Pyrite classification test method

Abstract

The invention relates to the technical field of material detection and analysis, and discloses a pyrite classification test method, which comprises the following steps: crushing and grinding a to-be-detected sample to prepare a standard detection sample; performing parameter detection on the standard detection sample to obtain spectrum time sequence data, component analysis data and microscopic image data; extracting spectral dynamic characteristics from the spectral time sequence data, extracting component composition characteristics from the component analysis data, extracting microscopic morphological characteristics from the microscopic image data, and then generating a phase determination index and a core characteristic subset; determining a phase classification result through a tree classification decision model according to the core feature subset; and generating a confidence label of a phase classification result according to the phase judgment index, and performing parameter detection on the standard detection sample again when the phase judgment index is lower than a preset threshold value. And the pyrite classification efficiency and accuracy are improved.

Description

Technical Field

[0001] This invention relates to the field of materials testing and analysis technology, and in particular to a method for classifying and analyzing pyrite. Background Technology

[0002] The inspection techniques for pyrite primarily rely on traditional mineralogical methods and chemical analysis. For unprocessed raw ore, customs and inspection agencies typically use visual observation, making preliminary judgments based on surface characteristics such as crystal morphology (cubic, octahedral, etc.), metallic luster, and streak color (greenish-black). For compositional analysis, current techniques generally employ wet chemical analysis, determining the content of major elements such as iron, sulfur, and arsenic through titration and colorimetry after acid dissolution of the sample, supplemented by X-ray diffraction (XRD) for further analysis of phase composition. In terms of instrumental testing, some laboratories have introduced X-ray fluorescence spectroscopy (XRF) for rapid elemental screening or use scanning electron microscopy (SEM) to observe the microscopic morphology of minerals. However, these techniques are fragmented in practical application and have not yet formed a systematic, integrated testing system.

[0003] Current methods for inspecting pyrite have the following drawbacks: Imported pyrite concentrate is crushed and ground to micron-sized particles, rendering traditional visual inspection completely ineffective. Furthermore, the detection of physical density, magnetic properties, and other characteristics is easily interfered with by associated minerals, making accurate mineral identification impossible. Chemical wet methods require complex procedures such as sample digestion and multi-element stepwise determination, with a single sample testing cycle exceeding 48 hours. Moreover, the detection of toxic elements such as arsenic poses safety risks, making it difficult to meet the rapid customs clearance requirements for bulk minerals. Existing chemical methods can only provide total elemental data and cannot distinguish between pyrite (FeS2) and pyrrhotite (Fe2+). 1-x The phase differences of minerals related to their genesis and occurrence (S) are significant. However, XRD phase analysis is affected by the oxide layer on the surface of concentrate particles, resulting in a significant decrease in the intensity of characteristic peaks and a high misjudgment rate. In the existing detection process, the technical modules such as elemental analysis, morphological observation, and phase identification are isolated from each other and lack a data fusion mechanism, making it difficult to establish an accurate mapping relationship between mineral properties and customs classification codes, leading to frequent classification disputes.

[0004] Therefore, there is an urgent need for a pyrite classification and testing method to improve the efficiency and accuracy of pyrite classification. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention provides a method for classifying and testing pyrite, which can improve the accuracy, efficiency, and safety of pyrite testing.

[0006] This invention provides a method for classifying and testing pyrite, comprising the following steps: S1. The sample to be tested is crushed and ground to prepare a standard test sample; S2. Perform parameter detection on the standard test sample to obtain spectral time series data, component analysis data and microscopic image data; S3. Extract spectral dynamic features from the spectral time series data, extract component composition features from the component analysis data, and extract microscopic morphological features from the microscopic image data; S4. Based on mineralogical properties, generate phase determination indices and a subset of core features for the spectral dynamics, compositional features and microscopic morphological features; S5. Based on the core feature subset, determine the phase classification result through a tree-based classification decision model; S6. Generate a confidence label for the phase classification result based on the phase determination index, and re-test the parameters of the standard test sample when the phase determination index is lower than a preset threshold.

[0007] Furthermore, the crushing and grinding process includes using mechanical crushing equipment to crush the sample to be tested to a preset particle size range.

[0008] Furthermore, in S2, The spectral time-series data were acquired using a laser-induced breakdown spectrometer. The component analysis data were obtained by detecting the content of sulfur, iron and associated elements using X-ray fluorescence spectrometry; the microscopic image data were obtained by scanning electron microscopy.

