Ore type division and ore body delineation method and system, medium and product

Abstract

The invention relates to the technical field of ore type division, in particular to an ore type division and ore body delineating method and system, a medium and a product. According to the ore type division and ore body delineation method, a plurality of groups of exploration auxiliary samples with determined ore types in an ore deposit are used for constructing sample sets, target element content data in each sample set are obtained by using p-XRF, and the target element content data are input into a machine learning classification model for training after standardization processing. And on the basis of the trained machine learning classification model, performing automatic classification on the exploration auxiliary samples of unknown types in the ore deposit. And further, sequentially drawing a borehole histogram, a geological profile map and an ore body distribution map according to a classification result. According to the method, intelligent and efficient division of ore types is realized, the subjective influence of manual division is effectively reduced, and the division efficiency and accuracy are remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of ore type classification technology, specifically to a method, system, storage medium, and computer program product for ore type classification and ore body delineation. Background Technology

[0002] Mineral deposit exploration is a crucial step before the development and utilization of mineral deposits. Systematic geological exploration can delineate ore bodies, classify ore types, and identify the distribution characteristics of valuable and harmful elements, thus providing a scientific basis for subsequent mining and production. However, current ore type classification in borehole logging primarily relies on visual observation by geologists, lacking quantitative data support. This results in relatively coarse logging results, susceptible to subjective interference, and difficult to accurately define the boundaries of different ore types. Furthermore, existing geological exploration reports often delineate ore bodies based on ore grade, failing to systematically reveal the internal structure of the ore body and the spatial distribution characteristics of ore types. If traditional methods are used for fine ore type classification, extensive chemical analysis is typically required, which is not only time-consuming and labor-intensive but also suffers from data acquisition delays, failing to meet the demands of efficient modern mining production. Simultaneously, existing ore classification technologies often rely on single elements or single indicators, lacking comprehensive utilization of multi-element information, leading to insufficient accuracy and reliability of classification results. Moreover, these methods often involve complex sample pretreatment processes, further increasing operational difficulty and time costs. In actual production, due to the complex geological conditions of mining areas, mining operations usually require the mixed mining and beneficiation of multiple types of ores. Traditional ore classification methods are difficult to adapt to the needs of fine classification and rapid identification of multiple types of ores, thus restricting their promotion and application in large-scale mineral deposit exploration and production practices.

[0003] In recent years, data-driven technologies have been increasingly widely applied in geological research. As an efficient data processing method, machine learning has significantly improved the processing and analysis capabilities of massive geochemical data, and has gradually become an important research hotspot in the field of quantitative geoscience. The application of various machine learning algorithms in geoscience problems is constantly expanding and deepening. In the process of mine production, a large amount of ore geochemical data has been accumulated, mainly from conventional laboratory chemical analysis and p-XRF testing. Among them, conventional laboratory analysis methods still dominate in core sample elemental analysis and ore geochemical anomaly identification due to their high testing accuracy. However, these methods usually require multiple steps such as "sample collection—transportation—processing—chemical analysis" before test results can be obtained. The overall process cycle is long and costly, making it difficult to meet the actual needs of rapid data acquisition and efficient production decision-making. Summary of the Invention

[0004] To address the technical problems of existing ore type classification processes relying heavily on researchers' subjective judgment and resulting in insufficient accuracy, this invention provides a method, system, storage medium, and computer program product for ore type classification and ore body delineation.

[0005] This invention employs the following technical solution: a method for ore type classification and ore body delineation, comprising: acquiring multiple sets of exploration samples from an ore deposit and determining their ore types, and using these as a sample set. The target element content data of each sample in the sample set is obtained through p-XRF; after standardizing the content data, the content data of each target element in each sample is used as a feature variable, and the known ore types in the ore deposit are used as the true labels. A machine learning classification model is constructed and trained to obtain a trained machine learning classification model. The target element content data of an exploration sample of an unknown ore type in any borehole is obtained through p-XRF, and after standardization, it is input into the trained machine learning classification model to output the ore type to which the exploration sample belongs. The ore types corresponding to the exploration samples in each borehole are mapped to the corresponding borehole columnar section according to their spatial location, forming a borehole columnar section with ore type layering labels. Based on the borehole columnar section with ore type layering labels, the ore body boundaries are delineated and connected in combination with geological constraints to draw a geological profile of the deposit. Based on the geological profile, an ore body distribution map is compiled to obtain the spatial distribution characteristics of different ore types in the deposit.

[0006] As a further improvement of the present invention, the known ore types in the deposit are classified and defined based on the deposit exploration report and ore characteristics. The principle for classifying ore types in the deposit is as follows: based on geological research, the ore types of the deposit are classified according to industrial indicators, ore characteristics, mineral composition, copper and sulfur grades and the content of harmful element magnesium.

