In-situ detection method and system for lead-zinc tailing aggregate based on LIBS technology

Abstract

The application provides a lead-zinc tailing aggregate in-situ detection method and system based on LIBS technology, and relates to the field of mineral element detection. The method comprises the following steps: optimizing and correcting the reference proofing point quantity according to the Q sample detection complexity indexes to obtain Q adaptive proofing point quantities; based on the to-be-detected sample sequence and the Q adaptive proofing point quantities, in-situ detection is sequentially performed on the Q to-be-detected samples by using the LIBS technology, and Q in-situ detection results are output, wherein, in the in-situ detection process, the adaptive proofing point quantity of the subsequent to-be-detected sample is secondarily optimized according to the consistency of the historical detection results. The technical problem that the existing detection method generally adopts a fixed rule for detection, the detection efficiency and precision are difficult to balance, and the detection demand of batch, on-site and rapidization in highway engineering cannot be met is solved; the efficiency of the lead-zinc tailing aggregate on-site in-situ detection can be significantly improved under the premise of ensuring the detection precision.

Description

Technical Field

[0001] This invention relates to the field of mineral element detection, and in particular to an in-situ detection method and system for lead-zinc tailings aggregates based on LIBS technology. Background Technology

[0002] Lead-zinc tailings are solid waste remaining after lead-zinc ore beneficiation. Their resource utilization in highway pavement aggregates can both dispose of solid waste and reduce environmental pressure, while also reducing engineering material costs. However, lead-zinc tailings have a complex composition, containing harmful mineral elements such as lead, zinc, and sulfur. Furthermore, tailings from different sources and batches exhibit significant differences in moisture content, surface roughness, and mineral uniformity. Direct application without screening may affect pavement performance and environmental safety.

[0003] Currently, existing laser-induced breakdown spectroscopy (LIBS) detection methods for batch on-site testing of lead-zinc tailings aggregates typically use a fixed number of sampling points, and the setting of detection parameters mainly relies on empirical values ​​under ideal laboratory conditions. They lack an effective adaptive adjustment mechanism for changes in on-site environmental factors and compositional differences between different batches of samples, making it difficult to balance accuracy and efficiency in the detection process, and the overall detection effect needs further improvement. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for in-situ detection of lead-zinc tailings aggregates based on LIBS technology. This addresses the technical problem that existing LIBS detection methods typically employ fixed rules, resulting in a trade-off between detection efficiency and accuracy, and failing to meet the demands of batch, on-site, and rapid detection in highway engineering projects. The method includes:

[0005] In a first aspect, the present invention provides an in-situ detection method for lead-zinc tailings aggregate based on LIBS technology, comprising: obtaining the number of benchmark sampling points under standard detection conditions based on the detection error index mapping of lead-zinc tailings aggregate; collecting Q raw material attribute information of Q samples to be tested from the lead-zinc tailings aggregate, and monitoring and acquiring the current detection environment parameters; using the standard detection conditions as a benchmark, performing sample detection complexity analysis based on the Q raw material attribute information and the current detection environment parameters, obtaining Q sample detection complexity indices, and generating a sample detection complexity index by sorting the samples according to the detection complexity index from smallest to largest. The sample sequence is tested; the number of baseline sampling points is optimized and corrected according to the Q sample detection complexity indices to obtain Q suitable sampling points, wherein the number of suitable sampling points is the floor value of the product of the sample detection complexity index and the number of baseline sampling points; based on the sample sequence to be tested and the Q suitable sampling points, LIBS technology is used to sequentially perform in-situ detection on the Q samples to be tested, and output Q in-situ detection results, wherein, during the in-situ detection process, the number of suitable sampling points for subsequent samples to be tested is further optimized based on the consistency of historical detection results.

[0006] Secondly, this invention also provides an in-situ detection system for lead-zinc tailings aggregate based on LIBS technology, comprising: a benchmark sampling point acquisition module, used to acquire the number of benchmark sampling points under standard detection conditions based on the detection error index mapping of lead-zinc tailings aggregate; a current detection information acquisition module, used to acquire Q raw material attribute information of Q samples to be tested from the lead-zinc tailings aggregate, and to monitor and acquire current detection environment parameters; and a sample detection complexity analysis module, used to perform sample detection complexity analysis based on the standard detection conditions, according to the Q raw material attribute information and the current detection environment parameters, to acquire Q sample detection complexity indices, and to analyze the sample detection complexity according to the sample detection complexity indices. A sequence of samples to be tested is generated by sorting them from smallest to largest. A module for generating the number of suitable sampling points is used to optimize and correct the number of baseline sampling points based on the detection complexity indices of the Q samples to obtain Q suitable sampling points, where the number of suitable sampling points is the floor value of the product of the detection complexity index and the number of baseline sampling points. An in-situ sample detection module is used to perform in-situ detection on the Q samples to be tested sequentially using LIBS technology, based on the sequence of samples to be tested and the number of Q suitable sampling points, and output Q in-situ detection results. During the in-situ detection process, the number of suitable sampling points for subsequent samples to be tested is further optimized based on the consistency of historical detection results.

[0007] The embodiments of the present invention have the following advantages: By constructing a detection complexity assessment model, using standard detection conditions as a benchmark, and calculating the sample detection complexity index in real time based on current detection environment parameters and raw material properties, the number of benchmark sampling points is optimized and corrected based on this index. This ensures that the number of sampling points matches the actual complexity of the sample, thereby avoiding over-detection of simple samples and under-sampling of complex samples while ensuring detection accuracy, and reducing invalid detection steps. Furthermore, during batch detection, this invention dynamically adjusts the number of appropriate sampling points for the next sample based on the consistency coefficient of the detection results of the previous sample, forming a continuous optimization closed loop. This fully utilizes the detection information of previous samples to guide subsequent detections, further improving the overall efficiency of batch detection. Through the above technical means, this invention can achieve synergistic optimization of accuracy and efficiency in the batch detection of lead-zinc tailings aggregates, achieving significant improvements in detection efficiency, reduction in detection costs, and guarantee of representativeness of detection results, meeting the practical application needs of rapid screening in highway pavement engineering. Attached Figure Description

[0008] Figure 1 This is a flowchart illustrating the steps of the in-situ detection method for lead-zinc tailings aggregates based on LIBS technology according to the present invention. Figure 2 This is a schematic diagram of the in-situ detection system for lead-zinc tailings aggregates based on LIBS technology according to the present invention.

[0009] Explanation of reference numerals in the attached figures: The module includes: 11 for obtaining the number of reference sampling points; 12 for collecting current detection information; 13 for analyzing the complexity of sample detection; 14 for generating the number of adapted sampling points; and 15 for in-situ sample detection. Detailed Implementation

[0010] The technical solutions of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. It should be understood that the present invention is not limited to the exemplary embodiments described herein. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention. It should also be noted that, for ease of description, only the parts related to the present invention are shown in the accompanying drawings, not all of them.

[0011] Example 1, please refer to the appendix. Figure 1 This invention provides an in-situ detection method for lead-zinc tailings aggregates based on LIBS technology, specifically including the following steps: S100: Based on the detection error index mapping of lead-zinc tailings aggregate, obtain the number of benchmark sampling points for LIBS detection under standard detection conditions.