[0009] Furthermore, S3 includes: The spectral dynamics features include the decay rate of the sulfur characteristic peak, the stability of the intensity ratio of the iron-sulfur characteristic peak, and the characteristic peak area ratio. The compositional characteristics include the sulfur-iron atomic ratio, the distribution coefficient of associated elements, and the elemental deviation index; The microscopic morphological features include grain boundary tortuosity, the proportion of oxide regions, and grain size dispersion.

[0010] Furthermore, in S4, the core feature subset includes the sulfur-iron atomic ratio, the distribution coefficient of associated elements, and the proportion of oxidation regions; the phase determination index is generated by a weighted linear combination.

[0011] Furthermore, the phase determination index is generated through a weighted linear combination, including: Normalize the feature parameters obtained in S3. Weighting factors are assigned to the importance of pyrite identification based on characteristic parameters; The phase determination index is generated by weighted linear combination of the normalized characteristic parameters.

[0012] Furthermore, in S5, the tree-structured classification decision model specifically includes: First-level phase classification node based on sulfur-iron atomic ratio; Identification nodes for coexisting minerals based on the distribution coefficient of associated elements; Reliability verification node with the percentage of oxidized region as the threshold.

[0013] Furthermore, in step S5, based on the core feature subset, the phase classification result is determined through a tree-based classification decision model. The specific execution steps include: S51. Determination of primary phase classification nodes: If the sulfur-iron atomic ratio is greater than or equal to the preset sulfur-iron atomic ratio threshold, it is determined to be a pyrite mineral branch, and proceeds to S52; If the sulfur-iron atomic ratio is less than the preset sulfur-iron atomic ratio threshold, it is determined to be a non-pyrite mineral branch, and the initial screening result of pyrrhotite or arsenopyrite is output. S52. Identification of associated minerals, based on the correlation characteristics between the distribution coefficients of the associated elements and typical deposit types: When the degree of arsenic enrichment meets the indication conditions for arsenopyrite and the ratio of base metal elements conforms to the genetic rules of arsenopyrite, it is identified as arsenopyrite. When the ratio of transition metal elements meets the characteristics of sedimentary genesis, and the fluctuation of the sulfur-iron atomic ratio is within a preset stable range, it is determined to be sedimentary pyrite. S53. Reliability verification node judgment, based on the comparison result of the oxidation region ratio and the surface interference critical value: If the proportion of the oxidized region is less than the surface interference critical value, the S52 classification result is directly output. If the proportion of the oxidized region is greater than or equal to the surface interference critical value, proceed to step S6.

[0014] Furthermore, the confidence label generation rule in S6 is as follows: If the phase determination index is ≥0.9, it is marked as high confidence. If 0.6 ≤ the phase determination index < 0.9, it is marked as medium confidence level; If the phase determination index is <0.6, it is marked as low confidence and the parameters of the standard test sample are retested.

[0015] The present invention has the following technical effects: The samples to be tested were crushed and ground to prepare standard test specimens. Then, laser-induced breakdown spectroscopy, X-ray fluorescence spectroscopy, and scanning electron microscopy were used to acquire spectral time-series data, component analysis data, and microscopic image data, respectively. Next, the extracted spectral dynamic characteristics (such as the intensity ratio and decay rate of pyrite characteristic peaks), component composition characteristics (such as the atomic ratio of pyrite and iron and the distribution coefficient of associated elements), and microscopic morphological characteristics (such as the tortuosity of grain boundaries and the proportion of oxidation regions) were normalized and weighted to construct a comprehensive phase determination index. Based on the mineral genesis law, a tree-shaped classification decision model was established to achieve intelligent identification and classification of pyrite. This method applies the principles of multi-source data fusion and pattern recognition, effectively overcoming problems such as failure of traditional visual discrimination, long chemical analysis cycle, and susceptibility of XRD to oxidation interference. It improves the classification accuracy and detection efficiency of fine-grained pyrite samples, solves the classification disputes caused by fragmented detection methods and weak phase differentiation capabilities in existing technologies, reduces the safety risks brought by wet analysis of toxic elements, and meets the needs of customs for rapid clearance. Attached Figure Description

[0016] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0017] Figure 1 This is a flowchart of a pyrite classification and testing method provided in an embodiment of the present invention. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below. Obviously, the described embodiments are only a part of the embodiments of this invention, and not all of them. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0019] Figure 1 This is a flowchart of a pyrite classification and testing method provided in an embodiment of the present invention. See also... Figure 1 Specifically, it includes: 1. A method for classifying and testing pyrite, comprising the following steps: S1. The sample to be tested is crushed and ground to prepare a standard test sample.