[0007] As a further improvement of the present invention, the ore types in the deposit are divided into the following nine categories: (1) Copper-bearing pyrrhotite serpentine ore: Cu>0.5%, S>12%, Mg>5%; (2) High-grade copper-bearing pyrrhotite ore: Cu>1%, S>12%, Mg<1%; (3) Copper-bearing pyrrhotite ore: 0.5% <Cu<1%,S> 10%, Mg<1%; (4) Copper-bearing diorite type ore: Cu>0.5%, S<6%, Si>14%, Al>5%; (5) Copper-bearing skarn type ore: Cu>0.5%, S<6%, Mg<1%; <Mg<5%,Mn> 0.1%; (6) Copper-bearing siltstone type ore: Cu>0.5%, S<6%, Si>28%, Ca<2%, Mg<1%; (7) Pyrrhotite ore: Cu<0.3%, S>20%, Mg<1%; (8) Chalcopyrite serpentine ore: Cu>0.5%, S<8%, Mg>5%; (9) Pyrrhotite magnetite ore: Cu>0.5%, S>12%, Fe>50%.

[0008] As a further improvement of the present invention, the target elements include Cu, S, Mg, Al, Si, Fe, Ca, Mn, Mo, W, Zn, Ag, Sr, Ti, Cr, Co, Ni, As, Se, Rb, Y, Zr, Nb, Cd, Sn, Sb, Hg, Pb, Bi, Th, and U; wherein, Cu, S, Mg, Al, Si, and Fe are defined as the main limiting elements for ore type classification; and Mo, W, Zn, Ag, Sr, Ti, Cr, Co, Ni, and As are defined as the secondary limiting elements for ore type classification.

[0009] As a further improvement of the present invention, each exploration sample needs to be pretreated before p-XRF measurement. The pretreatment process is as follows: after crushing the exploration sample to 200 mesh, it is then subjected to reduction and drying treatment in sequence; subsequently, the dried exploration sample is placed in a powder press and pressed into shape to obtain the pretreated exploration sample piece.

[0010] As a further improvement of the present invention, the target element content data in each exploration sample obtained by p-XRF needs to be corrected before being input into the machine learning classification model, and the corrected target element content data is used as a feature variable to be input into the machine learning classification model.

[0011] As a further improvement of the present invention, the correction process includes: selecting representative exploration samples with elemental content, measuring them using the dissolution method to obtain the content data of each target element and using them as reference values; establishing a linear regression model between the elemental content measured by p-XRF and the reference values ​​measured by the dissolution method, and correcting the original data measured by p-XRF so that the correlation coefficient between the corrected data and the reference values ​​reaches 0.95 or higher.

[0012] As a further improvement of the present invention, the process of establishing the machine learning classification model is as follows: using the Jupyter Notebook platform, the Python machine learning algorithm is written based on the machine learning libraries SKlearn and pyTorch, and the model is established using four machine learning algorithms: lightGBM, random forest, support vector machine, and neural network, and the seven-fold cross-validation method is used.

[0013] This invention also provides an ore type classification system, which employs the aforementioned ore type classification and ore body delineation methods. The ore type classification system includes a sample acquisition module, a data collection and processing module, a model building and training module, a prediction module, and a drawing module. The sample acquisition module acquires multiple sets of exploration samples from the ore deposit and determines their corresponding ore types, thereby constructing a sample set. The data collection and processing module acquires the target element content data of each sample in the sample set using p-XRF and standardizes the content data. The model building and training module uses the target element content data from each sample obtained by the data collection and processing module as feature variables, and the known ore types in the ore deposit as true labels to construct a machine learning classification model and complete training, resulting in a trained machine learning classification model. The prediction module, based on the trained machine learning classification model, predicts the ore type of any exploration sample with an unknown ore type in any borehole and outputs the ore type corresponding to that exploration sample. The drawing module first maps the ore types corresponding to the exploration samples in each borehole to the corresponding borehole histograms according to their spatial locations, forming borehole histograms with stratified ore type labels. Based on the borehole columnar section with ore type layering and combined with geological constraints, the boundary of the ore body is delineated and connected, and a geological profile of the deposit is drawn. Based on the geological profile, an ore body distribution map is compiled to obtain the spatial distribution characteristics of different ore types in the deposit.

[0014] The present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the above-described method for classifying ore types and delineating ore bodies.

[0015] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the above-described method for classifying ore types and delineating ore bodies.

[0016] The technical solution provided by this invention has the following beneficial effects: (1) The ore type classification and ore body delineation method provided by this invention organically integrates p-XRF technology with machine learning algorithms to construct a new ore type discrimination model. A portable p-XRF device enables rapid on-site acquisition of sample element content, and combined with a machine learning model trained on large-scale data, intelligent identification and classification of ore types are achieved. Compared with traditional methods, this method effectively overcomes the problems of strong subjectivity and low efficiency in manual logging, while continuously optimizing model performance based on actual mine production, exhibiting good adaptability and scalability. Furthermore, through comprehensive mining and analysis of multi-source geochemical data, this method provides a more scientific and reliable basis for ore body boundary delineation and ore type classification, significantly improving the accuracy and credibility of mineral resource evaluation.