[0012] Furthermore, step S100 of the present invention further includes: Obtain the detection error index of lead-zinc tailings aggregate required for road construction; obtain the instrument model of the current LIBS testing method; based on the mapping between the detection error index and the instrument model, obtain the minimum number of sampling points under standard testing conditions for the LIBS testing method, as the benchmark number of sampling points, wherein the standard testing conditions include standard testing environment parameters and standard sample raw material property information.

[0013] Specifically, the first step is to obtain the required testing error index for lead-zinc tailings aggregate testing as specified in the road construction specifications or design documents. For example, the relative standard deviation (RSD) for sulfur content testing should not exceed 10%, and the RSD for zinc content testing should not exceed 8%. This index measures the accuracy requirements of the test results. Simultaneously, the specific model of the LIBS testing instrument currently in use should be obtained, such as a handheld LIBS analyzer (model A) or a benchtop LIBS analyzer (model B). Because different instrument models have inherent differences in laser energy, spectral resolution, and detector sensitivity, the minimum number of sampling points required for the same testing error index varies depending on the instrument model. By simultaneously obtaining the testing error index and the instrument model, the minimum number of sampling points required to meet the error requirement under standard testing conditions can be mapped to the specific instrument being used. This serves as a baseline sampling point quantity, eliminating the influence of instrument differences on the test results and providing a unified and reliable benchmark for subsequent adaptive sampling.

[0014] Next, a mapping database is pre-constructed, which records the correspondence between different instrument models, different detection error requirements, and the minimum number of sampling points under standard detection conditions. The mapping database is constructed as follows: at least one model of LIBS testing instrument is selected, and repeated testing experiments are performed on standard samples under standard detection conditions. The relative standard deviation of the test results under different numbers of sampling points is calculated. The minimum number of sampling points that meets the preset detection error requirements is taken as the benchmark number of sampling points for that instrument model under that error requirement. The above calibration experiment is performed on multiple different models of LIBS testing instruments to form a mapping table covering multiple instrument models and multiple error requirements.

[0015] During actual testing, based on the currently acquired testing error index and the instrument model of the LIBS testing instrument currently being used, the corresponding number of benchmark sampling points is quickly obtained by querying the mapping relationship database. For example, when road construction requires the relative standard deviation of sulfur content detection in lead-zinc tailings aggregate to be no more than 10%, and a handheld LIBS instrument of model A is currently being used, the number of benchmark sampling points automatically mapped and obtained is 20; if a benchtop LIBS instrument of model B is currently being used, the number of benchmark sampling points mapped and obtained is 8. It should be noted that the standard testing conditions include standard testing environmental parameters and standard sample raw material property information. Among them, standard testing environmental parameters refer to the pre-set ideal environmental conditions, such as an ambient temperature of 25℃±2℃, an ambient relative humidity of 40%±5%, and an ambient dust concentration of less than 0.1mg / m³; standard sample raw material property information refers to the property parameters of the standard sample after homogenization treatment, such as a standard sample moisture content of less than 0.5%, good surface smoothness, and a mineral composition homogeneity index close to 1.0. By using standardized testing conditions, we can ensure that the number of benchmark sampling points obtained under different instruments and different error requirements is comparable and consistent.

[0016] Furthermore, step S100 of the present invention further includes: The environmental parameters to be tested include ambient temperature, ambient humidity and ambient dust concentration. The raw material property information includes raw material moisture content, raw material surface roughness and raw material mineral composition uniformity index. The calculation steps of the raw material mineral composition uniformity index include: collecting LIBS spectra of N pre-sampled points at preset positions on the surface of the sample to be tested, and calculating the relative standard deviation of the intensity of the characteristic spectral lines of the target element as the raw material mineral composition uniformity index.

[0017] Specifically, the detection environment parameters include ambient temperature, ambient humidity, and ambient dust concentration. Ambient temperature affects the stability of laser output and plasma cooling rate; deviations from the standard temperature will cause spectral line intensity drift. Ambient humidity affects laser energy transmission and spectral signal intensity; when humidity is too high, water molecules will absorb laser energy and weaken characteristic spectral lines. Ambient dust concentration affects the transmission quality of laser and spectral signals; excessive dust will cause signal attenuation and introduce background noise. The above three environmental parameters together reflect the impact of external detection conditions on the stability of LIBS detection and are important environmental factors for assessing detection complexity.

[0018] The raw material properties include moisture content, surface roughness, and mineral composition uniformity index. Moisture content affects laser ablation efficiency and spectral quality; excessive moisture absorbs laser energy, leading to a decrease in spectral line intensity. Surface roughness affects laser focusing accuracy and ablation stability; excessively rough or smooth surfaces both affect detection reproducibility. The mineral composition uniformity index quantifies the spatial distribution uniformity of different mineral phases in the sample. It is obtained by calculating the relative standard deviation of the characteristic spectral line intensity of the target element using LIBS spectra from multiple pre-sampled points. A lower index indicates a less uniform sample, requiring more sampling points to obtain representative detection results. These three raw material properties collectively reflect the influence of the sample's inherent characteristics on the complexity of the detection process.

[0019] The raw material mineral composition uniformity index is used to quantify the spatial uniformity of different mineral phases in lead-zinc tailings aggregate samples. The calculation steps are as follows: First, LIBS spectra of N pre-sampling points are collected on the surface of the sample to be tested according to the preset sampling positions. The preset positions usually include the central area and the four corner areas of the sample to cover the main distribution range of the sample surface. The value of N is preferably 5 to 9 points, which can reflect the overall uniformity of the sample without consuming too much detection time. Each pre-sampling point independently collects a LIBS spectrum and records the intensity value of the characteristic spectral lines of the target element (such as lead, zinc, sulfur, etc.). Then, the relative standard deviation of the characteristic spectral line intensity of the target element corresponding to the above N pre-sampling points is calculated. The formula for calculating the relative standard deviation is: RSD = (standard deviation / mean) × 100%, where the standard deviation reflects the dispersion of the spectral line intensity of each sampling point, and the mean reflects the overall level of the spectral line intensity. The calculated relative standard deviation is used as the quantitative value of the uniformity index of the raw material mineral composition. The smaller the index, the closer the element spectral line intensity at different positions on the sample surface is, the more uniform the mineral composition distribution is, the better the representativeness of the detection results, and the lower the detection complexity. Conversely, the larger the index, the more uneven the mineral distribution in the sample, the existence of local enrichment or depletion, the more sampling points are needed to obtain statistically representative detection results, and the higher the detection complexity.

[0020] S200: Collect the raw material attribute information of Q samples of lead-zinc tailings aggregate to be tested, and monitor and obtain the current testing environment parameters.

