixture collected by the LIBS equipment, as well as the actual concentration of the target element in the finished sinter obtained after the mixture reacts. The first calculation module is used to obtain the actual concentration of the target element in the mixture based on the actual concentration of the target element in the finished sinter. The second calculation module is used to obtain the predicted concentration of the target element in the mixture based on the spectral data and the material detection model. The third calculation module is used to obtain a calibration function based on the actual and predicted concentrations of the target elements in the mixture; A calibration module is used to calibrate the material detection model using the calibration function.

Images