[0020] In some embodiments, the crushing and grinding process includes crushing the sample to be tested to a preset particle size range using a mechanical crushing device.

[0021] Select appropriate mechanical crushing equipment based on the hardness and particle size of the sample to be tested. Common equipment includes jaw crushers, cone crushers, and ball mills. To ensure the effectiveness of subsequent testing steps, the sample needs to be crushed to a specific particle size range. For example, it can be set to less than 74 micrometers (200 mesh sieve) to ensure sample homogeneity and representativeness. Place the large sample into the selected mechanical crushing equipment, start the equipment, and gradually adjust the speed or pressure to gradually crush the sample into smaller particles. Periodically sample and check the particle size until the preset particle size range is reached. Use a standard sieve to sieve the crushed sample to remove oversized or undersized particles, ensuring that all particles are within the preset range. Then, thoroughly mix the qualified particles to prepare a standard test sample.

[0022] The above steps effectively prepare standard test samples suitable for subsequent detailed analysis, laying a solid foundation for later steps such as spectral time-series data acquisition, component analysis, and microscopic image acquisition. This not only solves the problem of traditional visual inspection failure on micron-sized particles but also improves the efficiency and accuracy of the entire testing process.

[0023] S2. Perform parameter detection on the standard test sample to obtain spectral time series data, component analysis data and microscopic image data.

[0024] During the parameter detection phase, the standard test sample undergoes collaborative testing by three types of instruments to acquire spectral time-series data, component analysis data, and microscopic image data. Traditional detection methods often rely on a single data source. Therefore, this solution introduces a collaborative working mechanism of three types of instruments: laser-induced breakdown spectrometer, X-ray fluorescence spectrometer, and scanning electron microscope. These instruments collect data on elemental dynamic behavior, component quantification, and microscopic morphology, respectively. The fusion of multimodal data enhances the discriminative ability of the classification model.

[0025] In some embodiments, spectral time-series data are acquired using a laser-induced breakdown spectrometer; component analysis data are obtained by detecting the content of sulfur, iron, and associated elements using an X-ray fluorescence spectrometer; and microscopic image data are acquired using a scanning electron microscope.

[0026] Laser-induced breakdown spectrometry (LASDS) continuously acquires spectral signals within a preset time window, generating spectral time-series data that includes peak intensity variations. Specifically, a high-energy pulsed laser is focused onto the sample surface, instantly vaporizing and ionizing the sample material to form plasma. This plasma emits characteristic spectral signals during cooling. The spectral signals are continuously acquired according to the preset time window, covering the entire lifecycle of plasma formation, expansion, and decay. During this process, the characteristic peak intensities of elements such as sulfur and iron dynamically change over time. For example, the peak intensity of sulfur at a specific wavelength may exhibit different decay rates due to differences in chemical morphology (such as sulfides and sulfates). By capturing these temporal variation characteristics, the chemical valence state and bonding state of sulfur can be indirectly reflected, providing dynamic data support for distinguishing pyrite from pyrrhotite.

[0027] The associated elements detected by X-ray fluorescence spectrometry include at least three elements: arsenic, lead, and zinc. Specifically, the content and ratio of these elements are closely related to the genesis of the deposit and the mineral coexistence. For example, a high content of arsenic may indicate the presence of arsenopyrite (FeAsS), while abnormal enrichment of lead and zinc may reflect a polymetallic sulfide coexistence environment. By simultaneously detecting major and associated elements, X-ray fluorescence spectrometry not only provides a precise basis for calculating the sulfur-iron atomic ratio, but also establishes a mineralogical correlation model based on the types and contents of associated elements, assisting the classification model in identifying complex coexisting minerals.