[0021] Specifically, before starting batch testing, it is necessary to collect the raw material property information of Q samples of lead-zinc tailings aggregate to be tested. The raw material property information of each sample includes the raw material moisture content, raw material surface roughness, and raw material mineral composition uniformity index. The raw material moisture content can be obtained through a rapid moisture analyzer or indirectly estimated through the intensity of hydrogen and oxygen spectral lines in LIBS spectroscopy. The raw material surface roughness can be measured by a surface roughness meter or evaluated through the stability of LIBS pre-scanning spectra. The raw material mineral composition uniformity index is obtained by calculating the relative standard deviation of the characteristic spectral line intensity of the target element using the aforementioned method, by collecting LIBS spectra from N pre-sampling points. The above collection steps are performed on the Q samples to be tested to obtain Q sets of raw material property information. On the other hand, it is necessary to monitor and acquire the current testing environment parameters in real time, including ambient temperature, ambient humidity, and ambient dust concentration. These parameters can be collected in real time by sensors installed at the testing site. For example, ambient temperature is acquired through a temperature sensor, ambient relative humidity through a humidity sensor, and the concentration of suspended particulate matter in the air through a dust concentration sensor. Environmental parameters are typically recorded at the start of each batch of testing, and can be dynamically updated if environmental conditions change significantly during the testing process. The aforementioned raw material properties and environmental parameters together constitute the input data for sample testing complexity analysis, providing a basis for subsequently calculating the testing complexity index of each sample.

[0022] S300: Based on the standard testing conditions, perform sample testing complexity analysis according to the Q raw material attribute information and the current testing environment parameters, obtain Q sample testing complexity indices, and generate a sequence of samples to be tested by sorting the samples according to the sample testing complexity indices from smallest to largest.

[0023] Furthermore, step S300 of the present invention further includes: Based on the standard detection environment parameters, an environment-detection complexity analysis is performed according to the current detection environment parameters, and the environment-detection complexity coefficient is output.

[0024] Furthermore, the steps of the present invention also include: Using the standard detection environment parameters as a benchmark, the parameter deviations between the current detection environment parameters and the standard detection environment parameters are calculated to obtain temperature deviation, humidity deviation, and dust concentration deviation. The temperature deviation is multiplied by a preset temperature sensitivity coefficient and then incremented by one to obtain the temperature deviation coefficient. The humidity deviation is multiplied by a preset humidity sensitivity coefficient and then incremented by one to obtain the humidity deviation coefficient. The dust concentration deviation is multiplied by a preset dust concentration sensitivity coefficient and then incremented by one to obtain the dust concentration deviation coefficient. The temperature deviation coefficient, humidity deviation coefficient, and dust concentration deviation coefficient are weighted and calculated to obtain the environment-detection complexity coefficient.

[0025] Specifically, when calculating the environmental-detection complexity coefficient, it is first necessary to use standard detection environmental parameters as a benchmark and calculate the deviation between the current detection environmental parameters and the standard detection environmental parameters. The standard detection environmental parameters are pre-set ideal environmental conditions, such as a standard ambient temperature of 25℃, a standard ambient humidity of 40%RH, and a standard ambient dust concentration of 0.1mg / m³. Assuming the current measured ambient temperature is 28℃, the temperature deviation is 28℃ minus 25℃, which equals 3℃; the current measured ambient humidity is 55%RH, which equals 55%RH minus 40%RH, which equals 15%RH; and the current measured ambient dust concentration is 0.3mg / m³, which equals 0.3mg / m³ minus 0.1mg / m³, which equals 0.2mg / m³. By calculating the deviation, the degree to which the current environmental conditions deviate from the standard environment can be quantified.

[0026] Next, after obtaining the deviations, the temperature deviation coefficient, humidity deviation coefficient, and dust concentration deviation coefficient are calculated separately. Each deviation coefficient is calculated by multiplying the corresponding deviation by a preset sensitivity coefficient and then adding one. The sensitivity coefficient reflects the degree of influence of each environmental parameter on the detection complexity and can be obtained through prior experimental calibration. For example, if the preset temperature sensitivity coefficient is 0.02 / ℃, then the temperature deviation coefficient equals 1 plus 0.02 multiplied by 3, resulting in 1.06; if the preset humidity sensitivity coefficient is 0.01 / %RH, then the humidity deviation coefficient equals 1 plus 0.01 multiplied by 15, resulting in 1.15; and if the preset dust concentration sensitivity coefficient is 0.5 / (mg / m³), then the dust concentration deviation coefficient equals 1 plus 0.5 multiplied by 0.2, resulting in 1.10. This calculation method ensures that when the environmental parameters are consistent with the standard conditions, the deviation is zero, and the corresponding deviation coefficient is 1, without any additional impact on the detection complexity. When the environmental parameters deviate from the standard conditions, the deviation coefficient is greater than 1; the greater the deviation, the larger the coefficient, and the higher the detection complexity.

[0027] Finally, the temperature deviation coefficient, humidity deviation coefficient, and dust concentration deviation coefficient are weighted and calculated to obtain the environmental-detection complexity coefficient. The weight coefficients of each environmental parameter can be preset according to their influence on the detection results. For example, if the weight of temperature is set to 0.25, humidity to 0.40, and dust concentration to 0.35, then the environmental-detection complexity coefficient is equal to 0.25 multiplied by 1.06 plus 0.40 multiplied by 1.15 plus 0.35 multiplied by 1.10, resulting in a coefficient of 1.11. A coefficient greater than 1 indicates that the current environmental conditions are more complex than the standard conditions, requiring a corresponding increase in the number of baseline sampling points; a coefficient equal to 1 indicates that the environmental conditions are consistent with the standard conditions, requiring no adjustment; a coefficient less than 1 indicates that the environmental conditions are better than the standard conditions, allowing for a reduction in the number of sampling points. In this way, the influence of environmental factors on detection complexity is quantified into a calculable coefficient, providing a basis for subsequent adaptive adjustment of the number of sampling points.

[0028] Based on the material attribute information of the standard sample, attribute-detection complexity analysis is performed on each of the Q material attribute information, and Q attribute-detection complexity coefficients are output.

[0029] Furthermore, the steps of the present invention also include: Randomly select one of the Q raw material attribute information as the first raw material attribute information; using the standard sample raw material attribute information as a benchmark, calculate the attribute deviation between the first raw material attribute information and the standard sample raw material attribute information to obtain the first raw material moisture content deviation, the first raw material surface roughness deviation, and the first raw material mineral composition uniformity index deviation; multiply the first raw material moisture content deviation by a preset moisture content sensitivity coefficient and add one to obtain the first moisture content deviation coefficient; multiply the first raw material surface roughness deviation by a preset surface roughness sensitivity coefficient and add one to obtain the first surface roughness deviation coefficient; multiply the first raw material mineral composition uniformity index deviation by a preset uniformity index sensitivity coefficient and add one to obtain the first uniformity index deviation coefficient; perform a weighted calculation on the first moisture content deviation coefficient, the first surface roughness deviation coefficient, and the first uniformity index deviation coefficient to obtain the first attribute-detection complexity coefficient.

[0030] Specifically, when calculating the property-detection complexity coefficient of a single sample, it is first necessary to select one of the raw material property information from Q samples to be tested as the current processing object. For example, the raw material property information of the first sample is randomly selected as the first raw material property information. Then, based on the raw material property information of the standard sample, the property deviation between the first raw material property information and the raw material property information of the standard sample is calculated. The raw material property information of the standard sample is a pre-set benchmark value. For example, the standard raw material moisture content is 0.5%, the standard raw material surface roughness is 1.0 μm, and the standard raw material mineral composition uniformity index is 5%. If the measured raw material moisture content of the first sample is 2.5%, then the raw material moisture content deviation is 2.5% minus 0.5% equals 2.0%; if the measured surface roughness of the first sample is 2.5 μm, then the surface roughness deviation is 2.5 μm minus 1.0 μm equals 1.5 μm; if the measured mineral composition uniformity index of the first sample is 15%, then the uniformity index deviation is 15% minus 5% equals 10%. By calculating the above deviation, the degree to which the raw material properties of the current sample deviate from those of the standard sample can be quantified.