[0017] (2) The ore type classification and ore body delineation method provided by this invention achieves full-process coverage from sample preparation to intelligent ore type identification through multi-step collaboration. Practical application shows that this method can not only significantly improve the efficiency and accuracy of ore type classification, but also provide strong technical support for resource evaluation and production management of mining enterprises. Especially under the conditions of large-scale ore deposit exploration and complex ore body development, the advantages of this method are more prominent. By integrating modern analytical testing technology and artificial intelligence algorithms, it realizes rapid, objective and precise identification of ore types, provides new means for ore body structure analysis and efficient resource utilization, and further expands the technical path of intelligent development of mineral resources, with good application prospects and promotion value.

[0018] (3) The ore type classification and ore body delineation method provided by this invention enables timely adjustment of ore beneficiation process parameters and dynamic optimization of production processes through rapid identification of ore types during the beneficiation process, thereby further improving resource utilization efficiency and economic benefits. The application of this technology provides strong technical support for mining enterprises to achieve refined management and intelligent production. Attached Figure Description

[0019] Figure 1 The flowchart illustrates the steps of the ore type classification and ore body delineation method provided in Embodiment 1 of the present invention.

[0020] Figure 2 This is a simplified flowchart illustrating the sample preprocessing, sample data acquisition, and data processing stages for inputting into a machine learning classification model in the ore type classification and ore body delineation method provided in Embodiment 1 of the present invention.

[0021] Figure 3 This is a schematic diagram of the training and prediction process of the machine learning classification model in Embodiment 1 of the present invention. Detailed Implementation

[0022] The present invention will now be further described in conjunction with specific embodiments. It should be noted that, without conflict, the various embodiments or technical features described below can be arbitrarily combined to form new embodiments.

[0023] In the description of this invention, it should be noted that directional terms such as "center," "lateral," "longitudinal," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," and "counterclockwise," etc., indicate the orientation and positional relationship based on the orientation or positional relationship shown in the accompanying drawings. These are used only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. They should not be construed as limiting the specific scope of protection of this invention. The terms "first," "second," etc., in the specification and claims of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. The terms "comprising" and "having," and any variations thereof, in the specification and claims of this invention, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to these processes, methods, products, or devices.

[0024] Example 1 This embodiment provides a method for ore type classification and ore body delineation, using the Dongguashan deposit as an example. This embodiment collects fine ore fragments from various boreholes during ore exploration as research samples, obtaining 3200 ore samples from 138 boreholes. In subsequent ore type classification research, 2528 of these ore samples were selected as training and analysis samples, accounting for approximately 40% of the total basic analysis samples from the mine, demonstrating good representativeness and statistical significance.

[0025] like Figure 1 and Figure 2 As shown, the methods for classifying ore types and delineating ore bodies include the following steps: (a) Classification and definition of ore types in mineral deposits Based on the preliminary geological research, and according to the characteristics of industrial indicators, ore features, mineral composition, copper and sulfur grades, and the content of harmful element magnesium, the ore types of this deposit are divided into the following 9 categories: (1) Copper-bearing pyrrhotite serpentine ore: Cu>0.5%, S>12%, Mg>5%; (2) High-grade copper-bearing pyrrhotite ore: Cu>1%, S>12%, Mg<1%; (3) Copper-bearing pyrrhotite ore: 0.5% <Cu<1%, S> 10%, Mg<1%; (4) Copper-bearing diorite type ore: Cu>0.5%, S<6%, Si>14%, Al>5%; (5) Copper-bearing skarn type ore: Cu>0.5%, S<6%, Mg<1%; <Mg<5%, Mn> 0.1%; (6) Copper-bearing siltstone type ore: Cu>0.5%, S<6%, Si>28%, Ca<2%, Mg<1%; (7) Pyrrhotite ore: Cu<0.3%, S>20%, Mg<1%; (8) Chalcopyrite serpentine ore: Cu>0.5%, S<8%, Mg>5%; (9) Pyrrhotite magnetite ore: Cu>0.5%, S>12%, Fe>50%.

[0026] Of these, ore type (1) accounts for approximately 20% of the total ore, while ore types (2) and (3) account for approximately 70%. Ore types (1), (2), and (3) are the main ore types in the magnesia-skarn deposit. Ore types (4) to (6) account for less than 10% of the total ore, and these three types are minor ore types in the magnesia-skarn deposit. Ore types (7) to (9) are present in very small amounts in the deposit and were only identified in the basic sample types in this study.

[0027] Furthermore, based on geological logging and petrochemical characteristics, this scheme identifies four rock types for the main top and bottom plates and interbedded rocks of the ore body, as follows: (I) Siltstone: Cu<0.3%, S<8%, Si>28%, Ca<2%, Mg<1%; (II) Skarn: Cu<0.3%, S<8%, Si>28%, Ca<2%, Mg<1%; <Mg<5%, Mn> 0.1%; (III) Marble: Cu<0.3%, S<8%, Ca >35%, Fe<2%; (IV) Diorite: Cu<0.3%, S<8%, Si >14%, Al>5%. Furthermore, considering the integrity of the ore body boundary rings and the potential future development and utilization of the mine, this scheme also classifies samples collected with copper grades greater than the boundary grade but less than the minimum industrial grade, and ore properties similar to copper-bearing pyrrhotite ore, into Category 10: Low-grade copper-bearing pyrrhotite ore: 0.3% <Cu<0.5%, S> 8%, Mg<1%, i.e., ore type (10).