[0028] The microscopic imaging equipment is a scanning electron microscope, which acquires the microscopic image data in backscatter mode. Specifically, the intensity of the backscattered electron signal is positively correlated with the atomic number of a local region of the sample; high atomic number regions (such as sulfides containing heavy metals) appear bright, while low atomic number regions (such as iron oxide or silicate impurities) appear dark. Microscopic images acquired in this mode can clearly show the boundary morphology of pyrite grains, surface oxidation regions, and the distribution characteristics of associated minerals. For example, pyrrhotite, due to its higher iron content, has a higher backscattered signal intensity than pyrite, while surface oxidation regions exhibit grayscale differences due to oxygen enrichment. These image features provide direct morphological discrimination criteria for convolutional neural network branches, compensating for the shortcomings of chemical composition data in microstructural characterization.

[0029] By acquiring multi-dimensional data (spectral time series, component analysis, microscopic images), we can gain a comprehensive understanding of sample characteristics from multiple perspectives, including elemental composition, phase distribution, and microstructure. This significantly improves the accuracy of pyrite classification and testing, helps build more complex models for classification and prediction, distinguishes similar minerals (such as pyrite and pyrrhotite), and reduces the misclassification rate.

[0030] S3. Extract spectral dynamic features from the spectral time series data, extract component composition features from the component analysis data, and extract microscopic morphological features from the microscopic image data.

[0031] In some embodiments, S3 includes: The spectral dynamics features include the decay rate of the sulfur characteristic peak, the stability of the intensity ratio of the iron-sulfur characteristic peak, and the characteristic peak area ratio. The compositional characteristics include the sulfur-iron atomic ratio, the distribution coefficient of associated elements, and the elemental deviation index; The microscopic morphological features include grain boundary tortuosity, the proportion of oxide regions, and grain size dispersion.

[0032] In a preferred embodiment of the present invention, feature extraction processing is performed on the three types of raw data (spectral time-series data, compositional analysis data, and microscopic image data) to obtain key mineral identification features for subsequent fusion and determination. Specifically, this includes: Spectroscopic dynamics features were extracted from the spectral time-series data: The spectral time-series data was acquired by laser-induced breakdown spectroscopy (LIBS), recording the change of the plasma spectral signal generated after the sample was excited by a laser over time. Based on this data, spectral dynamics features were extracted: the characteristic peak decay rate, defined as the rate at which the intensity of the characteristic emission peak of sulfur decays exponentially over time; this was extracted by calculating the half-life or decay constant through exponential fitting of the spectral signal; the characteristic peak decay rate reflects the binding state and excitation response characteristics of sulfur in the sample, and helps to distinguish between pyrite (FeS2) and pyrrhotite (Fe2+). 1-x S) and other sulfur-containing minerals. The stability of the iron-sulfur characteristic peak intensity ratio is defined as selecting multiple iron and sulfur characteristic peaks, measuring their intensity ratio at different time points, and calculating the standard deviation; through multi-time sampling, the fluctuation range of the iron / sulfur characteristic peak intensity ratio is statistically analyzed; the stability of the iron-sulfur characteristic peak intensity ratio can reflect the homogeneity of the sample and the stability of the sulfur-iron ratio, helping to determine whether it is a pure phase of pyrite. Characteristic peak area ratio: defined as the ratio of the total area of ​​iron and sulfur characteristic peaks over multiple periods; by integrating the entire spectral curve, the Fe / S area ratio is calculated; the characteristic peak area ratio can compensate for the influence of instantaneous spectral fluctuations, providing a more stable quantitative basis.

[0033] Compositional characteristics were extracted from the component analysis data: The component analysis data was acquired using X-ray fluorescence spectrometry (XRF) and included information on the content of sulfur, iron, and other associated elements (such as As, Cu, Zn, Pb, etc.). The following compositional characteristics were extracted: the sulfur-iron atomic ratio (Fe / S), which is calculated by converting the mass percentage measured by XRF into a molar ratio, and then obtaining the Fe / S atomic ratio. The sulfur-iron atomic ratio directly reflects the mineral type; the theoretical value for pyrite is 0.5, pyrrhotite is usually higher than this value, and arsenopyrite is lower than this value due to As replacing some of the S. The distribution coefficient of associated elements, i.e., the relative enrichment of elements such as As, Cu, Zn, and Pb in the sample, was constructed by comparing the content of each element with typical deposit types to construct a distribution coefficient vector, used to distinguish different types of pyrite, such as sedimentary pyrite, hydrothermal pyrite, and arsenopyrite. The elemental deviation index compares the actual detected elemental proportions with the ideal pyrite chemical formula (FeS2) and calculates the deviation. It uses Euclidean or Mahalanobis distance to measure the difference in elemental proportions and can assess whether a sample is pure pyrite or contains other phases or impurities.