[0031] Next, after obtaining the deviation amounts of each attribute, the deviation coefficients for moisture content, surface roughness, and uniformity index are calculated. Each deviation coefficient is calculated as follows: multiply the corresponding deviation amount by a preset sensitivity coefficient and add one. For example, if the preset moisture content sensitivity coefficient is 0.1%, then the moisture content deviation coefficient equals 1 plus 0.1 multiplied by 2, resulting in 1.2; if the preset surface roughness sensitivity coefficient is 0.05 / μm, then the surface roughness deviation coefficient equals 1 plus 0.05 multiplied by 3, resulting in 1.3; and if the preset uniformity index sensitivity coefficient is 0.02%, then the uniformity index deviation coefficient equals 1 plus 0.02 multiplied by 10, resulting in 1.2. When the sample attributes are consistent with the standard sample, the deviation amount is zero, and the deviation coefficient is 1, which does not affect the detection complexity. When the sample attributes deviate from the standard sample, the deviation coefficient is greater than 1; the larger the deviation, the larger the coefficient, indicating higher detection complexity.

[0032] Finally, the moisture content deviation coefficient, surface roughness deviation coefficient, and uniformity index deviation coefficient are weighted and calculated to obtain the first attribute-detection complexity coefficient. The weight coefficients of each attribute can be obtained through preliminary experimental calibration based on the influence of each raw material attribute on the LIBS test results, or they can be preset based on expert experience. For example, if the preset weights for moisture content are 0.3, surface roughness 0.2, and uniformity index 0.5, then the first attribute-detection complexity coefficient equals 0.3 multiplied by 1.2 plus 0.2 multiplied by 1.3 plus 0.5 multiplied by 1.2, resulting in 1.22. A coefficient greater than 1 indicates that the raw material properties of this sample are more complex than those of the standard sample, requiring a corresponding increase in the number of benchmark sampling points. Performing the above calculations on Q samples to be tested yields Q attribute-detection complexity coefficients.

[0033] The environmental-detection complexity coefficient and the Q attribute-detection complexity coefficients are weighted and calculated to obtain the Q sample detection complexity indices.

[0034] Specifically, for any one of the Q samples to be tested, its corresponding attribute-detection complexity coefficient is weighted and summed with the previously calculated environment-detection complexity coefficient. The environment-detection complexity coefficient reflects the deviation of the current detection environment from the standard environment, while the attribute-detection complexity coefficient reflects the deviation of the sample's own material properties from the standard sample. Through weighted calculation, external environmental factors and sample-specific factors are integrated into a unified quantitative index to characterize the overall detection complexity of the sample under the current detection conditions. The weighting coefficients of the environment-detection complexity coefficient and the attribute-detection complexity coefficient can be pre-set according to their respective influence on the detection results. Generally, the influence of the sample's own attribute differences on detection complexity is greater than that of environmental factors; therefore, the attribute-detection complexity coefficient can be assigned a higher weight, and the environment-detection complexity coefficient a lower weight. For example, if the environmental weight is set to 0.3 and the attribute weight to 0.7, then the sample detection complexity index is equal to 0.3 multiplied by the environment-detection complexity coefficient plus 0.7 multiplied by the attribute-detection complexity coefficient. By performing the weighted calculations described above on Q samples, we obtain Q sample detection complexity indices. The larger the index, the higher the detection complexity of the sample, requiring more sampling points to ensure detection accuracy; the smaller the index, the lower the detection complexity, requiring fewer sampling points.

[0035] Next, after obtaining the sample detection complexity index of Q samples, the Q samples to be tested are sorted in ascending order of their detection complexity index to generate a sample sequence. The purpose of sorting is to place samples with lower detection complexity at the beginning of the sequence and samples with higher detection complexity at the end. For example, if the detection complexity indices of 5 samples in a batch are 1.10, 0.95, 1.25, 0.88, and 1.05, then after sorting in ascending order, the sample sequence will be: sample with index 0.88, sample with index 0.95, sample with index 1.05, sample with index 1.10, and sample with index 1.25. Through this sorting method, in subsequent batch testing, the system will process samples with lower detection complexity first and then samples with higher detection complexity. This facilitates the rapid accumulation of high-quality historical test result consistency data in the early stages of testing, providing a reliable reference for secondary optimization of the number of sampling points for high-complexity samples.

[0036] S400: Based on the Q sample detection complexity indices, the number of baseline sampling points is optimized and corrected to obtain Q suitable sampling points, wherein the number of suitable sampling points is the floor value of the product of the sample detection complexity index and the number of baseline sampling points.

[0037] Specifically, the sample detection complexity index is multiplied by the baseline number of sampling points, and then the product is rounded down to the nearest integer, discarding the decimal part. This rounded result is the optimal number of sampling points for that sample. For example, if a sample has a detection complexity index of 1.25 and a baseline number of sampling points of 20, the product is 25, and after rounding down, the optimal number of sampling points is 25. If another batch of samples has a detection complexity index of 0.85 and a baseline number of sampling points of 20, the product is 17, and after rounding down, the optimal number of sampling points is 17. Through this calculation method, samples with detection complexity higher than the baseline level will receive more sampling points than the baseline to ensure detection accuracy; samples with detection complexity lower than the baseline level will receive fewer sampling points than the baseline to improve detection efficiency. This optimization correction is performed on Q samples to obtain Q optimal sampling point numbers, achieving a precise match between the number of sampling points and the actual detection complexity of each sample.

[0038] S500: Based on the sequence of samples to be tested and the number of Q matching sampling points, the Q samples to be tested are sequentially tested in situ using LIBS technology, and Q in situ test results are output. During the in situ test, the number of matching sampling points for subsequent samples to be tested is optimized again based on the consistency of historical test results.

[0039] Furthermore, step S500 of the present invention further includes: The first sample to be tested in the sequence of samples to be tested is selected as the first sample to be tested, and the number of first matching sampling points of the first sample to be tested is obtained. According to the number of first matching sampling points, the first sample to be tested is detected in situ using LIBS technology, and the first in-situ detection result is output. Specifically, this includes: collecting LIBS spectra of the first number of sampling points on the surface of the first sample to be tested according to a preset sampling path, and acquiring the elemental spectral characteristic signals of each sampling point in real time; performing spectral preprocessing on the collected LIBS spectra, the spectral preprocessing including background subtraction, noise filtering and spectral line normalization; performing feature analysis on the preprocessed spectra, identifying and determining the types of mineral elements contained in the sample based on the peak positions of the characteristic spectral lines; calculating the content of each mineral element based on the intensity of the characteristic spectral lines of the determined element types, combined with a preset quantitative analysis model, and outputting the first in-situ detection result.

[0040] Specifically, firstly, the first sample in the sequence of samples to be tested is selected as the first sample to be tested, and the number of first matching sampling points corresponding to the sample is obtained. The number of first matching sampling points is obtained by optimizing and correcting the number of benchmark sampling points based on the detection complexity index of the first sample to be tested, which reflects the optimal number of sampling points required by the sample under the current environmental conditions and its own properties.