[0028] When defining multiple ore types, it is necessary not only to specify the name of each ore type but also to generate corresponding label values ​​(such as "0, 1, 2") for each type. The defined ore type label values ​​can be used as real labels to input into the subsequent machine learning classification model for training and classification.

[0029] When classifying ore types, the target elements for ore classification can be rationally determined by combining the deposit type, existing geological research data, and comprehensive industrial indicators and ore characteristics. The deposit studied in this scheme is a magnesia-skarn deposit, so the target elements can be set as Cu, S, Mg, Al, Si, Fe, Ca, Mn, Mo, W, Zn, Ag, Sr, Ti, Cr, Co, Ni, As, Se, Rb, Y, Zr, Nb, Cd, Sn, Sb, Hg, Pb, Bi, Th, and U. Among these, Cu, S, Mg, Al, Si, and Fe are defined as the primary limiting elements for ore type classification; Mo, W, Zn, Ag, Sr, Ti, Cr, Co, Ni, and As are defined as secondary limiting elements for ore type classification.

[0030] (ii) Acquisition and processing of sample sets (2.1) Obtaining the sample set Multiple sets of exploration samples were obtained from the ore deposit, and the ore type corresponding to each exploration sample was determined. These samples were used as a sample set, with each exploration sample corresponding to an independent sample. The exploration samples were all collected from various boreholes within the ore deposit. During sample selection, it was ensured that the selected exploration samples covered all ore types in the deposit to guarantee the completeness and representativeness of the sample set during subsequent training and testing of the machine learning classification model, thereby improving the reliability and accuracy of the classification results.

[0031] (2.2) Sample preprocessing The exploration sample is crushed to about 200 mesh, and then subjected to reduction and drying processes. The dried exploration sample is then poured into a plastic ring (40*34*4.5mm) and flattened and compacted. It is then placed in a powder compactor to be pressed into a sample sheet and placed in a sealed bag for later use, thus obtaining the pre-treated sample.

[0032] The purpose of pressing the exploration samples is to improve the surface smoothness of the tested samples and enhance the uniformity of the internal material distribution, thereby reducing testing errors and improving the accuracy and stability of p-XRF detection results. A YY-600 laboratory powder compactor can be used.

[0033] (2.3) The content of each target element in each sample was obtained by p-XRF and then standardized. (2.3.1) Obtain the content data of each target element in each sample. By measuring the pretreated samples using p-XRF, the content data of the target element in each sample can be obtained. In this scheme, p-XRF can be performed using a portable X-ray fluorescence analyzer from the VANTA series manufactured by Olympus Technologies, Japan. This instrument can analyze 83 elements from magnesium (Mg) to plutonium (Pu) in the periodic table, and the detection limits for most elements are at the ppm level. The principle of portable X-ray fluorescence spectrometry (p-XRF) is the same as that of laboratory XRF. Its advantages are that it is lightweight, portable, fast, and highly accurate, and the elemental content of the target analyte can be obtained directly on-site. The results of p-XRF pellet analysis of powder samples show that it has a good correlation with laboratory analysis results. In this scheme, the measurement time for each p-XRF measurement can be set to 120s. When using this instrument for analysis, its working mode can be set to Geochem mode. The first exploration sample of each borehole is measured three times, and the stability and repeatability of the three analysis test results are verified to ensure that the instrument is working properly. Please refer to Table 1 for the data obtained from the repeated measurements.

[0034] Table 1. Experimental data on the stability test of p-XRF on exploration samples.

[0035] (2.3.2) Correcting the data acquired by p-XRF First, the raw data acquired by p-XRF needs to be preliminarily screened to remove interfering data. The screening rules are as follows: if more than half of the elements in a sample have a content greater than 10 times the detection limit of the corresponding element, the sample data is considered valid overall and retained. Conversely, if more than half (including half) of the elements in a sample have a content less than 10 times the detection limit of the corresponding element (i.e., low content or less than the detection limit, <LOD), the sample data quality is considered poor and the relevant data can be removed. Before performing multi-element statistical analysis, null values ​​(<LOD) also need to be processed, usually by replacing them with a fixed value less than the detection limit or a certain proportion less than the detection limit (such as 25% or 75%). The above screening and processing methods are common techniques in p-XRF data processing and will not be elaborated further in this solution.

[0036] After initial screening, the target element content data in each preprocessed sample needs to be corrected. The correction process is as follows: representative samples with high elemental content are selected, and the content of each target element is determined using the dissolution method, which is then used as a reference value. A linear regression model is established between the p-XRF determination values ​​and the reference values ​​obtained by the dissolution method to correct the original p-XRF data, ensuring that the correlation coefficients of each target element between the corrected data and the reference values ​​obtained by the dissolution method reach above 0.95.

[0037] Understandably, by performing preliminary screening and correction on the content of target elements in each preprocessed sample obtained by p-XRF, the accuracy and reliability of p-XRF data can be effectively improved. Using the corrected data as input to a machine learning classification model provides a high-quality data foundation for ore type classification.