[0034] Microscopic morphological features were extracted from microscopic image data acquired by scanning electron microscopy (SEM) to observe the microstructure of the sample surface. The following microscopic morphological features were extracted: grain boundary tortuosity, i.e., the complexity of the crystal edge contour; the fractal dimension or average curvature of grain boundaries was calculated using image segmentation and contour tracking algorithms; this reflects the crystal growth environment, with sedimentary pyrite having smoother grain boundaries and hydrothermal pyrite having more complex ones; and the proportion of oxidation area, i.e., the proportion of the sample surface area undergoing oxidation and alteration to the overall image, identified using energy dispersive spectroscopy (EDS) to identify oxide layer elements (such as Fe). 3+ The area ratio of oxidized regions is statistically analyzed using image recognition technology. The proportion of oxidized regions affects the accuracy of XRD detection; samples with high oxidation areas need to be reprocessed or re-examined. Grain size dispersion, i.e., the degree of variation in grain size in the sample, involves particle identification and statistical analysis of SEM images to calculate the standard deviation or coefficient of variation. Grain size dispersion reflects the degree of mineral crystallization; if the grain uniformity of a concentrate sample is poor, it may be a mixed mineral.

[0035] The three types of features described above characterize the essential properties of pyrite from three dimensions: elemental excitation behavior, chemical composition distribution, and microstructural features, respectively, and form the basis for subsequent feature fusion and classification decisions. Through the systematic extraction of these features, pyrite can be more accurately distinguished from its associated minerals (such as pyrrhotite and arsenopyrite); the robustness and anti-interference ability of the detection system can be improved; and the efficient operation of subsequent weighted fusion and tree-based classification models can be supported.

[0036] S4. Based on mineralogical characteristics, the spectral dynamics features, the compositional features, and the microscopic morphological features, generate phase determination indices and a subset of core features.

[0037] In some embodiments, in step S4, the core feature subset includes the sulfur-iron atomic ratio, the distribution coefficient of associated elements, and the proportion of oxidation regions; the phase determination index is generated by a weighted linear combination.

[0038] Specifically, firstly, three key features with the most representative and discriminative power are selected from all extracted features to form a core feature subset: the sulfur-iron atomic ratio, the distribution coefficient of associated elements, and the proportion of oxidized regions. This core feature subset serves as the main input variable for subsequent classification decisions. Subsequently, a weighted linear combination method is used to construct phase determination indicators.

[0039] In some embodiments, the phase determination index is generated by a weighted linear combination, including: Normalize each type of feature parameter obtained in S3. Since the dimensions and value ranges of each feature parameter are different, they need to be normalized first to make them comparable.

[0040] Weighting factors are assigned based on the importance of characteristic parameters in pyrite identification. Based on the importance of each characteristic in pyrite identification, corresponding weighting factors are set by expert analysis or in conjunction with experimental data analysis. For example: the weighting factor for the sulfur-iron atomic ratio can be set to 0.4; the weighting factor for the distribution coefficient of associated elements can be set to 0.35; and the weighting factor for the proportion of oxidized regions can be set to 0.25.

[0041] The phase determination index is generated by performing a weighted linear combination of the normalized feature parameters. The final phase determination index is obtained by multiplying the normalized feature values ​​with their corresponding weights and summing the results.

[0042] The phase identification index reflects the likelihood that a sample belongs to the pyrite group of minerals. A value closer to 1 indicates a stronger match to the typical characteristics of pyrite; a lower value suggests the presence of interfering factors (such as high oxidation or impurity contamination). In subsequent steps, the phase identification index will serve as an auxiliary criterion to improve the accuracy and stability of the classification.

[0043] S5. Based on the core feature subset, determine the phase classification result through a tree-based classification decision model.

[0044] In a preferred embodiment of the present invention, step S5 includes: systematically classifying the sample according to the core feature subset extracted from the multimodal detection data (including sulfur-iron atomic ratio, associated element distribution coefficient and oxidation region proportion) through a structured tree classification decision model, and finally determining whether it belongs to pyrite and its specific type.