[0041] Next, according to the first number of matching sampling points, in-situ detection of the first sample to be tested is performed using LIBS technology. The specific detection process is as follows: LIBS spectra of the first number of matching sampling points are collected on the surface of the first sample to be tested according to a preset sampling path. The preset sampling path can be a grid path, a serpentine path, or a random path to ensure that each sampling point is evenly distributed on the sample surface and to avoid sampling concentration in local areas. At each sampling point, a laser pulse ablates the sample surface to generate plasma. The characteristic spectra emitted during the plasma cooling process are collected by a spectrometer to obtain the elemental spectral characteristic signals of each sampling point in real time. By collecting the spectra of the first number of matching sampling points, a statistically representative spectral dataset can be obtained. Then, the acquired LIBS spectra are preprocessed, which includes three main steps: background subtraction, noise filtering, and spectral normalization. Background subtraction removes continuous background signals from the spectrum, typically using polynomial fitting or wavelet transform methods. Noise filtering suppresses high-frequency random noise, using smoothing or median filtering algorithms. Spectral normalization corrects for signal intensity variations caused by laser energy fluctuations and sample matrix effects, typically using internal standard normalization or full-spectrum normalization. The preprocessed spectral signal quality is significantly improved, providing a reliable data foundation for subsequent feature analysis.

[0042] Furthermore, the preprocessed spectrum undergoes feature analysis. First, the mineral elements present in the sample are identified and determined based on the peak positions of characteristic spectral lines. Each element has its characteristic emission spectral line wavelength; for example, lead's characteristic spectral line is located around 405.78 nm, zinc's around 213.86 nm, and sulfur's around 921.3 nm. By comparing the peak positions of the measured spectra with a standard spectral database, the element types present in the sample can be qualitatively identified. Then, based on the intensity of the characteristic spectral lines of the identified element types, combined with a pre-set quantitative analysis model, the content of each mineral element is calculated. The quantitative analysis model can be a univariate linear regression model based on standard samples, with the intensity value of the characteristic spectral line as the input and the element content value as the output. Finally, the calculated content of each mineral element is output as the first in-situ detection result, completing the detection of the first sample.

[0043] Furthermore, step S500 of the present invention further includes: During the in-situ detection process, the number of matching sampling points for the (N+1)th sample to be tested is optimized based on the consistency of the detection results of the Nth sample to be tested, resulting in the optimized number of sampling points, where N is a positive integer.

[0044] Furthermore, the steps of the present invention also include: The consistency coefficient of the Nth sample to be tested is calculated based on the dispersion of the detection results of each sampling point in the in-situ detection results of the Nth sample to be tested. A high consistency threshold and a low consistency threshold are set, wherein the high consistency threshold is greater than the low consistency threshold. If the consistency coefficient is greater than or equal to the high consistency threshold, the difference between the consistency coefficient and the high consistency threshold is used as the first coefficient deviation. The product of the first coefficient deviation and the preset sampling point number step size is rounded to obtain the downward adjustment range. The number of suitable sampling points for the (N+1)th sample to be tested is adjusted downward according to the downward adjustment range. If the consistency coefficient is less than or equal to the low consistency threshold, the difference between the low consistency threshold and the consistency coefficient is used as the second coefficient deviation. The product of the second coefficient deviation and the preset sampling point number step size is rounded to obtain the upward adjustment range. The number of suitable sampling points for the (N+1)th sample to be tested is adjusted upward according to the upward adjustment range. If the consistency coefficient is between the low consistency threshold and the high consistency threshold, the number of suitable sampling points for the (N+1)th sample to be tested remains unchanged.

[0045] Specifically, during the testing of the Nth sample, LIBS spectra were collected from multiple sampling points according to the number of appropriate sampling points. After spectral preprocessing and feature analysis, the content of the target mineral elements (such as lead, zinc, sulfur, etc.) could be obtained from each sampling point. Since these sampling points are evenly distributed on the sample surface, the test results of each sampling point should theoretically have good consistency. If the sample itself has a uniform composition and the testing process is stable, the difference between the results of each sampling point is small. If the sample composition is not uniform or there are fluctuations in the testing process, the difference between the results of each sampling point is large.

[0046] Next, the consistency coefficient of the Nth sample is calculated based on the dispersion of the in-situ detection results at each sampling point in the Nth sample. The consistency coefficient quantifies this dispersion, and the calculation process is as follows: The target element content detection results of all sampling points of the Nth sample are collected, forming a set containing M data points, where M is the number of suitable sampling points for the sample. Then, the statistical characteristics of these M data points are calculated. The relative standard deviation (RSD) is typically used as a measure of dispersion. The formula for RSD is: RSD = (Standard Deviation / Mean) × 100%, where the standard deviation reflects the degree to which the results of each sampling point deviate from the mean, and the mean reflects the overall level of the detection results. The calculated RSD is then appropriately transformed and used as the consistency coefficient. The RSD can be obtained by subtracting the normalized RSD value from the reciprocal of RSD or 1, resulting in a higher consistency coefficient indicating more stable detection results. By calculating the consistency coefficient, the stability and representativeness of the detection process for the Nth sample can be quantified, providing a basis for secondary optimization of the number of sampling points for subsequent samples. If the consistency coefficient is high, it means that the current number of matching sampling points is sufficient to obtain stable test results, and the number of sampling points can be appropriately reduced for subsequent samples; if the consistency coefficient is low, it means that the current number of matching sampling points may be insufficient, and the number of sampling points needs to be increased for subsequent samples to improve the representativeness of the test results.

[0047] Next, during the batch testing process, two thresholds for judging the consistency of the tests need to be pre-set: a high consistency threshold and a low consistency threshold. The value of the high consistency threshold is greater than that of the low consistency threshold. For example, a batch of representative lead-zinc tailings samples are selected, and repeated tests are performed multiple times under standard testing conditions using different numbers of sampling points. The consistency coefficient distribution of each test result is statistically analyzed. The upper quartile or 85th percentile of the consistency coefficient under normal testing conditions is set as the high consistency threshold, and the lower quartile or 15th percentile is set as the low consistency threshold. Alternatively, the thresholds can be set directly based on engineering experience, for example, setting the high consistency threshold to 0.85 and the low consistency threshold to 0.65.

[0048] When the consistency coefficient of the Nth sample to be tested is greater than or equal to the high consistency threshold, it indicates that the test result of the sample is very stable, the results of each sampling point are highly consistent, and the current number of matching sampling points is redundant. The number of sampling points for subsequent samples can be appropriately reduced. At this time, the difference between the consistency coefficient and the high consistency threshold is taken as the first coefficient deviation. This deviation reflects the extent to which the current consistency coefficient exceeds the high threshold. The greater the excess, the more serious the redundancy. The first coefficient deviation is multiplied by the preset sampling point number step size and rounded to obtain the reduction adjustment range. The preset sampling point number step size is a pre-set positive integer. For example, a step size of 2 means that each unit of coefficient deviation corresponds to the adjustment of 2 sampling points. The rounding method can be rounding down, rounding up, or rounding to the nearest integer. Then, the number of matching sampling points for the (N+1)th sample to be tested is adjusted downward according to the reduction adjustment range, that is, the number of matching sampling points for the Nth sample is reduced by the reduction adjustment range. However, the adjusted number of sampling points should not be lower than the minimum number of sampling points set by the system.