[0038] Furthermore, after obtaining the corrected data, it is necessary to standardize the content data of each target element retained after correction. Since this scheme does not improve the standardization process, conventional standardization methods for p-XRF element content data in existing technologies can be used.

[0039] (III) Constructing and training machine learning classification models (3.1) The process of building the machine learning classification model is as follows: Using the Jupyter Notebook platform, Python machine learning algorithms are written based on the machine learning libraries SKlearn and pyTorch. Four machine learning algorithms are used: lightGBM, random forest, support vector machine, and neural network, and the seven-fold cross-validation method is used to build the model. Since this solution does not improve the construction and training of the machine learning classification model, the construction and training of the machine learning classification model can refer to the common construction and training operations of existing technologies.

[0040] (3.2) Please refer to Figure 3 Based on the preprocessed samples (i.e. all chemical analysis data of the exploration subsamples, the content of all elements contained in each exploration subsample, ore type, etc.), data feature engineering is performed. The content data of each target element in each standardized sample is set as feature variables, and the ore type value is set as the true label. The data are then input into the constructed machine learning classification model for training, thereby obtaining the trained machine learning classification model.

[0041] During the training of the machine learning classification model, the sample set is divided into a training set and a test set in a 7:3 ratio, and the hyperparameter settings are optimized using cross-validation to ensure that the model has good generalization ability.

[0042] To scientifically evaluate the performance of the machine learning classification model, this scheme selects three multi-classification evaluation metrics—classification accuracy, kappa coefficient, and Harman distance—for model validation and comparison. Classification accuracy reflects the proportion of correct classifications overall, embodying the overall classification precision. The kappa coefficient is used to eliminate the influence of uneven ore type sample distribution on the evaluation results, with a value ranging from 0 to 1; the closer the coefficient is to 1, the higher the classification consistency and the stronger the model reliability. The Harman distance represents the average difference between the predicted label and the true label; the smaller the value, the smaller the classification error and the more refined the classification result. In this scheme, the machine classification model achieves an average classification accuracy of approximately 90.02%, an average kappa coefficient of approximately 0.8837, and an average Harman distance of approximately 0.0998, thus proving that the machine learning classification model trained under this scheme can effectively classify the ore types contained in the deposit.

[0043] After the machine learning classification model was trained, data from 2872 randomly selected ore samples out of 3200 ore samples were processed according to steps (I) and (II) before being used for machine learning. Cu, S, Mg, Al, Si, Fe, and Ca in each ore sample were the main limiting elements for ore type identification, while Mo, W, Zn, Ag, Sr, Ti, Cr, Co, Ni, and As were the secondary limiting elements for ore type identification.

[0044] The elemental content data were exported from p-XRF to the computer, organized according to standard data format, and then imported into a software system integrating a machine learning classification model for calculation and analysis to obtain the classification results of ore type and surrounding rock lithology. Furthermore, the model classification results were compared and analyzed with the double-blind manual classification results, resulting in Table 2.

[0045] Table 2 Comparison of Ore Types Obtained by Trained Machine Learning Classification Model

[0046] The results in Table 2 show that the ore type classification results obtained based on the machine learning classification model are generally consistent with the geological logging results, and the classification results of the vast majority of samples are in agreement, indicating that the machine learning classification model constructed in this scheme has good classification effect and reliability.

[0047] Furthermore, through multiple cross-validation and self-testing analyses of the machine learning classification model, the classification accuracy of various ores and surrounding rock lithology can be obtained, thereby further evaluating the stability and predictive performance of the machine learning classification model, as shown in Table 3.

[0048] Table 3. Machine Learning Big Data Accuracy for Various Types of Ores and Rocks

[0049] Analysis of Table 3 shows that the machine learning classification model constructed in this scheme can effectively classify various ore types and rock types. Its classification accuracy is stable at about 94% in multiple cross-validation and self-testing analyses, indicating that the machine learning classification model has high recognition accuracy and stability and can classify different ore types and rock types relatively accurately.

[0050] Furthermore, during the training of the machine learning classification model, feature importance analysis revealed that the top 13 elements with the greatest influence on the classification results were Cu, S, Si, Mg, Ca, Ag, Al, Mn, Zn, Sr, W, As, and Fe. Comparative analysis with the geological characteristics of the ore type shows that these elements have good geological significance and rationality as discriminant indicators. Cu and S are the main beneficial elements in this deposit, ranking first and second in feature influence. Si, Mg, Al, and Ca mainly constitute gangue minerals and some ore minerals in the ore and surrounding rocks. Sr has a relatively low overall content in the ore, mainly originating from the surrounding rocks or intrusive bodies, thus serving as an important discriminant indicator for distinguishing between ore and non-ore. Ag is usually enriched in the ore, providing high indicative significance for ore identification. Fe has a low content in marble but a relatively high content in the ore, also demonstrating good discriminative ability. The above results show that the machine learning classification model constructed in this scheme not only has high classification accuracy, but also has a clear geological interpretation in its discrimination basis, further verifying the scientificity and feasibility of the ore type classification and ore body delineation method.