[0045] The tree-structured classification decision model in S5 specifically includes: First-level phase classification node based on sulfur-iron atomic ratio; Identification nodes for coexisting minerals based on the distribution coefficient of associated elements; Reliability verification node with the percentage of oxidized region as the threshold.

[0046] In some embodiments, in step S5, the phase classification result is determined by a tree-based classification decision model based on the core feature subset. The specific execution steps include: S51. Determination of primary phase classification nodes: If the sulfur-iron atomic ratio is greater than or equal to a preset sulfur-iron atomic ratio threshold (e.g., 0.5, corresponding to the stoichiometric ratio of ideal pyrite FeS2), it is determined to be a pyrite mineral branch and proceeds to S52; If the sulfur-iron atomic ratio is less than the preset sulfur-iron atomic ratio threshold, it is determined to be a non-pyrite mineral branch, and the preliminary screening result of pyrrhotite or arsenopyrite is output. It is initially identified as pyrrhotite or arsenopyrite and the preliminary screening result is output.

[0047] S52. Association mineral identification node: Based on the correlation characteristics between the associated element distribution coefficient and typical deposit types, this node further refines the classification based on the associated element distribution coefficient to distinguish different genetic types of pyrite and the types of minerals associated with it. When the degree of arsenic enrichment meets the indication conditions for arsenopyrite (such as the As content being higher than a set threshold), and the ratio of base metal elements (such as Cu / Zn, Zn / Pb, etc.) conforms to the genetic rules of arsenopyrite, it is identified as arsenopyrite. When the ratio of transition metal elements (such as Fe, Cu, Zn) conforms to the characteristics of sedimentary genesis, and the fluctuation of the sulfur-iron atomic ratio is within a preset stable range (such as ±0.05), it is determined to be sedimentary pyrite. S53. Reliability verification node judgment: Based on the comparison result between the oxidation region ratio and the surface interference threshold, this node evaluates the reliability of the classification result based on the oxidation region ratio to prevent misjudgment due to sample surface oxidation.

[0048] If the proportion of the oxidized area is less than the surface interference threshold (e.g., 15%), the S52 classification result is directly output, indicating that the sample surface condition is good and the classification result is reliable. If the proportion of the oxidized region is greater than or equal to the surface interference threshold, it indicates that the sample may be affected by oxidation, and the classification result may be biased. Proceed to step S6.

[0049] S6. Generate a confidence label for the phase classification result based on the phase determination index, and re-test the parameters of the standard test sample when the phase determination index is lower than a preset threshold.

[0050] In a preferred embodiment of the present invention, step S6 includes: assessing the confidence level of the phase classification result obtained in S5 based on the phase determination index generated in S4, and activating a feedback mechanism to re-detect the parameters of the sample when the confidence level is insufficient.

[0051] In some embodiments, the confidence label generation rule in S6 is: If the phase determination index is ≥0.9, it is marked as high confidence. If 0.6 ≤ the phase determination index < 0.9, it is marked as medium confidence level; If the phase determination index is <0.6, it is marked as low confidence and the parameters of the standard test sample are retested.

[0052] When the classification result is marked as low confidence, the system automatically returns to stage S2 to retest the parameters of the original standard test sample. During the retesting process, the test parameters can be adjusted or the sample processing method can be optimized based on the problems found in the previous test to improve the accuracy of subsequent tests. The data after the retesting is input into the S3 to S5 process again to form a closed-loop intelligent testing mechanism.

[0053] Through the synergistic effect of the aforementioned tree-structured classification decision model and confidence feedback mechanism, this invention significantly improves the classification accuracy in pyrite analysis. It uses multi-level classification logic combined with core feature subsets (such as sulfur-iron atomic ratio, associated element distribution coefficient, and oxidation region proportion) for systematic judgment, effectively enhancing the ability to distinguish pyrite and its associated minerals. Simultaneously, by introducing the oxidation region proportion as a reliability assessment factor, the system's ability to identify and respond to interference factors such as sample surface oxidation is improved, avoiding the risk of misjudgment. Furthermore, the confidence label generated based on phase determination indicators enables a re-examination mechanism for low-quality samples, forming a closed-loop feedback control that further ensures the reliability of the test results. The entire classification and evaluation process requires no manual intervention, possessing a high degree of automation, and is suitable for rapid batch sample testing in scenarios such as customs clearance.

[0054] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the technical solutions of the embodiments of the present invention.