[0049] When the consistency coefficient of the Nth sample to be tested is less than or equal to the low consistency threshold, it indicates that the test result of the sample is not stable enough, the results of each sampling point are highly dispersed, and the current number of suitable sampling points may be insufficient. The number of sampling points for subsequent samples needs to be appropriately increased. At this time, the difference between the low consistency threshold and the consistency coefficient is taken as the second coefficient deviation. This deviation reflects the extent to which the current consistency coefficient is lower than the low threshold. The greater the deviation, the more serious the deficiency. The second coefficient deviation is multiplied by the preset sampling point number step size and rounded to obtain the upward adjustment range. Then, the number of suitable sampling points for the (N+1)th sample to be tested is adjusted upward according to the upward adjustment range, that is, the number of suitable sampling points for the Nth sample is increased by the upward adjustment range. However, the adjusted number of sampling points should not exceed the maximum number of sampling points set by the system.

[0050] When the consistency coefficient of the Nth sample to be tested is between the low consistency threshold and the high consistency threshold, it indicates that the test result of the sample is within the normal range, and the current number of matching sampling points is appropriate, with neither obvious redundancy nor obvious deficiency. At this time, there is no need to adjust the number of sampling points, and the number of matching sampling points for the (N+1)th sample to be tested remains the same as that for the Nth sample. Through the above three-zone judgment and corresponding adjustment strategy, dynamic optimization of the number of sampling points for subsequent samples based on the consistency of the test results of previous samples is achieved, forming a continuous improvement closed loop in the batch testing process.

[0051] According to the optimized number of sampling points, the (N+1)th sample to be tested is tested in situ, and the (N+1)th in situ test result is output; each sample to be tested in the sequence of samples to be tested is traversed in turn to form a continuous optimization process until the last sample to be tested in the sequence of samples to be tested is tested.

[0052] Specifically, after completing the secondary optimization of the number of sampling points for the (N+1)th sample to be tested, in-situ testing is performed on the sample according to the optimized number of sampling points. The specific testing process is the same as for the first sample to be tested: LIBS spectra of the optimized number of sampling points are collected on the sample surface according to the preset sampling path. The collected spectra are preprocessed with background subtraction, noise filtering, and spectral normalization. The types of mineral elements contained in the sample are identified based on the peak positions of the characteristic spectral lines. Then, the content of each mineral element is calculated based on the intensity of the characteristic spectral lines and a preset quantitative analysis model, and the in-situ detection result of the (N+1)th sample is output. At the same time, the consistency coefficient of the detection is calculated based on the dispersion of the detection results of each sampling point during the detection of the (N+1)th sample. This consistency coefficient will be used to perform secondary optimization of the number of sampling points for the (N+2)th sample.

[0053] The above process forms a continuous optimization closed loop: after each sample is tested, the number of appropriate sampling points for the next sample is adjusted based on the consistency coefficient of that sample. Then, the next sample is tested according to the adjusted number, and the number of sampling points for subsequent samples is adjusted based on the test results of the next sample. This process is repeated for each sample in the sequence to be tested until the last sample in the sequence is tested. Since the sequence of samples to be tested is sorted in ascending order of sample detection complexity index, samples with lower detection complexity are placed at the beginning of the sequence, and samples with higher detection complexity are placed at the end of the sequence. During the testing of simpler samples at the beginning, the system can quickly accumulate consistency coefficient data and gradually optimize the adjustment strategy. When complex samples at the end are detected, the adjustment strategy has stabilized and can provide a more reasonable configuration of the number of sampling points for complex samples. Through this continuous optimization process, the efficiency of the entire batch testing continuously improves as the testing process progresses.

[0054] In summary, the in-situ detection method for lead-zinc tailings aggregates based on LIBS technology provided by this invention has the following technical advantages: By constructing a detection complexity assessment model, using standard detection conditions as a benchmark, and calculating the sample detection complexity index in real time based on current detection environment parameters and raw material properties, the number of benchmark sampling points is optimized and corrected based on this index. This ensures that the number of sampling points matches the actual complexity of the sample, thereby avoiding over-detection of simple samples and under-sampling of complex samples while ensuring detection accuracy, and reducing invalid detection steps. Furthermore, during batch detection, this invention dynamically adjusts the number of appropriate sampling points for the next sample based on the consistency coefficient of the detection results of the previous sample, forming a continuous optimization closed loop. This fully utilizes the detection information of previous samples to guide subsequent detections, further improving the overall efficiency of batch detection. Through the above technical means, this invention can achieve synergistic optimization of accuracy and efficiency in the batch detection of lead-zinc tailings aggregates, achieving significant improvements in detection efficiency, reduction in detection costs, and guarantee of representativeness of detection results, meeting the practical application needs of rapid screening in highway pavement engineering.

[0055] Example 2: This invention also provides an in-situ detection system for lead-zinc tailings aggregates based on LIBS technology. Please refer to the appendix. Figure 2 ,include: The module 11 for obtaining the number of benchmark sampling points is used to obtain the number of benchmark sampling points for LIBS testing under standard testing conditions based on the mapping of the detection error index of lead-zinc tailings aggregate; the current testing information acquisition module 12 is used to acquire the Q raw material attribute information of Q samples to be tested from the lead-zinc tailings aggregate, and to monitor and acquire the current testing environment parameters; the sample testing complexity analysis module 13 is used to perform sample testing complexity analysis based on the standard testing conditions, according to the Q raw material attribute information and the current testing environment parameters, to obtain Q sample testing complexity indices, and to generate a sequence of samples to be tested by sorting them in ascending order of sample testing complexity indices; adaptation The sampling point number generation module 14 is used to optimize and correct the baseline sampling point number according to the Q sample detection complexity indices to obtain Q suitable sampling point numbers, wherein the suitable sampling point number is the floor value of the product of the sample detection complexity index and the baseline sampling point number; the in-situ sample detection module 15 is used to perform in-situ detection on the Q samples to be tested sequentially using LIBS technology based on the sample sequence to be tested and the Q suitable sampling point numbers, and output Q in-situ detection results, wherein, during the in-situ detection process, the suitable sampling point number of subsequent samples to be tested is further optimized based on the consistency of historical detection results.

[0056] Furthermore, the LIBS-based in-situ detection system for lead-zinc tailings aggregates is also used to: obtain the detection error index of lead-zinc tailings aggregates required for road construction; obtain the instrument model of the detection instrument used in the current LIBS detection method; and obtain the minimum number of sampling points under standard detection conditions for the LIBS detection method based on the mapping between the detection error index and the instrument model, as the benchmark number of sampling points. The standard detection conditions include standard detection environment parameters and standard sample raw material property information.

[0057] Furthermore, the LIBS-based in-situ detection system for lead-zinc tailings aggregates is also used to: detect environmental parameters including ambient temperature, ambient humidity, and ambient dust concentration; and raw material property information including raw material moisture content, raw material surface roughness, and raw material mineral composition uniformity index. The calculation steps for the raw material mineral composition uniformity index include: collecting LIBS spectra at N pre-sampled points at preset positions on the surface of the sample to be tested, and calculating the relative standard deviation of the intensity of the characteristic spectral lines of the target element as the raw material mineral composition uniformity index.