[0051] The ore classification method provided in this solution is based on a data-driven approach. It utilizes a portable X-ray fluorescence analyzer to collect comprehensive data from ore samples in boreholes at the Dongguashan deposit. Combined with big data analysis techniques, it systematically mines the geochemical characteristics of different ore types and surrounding rocks, providing a reliable basis for rapid ore type identification. This method significantly improves analytical efficiency and decision-making in the mining process, thereby enhancing overall production efficiency.

[0052] (iv) Predicting the type of an exploration sample of an unknown ore type within the deposit. By obtaining the content data of each target element in an exploration sample of an unknown ore type using p-XRF, and then standardizing the data before inputting it into a trained machine learning classification model, the ore type corresponding to the exploration sample can be output.

[0053] The machine learning classification model trained above can be used to automatically identify and classify exploration samples of any unknown ore type in a mineral deposit.

[0054] (v) Spatial distribution of various ore types in the deposit Based on the ore type results output by the trained machine learning classification model, manually classified ore types can be optimized and reclassified, and ore bodies can be re-delineated and their distribution maps revised. The specific operations are as follows: First, the ore type results of exploration samples from each borehole are obtained using the machine learning classification model and arranged according to depth (i.e., the spatial location of the exploration samples in the deposit). These samples are then mapped onto the corresponding borehole columnar section, forming a borehole columnar section with ore type stratification labels. This constructs continuous ore type stratification data, enabling refined reclassification of ore types. Second, using the exploration line as the basic unit, key information such as the top and bottom plate depths, thicknesses, ore types, and stratigraphic occurrences of the ore body are extracted from each borehole columnar section. This information is then projected onto the corresponding profile locations. Combining geological structural features (such as faults and folds) and the continuity of the ore body, the ore body boundaries between adjacent boreholes are delineated and connected, thereby drawing a geological profile reflecting the spatial morphology, dip angle, and distribution characteristics of the ore body. Finally, the boundaries of the ore bodies in each exploration profile are projected onto a plane and corrected by spatial interpolation and geological constraints. The discrete profile information is integrated into a continuous planar distribution area, and an ore body distribution map is compiled to intuitively reflect the planar location, distribution range, orientation and scale characteristics of the ore bodies.

[0055] Through the above process, existing orebody distribution maps can be revised, significantly improving the accuracy of orebody boundaries and the spatial distribution of ore types. Simultaneously, this method effectively enhances the efficiency of orebody delineation and reduces interference from subjective human factors. In practical applications, combining machine learning classification results with geological logging data further enhances the scientific rigor and reliability of orebody interpretation. Furthermore, this method achieves intelligent integration of the entire process from borehole data to the visualization of orebody spatial distribution. The resulting orebody distribution map intuitively reflects the spatial distribution characteristics of different ore types, providing a reliable basis for mine design. Moreover, statistical analysis of the spatial distribution patterns of various ores further verifies the rationality and stability of the machine learning classification results.

[0056] Furthermore, this solution first utilizes p-XRF equipment to conduct on-site elemental content determination of the mineral processing products, obtaining target element data. Subsequently, the data is input into a pre-trained machine learning classification model, which rapidly obtains ore type identification results through intelligent analysis. Compared to traditional laboratory analysis methods, this method significantly shortens the analysis cycle, effectively reduces human error, and improves the objectivity and accuracy of the identification results. Simultaneously, the rapid identification of ore types in the mineral processing products enables dynamic adjustment of mineral processing parameters and optimization of the production process, thereby further improving resource utilization and economic benefits, and providing strong technical support for the refined management and intelligent production of mining enterprises. In addition, this solution, through in-depth mining and comprehensive analysis of multi-source geochemical data, provides a more scientific basis for orebody boundary delineation and ore type classification, significantly improving the accuracy and reliability of mineral resource evaluation.

[0057] In summary, this solution organically integrates p-XRF technology with a machine learning classification model to construct a novel ore type discrimination mode. Utilizing a portable p-XRF device, it enables rapid on-site acquisition of elemental content data, and combined with a machine learning classification model trained on large datasets, achieves rapid and intelligent ore type classification. This method not only effectively overcomes the problems of high subjectivity and low efficiency associated with traditional manual logging, but also allows for continuous optimization of model performance based on actual mine needs. Furthermore, this solution achieves full-process coverage from sample preparation to intelligent ore type discrimination through multi-stage collaboration. In practical applications, it can significantly improve the efficiency and accuracy of ore type classification and provide reliable support for mine resource evaluation and production management. Its advantages are particularly prominent under conditions of large-scale ore deposit exploration and complex ore body development. By integrating modern analytical testing technologies with artificial intelligence methods, it provides a new technical path for the intelligent development of mineral resources, demonstrating promising application prospects and widespread application value.

[0058] Example 2 Based on Example 1, this solution provides an ore type classification and ore body delineation system, which adopts the ore type classification and ore body delineation method of Example 1. The ore type classification and ore body delineation system includes a sample acquisition module, a data collection and processing module, a model building and training module, a prediction module, and a drawing module.

[0059] The sample acquisition module is used to acquire multiple sets of exploration samples from the ore deposit and determine the ore type corresponding to each exploration sample, thereby constructing a sample set. These exploration samples are all collected from boreholes within the ore deposit. During sample selection, it should be ensured that the samples cover all ore types in the ore deposit to guarantee the completeness and representativeness of subsequent model training and testing.