Claims

1. A method for classifying and testing pyrite, characterized in that, The method comprises the following steps: S1, performing a crushing and grinding treatment on a to-be-detected sample to prepare a standard detection sample; S2, performing parameter detection on the standard detection sample to obtain spectral timing data, component analysis data, and microscopic image data; S3, extracting spectral dynamics features from the spectral timing data, extracting component composition features from the component analysis data, and extracting microscopic morphology features from the microscopic image data; S4, generating a phase determination index and a core feature subset based on mineralogical characteristics of the spectral dynamics features, the component composition features, and the microscopic morphology features; S5, determining a phase classification result by a tree-shaped classification decision model according to the core feature subset; S6, generating a confidence label of the phase classification result according to the phase determination index, and re-performing parameter detection on the standard detection sample when the phase determination index is lower than a preset threshold.

2. The method of claim 1, wherein the pyrite is classified by a test. In the S1, The crushing and grinding treatment comprises crushing the to-be-detected sample to a preset particle size range by using a mechanical crushing device.

3. The method of claim 1, wherein the pyrite is classified by a test. In the S2, The spectral timing data are collected by a laser-induced breakdown spectrometer; The component analysis data are detected by an X-ray fluorescence spectrometer to detect the contents of sulfur, iron, and associated elements; and the microscopic image data are collected by a scanning electron microscope.

4. The method of claim 1, wherein the pyrite is classified by a test. In the S3, The spectral dynamics features include a sulfur characteristic peak decay rate, an iron-sulfur characteristic peak intensity ratio stability, and a characteristic peak area ratio; The component composition features include a sulfur-iron atomic ratio, an associated element distribution coefficient, and an element deviation index; The microscopic morphology features include a grain boundary tortuosity, an oxidation area proportion, and a grain size dispersion.

5. The method of claim 1, wherein the pyrite is classified by a test. In the S4, the core feature subset includes a sulfur-iron atomic ratio, an associated element distribution coefficient, and an oxidation area proportion; and the phase determination index is generated by weighted linear combination.

6. A method of pyrite classification assay according to claim 5, characterised in that, The phase determination index is generated by weighted linear combination, comprising: normalizing each type of feature parameter obtained in S3; allocating a weight factor according to the importance degree of the feature parameter in pyrite identification; generating the phase determination index by weighted linear combination of the normalized feature parameters.

7. A method of pyrite classification assay according to claim 4, characterised in that, In the S5, the tree-shaped classification decision model specifically comprises: a first-level phase classification node based on a sulfur-iron atomic ratio; a paragenetic mineral identification node based on an associated element distribution coefficient; a reliability verification node with the oxidation area proportion as a threshold.

8. A method of pyrite classification assay according to claim 7, characterised in that, In the S5, the phase classification result is determined by the tree-shaped classification decision model according to the core feature subset, and the specific execution steps comprise: S51, first-level phase classification node judgment: if the sulfur-iron atomic ratio is greater than or equal to a preset sulfur-iron atomic ratio threshold, it is determined as a pyrite mineral branch, and S52 is entered; if the sulfur-iron atomic ratio is less than the preset sulfur-iron atomic ratio threshold, it is determined as a non-pyrite mineral branch, and a preliminary screening result of pyrrhotite or arsenic pyrite is outputted; S52, paragenetic mineral identification node judgment based on the correlation characteristics of the associated element distribution coefficient and typical ore deposit types: when the enrichment degree of arsenic element meets the indication condition of arsenic pyrite, and the ratio of base metal elements meets the genetic law of arsenic pyrite, it is determined as arsenic pyrite; When the transition metal element ratio meets the sedimentary origin characteristics, and the fluctuation of the sulfur-iron atomic ratio is in the preset stable interval, the deposit type pyrite is determined; S53, the reliability verification node determines whether the oxidation area proportion is less than the surface interference critical value according to the comparison result of the oxidation area proportion and the surface interference critical value: If the oxidation area proportion is less than the surface interference critical value, the S52 classification result is directly outputted; If the oxidation area proportion is greater than or equal to the surface interference critical value, S6 is executed.

9. The method of claim 1, wherein the pyrite is classified by a test. The confidence label generation rule in S6 is: If the phase determination index is greater than or equal to 0.9, it is marked as high confidence; If 0.6 is less than the phase determination index, it is marked as low confidence and the standard detection sample is re-detected. ​

Images