[0058] Furthermore, the LIBS-based in-situ detection system for lead-zinc tailings aggregates is also used for: using the standard detection environment parameters as a benchmark, performing an environment-detection complexity analysis based on the current detection environment parameters, and outputting an environment-detection complexity coefficient; using the standard sample raw material attribute information as a benchmark, performing attribute-detection complexity analysis based on the Q raw material attribute information respectively, and outputting Q attribute-detection complexity coefficients; and weighting the environment-detection complexity coefficients and the Q attribute-detection complexity coefficients to obtain Q sample detection complexity indices.

[0059] Furthermore, the LIBS-based in-situ detection system for lead-zinc tailings aggregates is also used for: calculating the parameter deviation between the current detection environment parameters and the standard detection environment parameters, based on the standard detection environment parameters, to obtain temperature deviation, humidity deviation, and dust concentration deviation; multiplying the temperature deviation by a preset temperature sensitivity coefficient and adding one to calculate the temperature deviation coefficient; multiplying the humidity deviation by a preset humidity sensitivity coefficient and adding one to calculate the humidity deviation coefficient; multiplying the dust concentration deviation by a preset dust concentration sensitivity coefficient and adding one to calculate the dust concentration deviation coefficient; and performing a weighted calculation of the temperature deviation coefficient, humidity deviation coefficient, and dust concentration deviation coefficient to obtain the environment-detection complexity coefficient.

[0060] Furthermore, the LIBS-based in-situ detection system for lead-zinc tailings aggregates is also used for: randomly selecting any one of the Q raw material attribute information as the first raw material attribute information; using the standard sample raw material attribute information as a benchmark, calculating the attribute deviation between the first raw material attribute information and the standard sample raw material attribute information to obtain the first raw material moisture content deviation, the first raw material surface roughness deviation, and the first raw material mineral composition uniformity index deviation; multiplying the first raw material moisture content deviation by a preset moisture content sensitivity coefficient and adding one to calculate the first moisture content deviation coefficient; multiplying the first raw material surface roughness deviation by a preset surface roughness sensitivity coefficient and adding one to calculate the first surface roughness deviation coefficient; multiplying the first raw material mineral composition uniformity index deviation by a preset uniformity index sensitivity coefficient and adding one to calculate the first uniformity index deviation coefficient; and performing a weighted calculation on the first moisture content deviation coefficient, the first surface roughness deviation coefficient, and the first uniformity index deviation coefficient to obtain the first attribute-detection complexity coefficient.

[0061] Furthermore, the LIBS-based in-situ detection system for lead-zinc tailings aggregates is also used for: selecting the first sample to be tested in the sequence of samples to be tested as the first sample to be tested, and obtaining the first number of matching sampling points for the first sample to be tested; performing in-situ detection on the first sample to be tested using LIBS technology according to the first number of matching sampling points, and outputting the first in-situ detection result, specifically including: collecting LIBS spectra of the first number of matching sampling points on the surface of the first sample to be tested according to a preset sampling path, and acquiring the elemental spectral characteristic signals of each sampling point in real time; performing spectral preprocessing on the collected LIBS spectra, the spectral preprocessing including background subtraction, noise filtering, and spectral line normalization; performing feature analysis on the preprocessed spectra, identifying and determining the types of mineral elements contained in the sample based on the peak positions of the characteristic spectral lines; calculating the content of each mineral element based on the intensity of the characteristic spectral lines of the determined element types, combined with a preset quantitative analysis model, and outputting the first in-situ detection result.

[0062] Furthermore, the LIBS-based in-situ detection system for lead-zinc tailings aggregates is also used for: during the in-situ detection process, optimizing the number of suitable sampling points for the (N+1)th sample based on the consistency of the detection results of the Nth sample to be tested, to obtain an optimized number of sampling points, where N is a positive integer; performing in-situ detection on the (N+1)th sample to be tested according to the optimized number of sampling points, and outputting the (N+1)th in-situ detection result; and sequentially traversing each sample in the sequence of samples to be tested to form a continuous optimization process until the last sample in the sequence of samples to be tested is detected.

[0063] Furthermore, the LIBS-based in-situ detection system for lead-zinc tailings aggregates is also used for: calculating the detection consistency coefficient of the Nth sample based on the dispersion of the detection results at each sampling point in the in-situ detection results of the Nth sample; setting a high consistency threshold and a low consistency threshold, wherein the high consistency threshold is greater than the low consistency threshold; if the detection consistency coefficient is greater than or equal to the high consistency threshold, the difference between the detection consistency coefficient and the high consistency threshold is used as the first coefficient deviation; the product of the first coefficient deviation and the preset sampling point number step size is rounded to obtain the reduction adjustment range, according to the... The decrease adjustment range is used to adjust the number of matching sampling points for the (N+1)th sample to be tested downwards; if the consistency coefficient is less than or equal to the low consistency threshold, the difference between the low consistency threshold and the consistency coefficient is used as the second coefficient deviation, and the product of the second coefficient deviation and the preset sampling point number step size is rounded to obtain the increase adjustment range, and the number of matching sampling points for the (N+1)th sample to be tested is adjusted upwards according to the increase adjustment range; if the consistency coefficient is between the low consistency threshold and the high consistency threshold, the number of matching sampling points for the (N+1)th sample to be tested remains unchanged.