[0060] The data collection and processing module is used to acquire the target element content data of each sample in the sample set through p-XRF and to standardize the data. This standardized data can be directly input into the machine learning classification model for training and analysis, thus providing a standardized data foundation for model construction.

[0061] The model building and training module, based on the Jupyter Notebook platform, utilizes the Python programming language and machine learning libraries (such as sklearn and PyTorch) to construct classification models using various algorithms including LightGBM, Random Forest, Support Vector Machine, and Neural Networks. A seven-fold cross-validation method is then used for model training and evaluation. Specifically, the content of target elements in each sample obtained from the data acquisition and processing module is used as feature variables, and known ore types in the deposit are used as ground truth labels to construct and train a machine learning classification model. During training, the dataset is divided into training and test sets in a 7:3 ratio, and cross-validation is used to optimize the model's hyperparameters to improve its generalization ability and stability.

[0062] The prediction module, based on a trained machine learning classification model, predicts the type of ore in any exploration sample of unknown ore type in any borehole and outputs the corresponding ore type. This module enables rapid and automatic identification based on p-XRF elemental content data, effectively replacing traditional manual visual identification methods and significantly improving the efficiency and accuracy of ore type classification.

[0063] The drawing module first maps the ore types corresponding to the exploration samples from each borehole to the corresponding borehole columnar section based on their spatial location, generating a borehole columnar section with ore type stratification labels. Based on the borehole columnar section with ore type stratification labels and combined with geological constraints, the orebody boundaries are delineated and connected, and a geological profile of the deposit is drawn. Finally, an orebody distribution map is compiled based on the geological profile, thereby obtaining the spatial distribution characteristics of various ore types in the deposit. This process not only improves the efficiency of orebody delineation and connection but also effectively reduces subjective errors introduced by human factors. The resulting orebody distribution map can intuitively reflect the spatial distribution patterns of ore types, providing a scientific basis for mine design.

[0064] Example 3 This embodiment discloses a readable storage medium storing computer program instructions. When the computer program instructions are read and executed by a processor, the steps of the ore type classification and ore body delineation method in Embodiment 1 are performed.

[0065] In Example 1, the method for classifying ore types and delineating ore bodies can be applied in the form of software, such as by designing a program that can run independently on a computer-readable storage medium, which can be a USB flash drive. The program can be designed to start the entire method via an external trigger.

[0066] Example 4 This embodiment provides a computer program product, which includes a computer program. When the computer program is executed by a processor, it implements the ore type classification and ore body delineation method as described in Embodiment 1.

[0067] The ore type classification and ore body delineation method in Example 1 can be applied in the form of software, such as by designing a computer-readable storage medium that can run independently. The computer-readable storage medium can be a USB flash drive, and the program can be designed to start the entire method through an external trigger.

[0068] The computer program product provided in this embodiment is essentially a computer device used to implement the ore type classification and ore body delineation method in Embodiment 1. This computer device can be an embedded module, or it can be a smart terminal, tablet computer, laptop computer, desktop computer, rack server, blade server, tower server, or cabinet server (including independent servers or server clusters composed of multiple servers), etc., capable of executing programs.

[0069] The computer device mentioned in this embodiment includes, but is not limited to, a memory and a processor that can be interconnected via a system bus.

[0070] In this embodiment, the memory (i.e., the readable storage medium) includes flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, disk, optical disk, etc. In some embodiments, the memory can be an internal storage unit of a computer device, such as the hard disk or RAM of the computer device. In other embodiments, the memory can also be an external storage device of the computer device, such as a plug-in hard disk, smart media card (SMC), secure digital card (SD), flash card, etc., equipped on the computer device. Of course, the memory can also include both internal storage units and external storage devices of the computer device. In this embodiment, the memory is typically used to store the operating system and various application software installed on the computer device. In addition, the memory can also be used to temporarily store various types of data that have been output or will be output.

[0071] In some embodiments, the processor may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chip. The processor is typically used to control the overall operation of a computer device. In this embodiment, the processor is used to run program code stored in memory or process data to implement the steps of the ore type classification and ore body delineation method in Embodiment 1.

[0072] The basic principles, main features, and advantages of this invention have been described above. Those skilled in the art should understand that this invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely principles of the invention. Various changes and modifications can be made without departing from the spirit and scope of the invention, and all such changes and modifications fall within the scope of the invention as claimed. The scope of protection claimed by this invention is defined by the appended claims and their equivalents.