[0064] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0065] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of this invention and its equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for in-situ detection of lead-zinc tailings aggregates based on LIBS technology, characterized in that, The methods include: The number of baseline sampling points for LIBS testing under standard testing conditions is obtained by mapping the detection error index of lead-zinc tailings aggregate, including: To obtain the detection error index of lead-zinc tailings aggregate required for road construction; Obtain the instrument model of the testing instrument used in the current LIBS testing method; The minimum number of sampling points for LIBS testing under standard testing conditions is obtained based on the aforementioned detection error index and instrument model mapping, and is used as the baseline number of sampling points. The standard testing conditions include standard testing environment parameters and standard sample material property information. Collect the raw material attribute information of Q samples of lead-zinc tailings aggregate to be tested, and monitor and obtain the current testing environment parameters; Based on the aforementioned standard testing conditions, and according to the Q raw material attribute information and current testing environment parameters, a sample testing complexity analysis is performed to obtain Q sample testing complexity indices. These indices are then sorted in ascending order to generate a sequence of samples to be tested, including: Based on the aforementioned standard testing environment parameters, an environment-testing complexity analysis is performed according to the current testing environment parameters, and the environment-testing complexity coefficient is output, including: Using the standard detection environment parameters as a benchmark, the parameter deviation between the current detection environment parameters and the standard detection environment parameters is calculated to obtain the temperature deviation, humidity deviation, and dust concentration deviation. The temperature deviation is calculated by multiplying the temperature deviation by a preset temperature sensitivity coefficient and then adding one. The humidity deviation coefficient is calculated by multiplying the humidity deviation by a preset humidity sensitivity coefficient and then adding one. The dust concentration deviation is calculated by multiplying the dust concentration deviation by a preset dust concentration sensitivity coefficient and then adding one. The environmental-detection complexity coefficient is obtained by weighting the temperature deviation coefficient, humidity deviation coefficient, and dust concentration deviation coefficient. Based on the material property information of the standard sample, attribute-detection complexity analysis is performed on each of the Q material property information, and Q attribute-detection complexity coefficients are output, including: Randomly select one of the Q raw material attribute information as the first raw material attribute information; Based on the attribute information of the standard sample raw material, the attribute deviation between the attribute information of the first raw material and the attribute information of the standard sample raw material is calculated to obtain the deviation of the moisture content of the first raw material, the deviation of the surface roughness of the first raw material, and the deviation of the mineral composition uniformity index of the first raw material. The first moisture content deviation coefficient is calculated by multiplying the moisture content deviation of the first raw material by a preset moisture content sensitivity coefficient and then adding one. The first surface roughness deviation coefficient is calculated by multiplying the surface roughness deviation of the first raw material by a preset surface roughness sensitivity coefficient and then adding one. The first uniformity index deviation coefficient is calculated by multiplying the first raw material mineral composition uniformity index deviation by the preset uniformity index sensitivity coefficient and adding one. The first attribute-detection complexity coefficient is obtained by weighting the first moisture content deviation coefficient, the first surface roughness deviation coefficient, and the first uniformity index deviation coefficient. The environmental-detection complexity coefficient and the Q attribute-detection complexity coefficients are weighted and calculated to obtain Q sample detection complexity indices; Based on the Q sample detection complexity indices, the number of baseline sampling points is optimized and corrected to obtain Q suitable sampling points, wherein the number of suitable sampling points is the floor value of the product of the sample detection complexity index and the number of baseline sampling points. Based on the sample sequence to be tested and the number of Q matching sampling points, LIBS technology is used to perform in-situ detection on the Q samples to be tested in sequence, and Q in-situ detection results are output. During the in-situ detection process, the number of matching sampling points for subsequent samples to be tested is optimized again based on the consistency of historical detection results.

2. The in-situ detection method for lead-zinc tailings aggregates based on LIBS technology according to claim 1, characterized in that, The environmental parameters to be tested include ambient temperature, ambient humidity and ambient dust concentration. The raw material property information includes raw material moisture content, raw material surface roughness and raw material mineral composition uniformity index. The calculation steps of the raw material mineral composition uniformity index include: collecting LIBS spectra of N pre-sampled points at preset positions on the surface of the sample to be tested, and calculating the relative standard deviation of the intensity of the characteristic spectral lines of the target element as the raw material mineral composition uniformity index.

3. The in-situ detection method for lead-zinc tailings aggregates based on LIBS technology according to claim 1, characterized in that, Based on the sample sequence to be tested and the number of Q matching sampling points, LIBS technology is used to sequentially perform in-situ detection on the Q samples to be tested, and output Q in-situ detection results, including: The first sample to be tested in the sequence of samples to be tested is selected as the first sample to be tested, and the number of first matching sampling points of the first sample to be tested is obtained. Based on the first number of matching sampling points, the first sample to be tested is subjected to in-situ detection using LIBS technology, and the first in-situ detection result is output, specifically including: LIBS spectra of a number of sampling points of the first matching sampling point are collected on the surface of the first sample to be tested according to a preset sampling path, and the elemental spectral characteristic signals of each sampling point are obtained in real time. The acquired LIBS spectra are preprocessed, including background subtraction, noise filtering, and spectral line normalization. The preprocessed spectrum is analyzed for characteristics, and the types of mineral elements contained in the sample are identified and determined based on the peak positions of the characteristic spectral lines. Based on the characteristic spectral line intensities of the identified element types, and combined with a pre-set quantitative analysis model, the content of each mineral element is calculated and output as the first in-situ detection result.

4. The in-situ detection method for lead-zinc tailings aggregates based on LIBS technology according to claim 3, characterized in that, The method further includes using LIBS technology to sequentially perform in-situ detection on the Q samples to be tested, outputting Q in-situ detection results, and also includes: During the in-situ detection process, the number of matching sampling points for the N+1th sample to be tested is optimized based on the consistency of the detection results of the Nth sample to be tested, resulting in the optimized number of sampling points, where N is a positive integer. Perform in-situ detection on the (N+1)th sample to be tested according to the optimized number of sampling points, and output the (N+1)th in-situ detection result; The process involves iterating through each sample in the sequence to be tested, forming a continuous optimization process, until the last sample in the sequence is tested.

5. The in-situ detection method for lead-zinc tailings aggregates based on LIBS technology according to claim 4, characterized in that, Based on the consistency of the test results of the Nth sample to be tested, the number of suitable sampling points for the (N+1)th sample to be tested is further optimized, including: The consistency coefficient of the Nth sample to be tested is calculated based on the dispersion of the test results at each sampling point in the in-situ test results of the Nth sample to be tested. Set a high consistency threshold and a low consistency threshold, wherein the high consistency threshold is greater than the low consistency threshold; If the detection consistency coefficient is greater than or equal to the high consistency threshold, the difference between the detection consistency coefficient and the high consistency threshold is taken as the first coefficient deviation. The product of the first coefficient deviation and the preset sampling point number step size is rounded to obtain the reduction adjustment range. The number of matching sampling points for the N+1th sample to be tested is adjusted downward according to the reduction adjustment range. If the consistency coefficient is less than or equal to the low consistency threshold, the difference between the low consistency threshold and the consistency coefficient is taken as the second coefficient deviation. The product of the second coefficient deviation and the preset sampling point number step size is rounded to obtain the upward adjustment range. The number of matching sampling points for the N+1th sample to be tested is adjusted upward according to the upward adjustment range. If the consistency coefficient is between the low consistency threshold and the high consistency threshold, then the number of matching sampling points for the N+1th sample to be tested remains unchanged.

6. A lead-zinc tailings aggregate in-situ detection system based on LIBS technology, characterized in that, The steps for implementing the in-situ detection method for lead-zinc tailings aggregates based on LIBS technology according to any one of claims 1 to 5 include: The module for obtaining the number of benchmark sampling points is used to obtain the number of benchmark sampling points under standard testing conditions based on the detection error index mapping of lead-zinc tailings aggregates. The current detection information acquisition module is used to collect the raw material attribute information of Q samples of lead-zinc tailings aggregate to be tested, and to monitor and acquire the current detection environment parameters; The sample testing complexity analysis module is used to perform sample testing complexity analysis based on the standard testing conditions, according to the Q raw material attribute information and the current testing environment parameters, to obtain Q sample testing complexity indices, and to generate a sequence of samples to be tested by sorting the sample testing complexity indices from smallest to largest. The module for generating the number of adaptive sampling points is used to optimize and correct the number of benchmark sampling points according to the Q sample detection complexity indices to obtain the number of Q adaptive sampling points, wherein the number of adaptive sampling points is the floor value of the product of the sample detection complexity index and the number of benchmark sampling points. The in-situ sample detection module is used to perform in-situ detection on the Q samples to be tested sequentially using LIBS technology based on the sample sequence to be tested and the number of Q matching sampling points, and output Q in-situ detection results. During the in-situ detection process, the number of matching sampling points for subsequent samples to be tested is optimized again based on the consistency of historical detection results.

Images