Claims

1. A method for classifying ore types and delineating ore bodies, characterized in that, It includes: Multiple sets of exploration samples from the ore deposit were obtained and their ore types were determined, and these were used as a sample set. The target element content data of each sample in the sample set is obtained by p-XRF; after standardizing the content data, the content data of each target element in each sample is used as the feature variable, and the known ore type in the deposit is used as the true label. A machine learning classification model is constructed and trained to obtain the trained machine learning classification model. The p-XRF method obtains the target element content data of an exploration sample of unknown ore type in any borehole, and after standardization, it is input into a trained machine learning classification model to output the ore type corresponding to the exploration sample. The ore types corresponding to the exploration samples in each borehole are mapped to the corresponding borehole columnar section according to their spatial location, forming a borehole columnar section with ore type layering labels. Based on the borehole columnar section with ore type layering labels, the ore body boundaries are delineated and connected in combination with geological constraints, thereby drawing a geological profile of the deposit. Based on the geological profile, an ore body distribution map is compiled to obtain the spatial distribution characteristics of different ore types in the deposit.

2. The method for classifying ore types and delineating ore bodies as described in claim 1, characterized in that, The known ore types in the deposit are classified and defined based on the deposit exploration report and ore characteristics. The principle for classifying ore types in the deposit is as follows: based on geological research, the ore types of the deposit are classified according to industrial indicators, ore characteristics, mineral composition, copper and sulfur grades, and the content of harmful element magnesium.

3. The method for classifying ore types and delineating ore bodies as described in claim 1, characterized in that, The ore types in the deposit are classified into the following nine categories: (1) Copper-bearing pyrrhotite serpentine ore: Cu>0.5%, S>12%, Mg>5%; (2) High-grade copper-bearing pyrrhotite ore: Cu>1%, S>12%, Mg<1%; (3) Copper-bearing pyrrhotite ore: 0.5% <Cu<1%,S> 10%, Mg<1%; (4) Copper-bearing diorite type ore: Cu>0.5%, S<6%, Si>14%, Al>5%; (5) Copper-bearing skarn type ore: Cu>0.5%, S<6%, Mg<1%; <Mg<5%,Mn> 0.1%; (6) Copper-bearing siltstone type ore: Cu>0.5%, S<6%, Si>28%, Ca<2%, Mg<1%; (7) Pyrrhotite ore: Cu<0.3%, S>20%, Mg<1%; (8) Chalcopyrite serpentine ore: Cu>0.5%, S<8%, Mg>5%; (9) Pyrrhotite magnetite ore: Cu>0.5%, S>12%, Fe>50%.

4. The method for classifying ore types and delineating ore bodies as described in claim 1, characterized in that, The target elements include Cu, S, Mg, Al, Si, Fe, Ca, Mn, Mo, W, Zn, Ag, Sr, Ti, Cr, Co, Ni, As, Se, Rb, Y, Zr, Nb, Cd, Sn, Sb, Hg, Pb, Bi, Th, and U; among them, Cu, S, Mg, Al, Si, and Fe are defined as the main limiting elements for ore type classification; Mo, W, Zn, Ag, Sr, Ti, Cr, Co, Ni, and As are defined as the secondary limiting elements for ore type classification.

5. The method for classifying ore types and delineating ore bodies as described in claim 1, characterized in that, Each exploration sample must be pretreated before p-XRF analysis. The pretreatment process is as follows: the exploration sample is crushed to 200 mesh, then reduced in size and dried. The dried exploration sample is then pressed into shape in a powder press to obtain the pretreated exploration sample piece.

6. The method for classifying ore types and delineating ore bodies as described in claim 1, characterized in that, The target element content data in each exploration sample obtained by p-XRF need to be corrected before being input into the machine learning classification model, and the corrected target element content data are then used as feature variables to be input into the machine learning classification model.

7. The method for classifying ore types and delineating ore bodies as described in claim 6, characterized in that, The correction process includes: selecting representative exploration samples with elemental content, measuring them using the dissolution method to obtain the content data of each target element and using it as a reference value; establishing a linear regression model between the elemental content measured by p-XRF and the reference value measured by the dissolution method, and correcting the original data measured by p-XRF so that the correlation coefficient between the corrected data and the reference value reaches above 0.

95.

8. A system for classifying ore types and delineating ore bodies, characterized in that, It employs the ore type classification and ore body delineation method as described in any one of claims 1-7; the ore type classification and ore body delineation system includes: The sample acquisition module is used to acquire multiple sets of exploration samples from the ore deposit, determine their corresponding ore types, and then construct a sample set. The data collection and processing module is used to acquire the target element content data of each sample in the sample set through p-XRF and to standardize the content data. The model building and training module is used to take the target element content data in each sample obtained by the data collection and processing module as feature variables, take the known ore type in the deposit as the real label, build a machine learning classification model and complete the training to obtain the trained machine learning classification model. The prediction module, based on a trained machine learning classification model, predicts the type of ore in any exploration sample of unknown ore type in any borehole and outputs the ore type corresponding to the exploration sample. The drawing module is used to first map the ore types corresponding to the exploration samples in each borehole to the corresponding borehole columnar section according to their spatial location, forming a borehole columnar section with ore type layering labels; based on the borehole columnar section with ore type layering labels and combined with geological constraints, the boundary of the ore body is delineated and connected, and a geological profile of the deposit is drawn; based on the geological profile, an ore body distribution map is compiled to obtain the spatial distribution characteristics of different ore types in the deposit.

9. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the ore type classification and ore body delineation method as described in any one of claims 1-7.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the ore type classification and ore body delineation method as described in any one of claims 1-7.

Images