Methods and equipment for probabilistic determination of tectonic environments based on multi-source geochemical data

Abstract

This invention provides a method and device for probabilistic tectonic environment discrimination based on multi-source geochemical data, relating to the field of geological exploration technology. It performs quality screening on whole-rock, zircon, and apatite data based on metadata of the target igneous rock sample, reducing the impact of raw data quality differences on the discrimination results from the data source. The detection limits of the screened whole-rock element vectors, selected zircon and apatite point data are made consistent through metadata to improve comparability across batches and laboratories and suppress spurious correlations. Features are calculated and aggregated for the transformed whole-rock element vectors, transformed zircon and apatite point data respectively, addressing the problems of non-standard multi-point data fusion and the inability to quantify cross-evidence consistency, thereby improving the stability of tectonic environment discrimination from multiple aspects.

Description

Technical Field

[0001] This invention relates to the field of geological exploration technology, and in particular to a method and device for probabilistic determination of tectonic environments based on multi-source geochemical data. Background Technology

[0002] Current engineering approaches for identifying the tectonic environment of magmatic rocks can be broadly categorized into three types: The first type is a system of discrimination diagrams and empirical rules based on whole-rock main beams and trace elements as core evidence. This approach relies on a large amount of regional geochemical and petrological research and can quickly provide environmental indications under conditions where sample information is relatively complete and alteration effects are controllable. Therefore, it is still widely used in regional tectonic analysis, magmatic rock correlation, and metallogenic geology research. The second type is a discrimination method based on the trace element and rare earth element distribution curves of accessory minerals (especially zircon and apatite). Zircon, due to its high sealing properties and sensitivity to magma source and evolution, is often used as an indicator of arc-related intraplate magmatic processes. Apatite has a strong response to volatile matter, oxygen fugacity, and magma evolution stages, and also has good environmental indication value in basic-intermediate magmatic systems. The third type is statistical learning / machine learning discrimination methods for big data. These methods typically use public databases or regional datasets as training samples and achieve automatic discrimination through feature engineering and classifiers. Some studies have begun to attempt to incorporate whole-rock and mineral information simultaneously.

[0003] However, the above approaches generally face several structural bottlenecks in practical applications, resulting in insufficient stability in situations involving "adjacent environmental types" or "multiple processes superimposed".

[0004] First, single data sources have a significant systematic bias: whole-rock composition is easily affected by multiple factors such as later alteration, metasomatism, contamination, and fractional crystallization, especially in ancient or multi-stage altered areas, where primary magma signals may be significantly rewritten; in contrast, zircon and apatite, as accessory minerals, can retain chemical information of the magma crystallization stage to some extent, but they record a "selective process window" and may not be strictly isomorphic to the whole rock in terms of time scale and material source, so relying solely on mineral evidence may also lead to representative bias.

[0005] Secondly, the "compositional data" attribute of geochemical data is often underestimated in existing methods: whether it is whole-rock major / trace elements or mineral trace elements, they are inherently affected by closed constraints. When directly using concentration values ​​or simple ratios for statistical analysis and machine learning, it is easy to introduce spurious correlations and scale biases, causing the model to perform well on the training set but with decreased generalization on cross-regional and cross-laboratory data. At the same time, the detection limits, background subtraction, and below-detection-limit processing rules of different laboratories vary significantly. Without a consistent strategy, the incomparability between samples will be further amplified, causing training-application bias, thereby affecting the stability and reproducibility of classification conclusions.

[0006] Furthermore, the fusion of multi-site and multi-particle data lacks a unified standard: in-situ mineral analysis, represented by micro-area elemental and isotopic analysis techniques, typically produces a hierarchical data structure of "site-particle-sample"; the same sample may contain multiple zircons, and each zircon may have multiple sites, as does apatite; common processing methods in existing studies include simple mean or median substitution, or selecting "representative particles / sites" based on experience. These approaches are feasible in the scientific research exploration stage, but they bring two types of problems in patented and engineering application scenarios: first, a lack of robustness, as extreme values, inclusion contamination, or local modifications can significantly affect the statistics; second, a lack of traceability, as it is difficult to determine which sites are included, how to exclude them, and how to set the parameters, making it difficult to verify and reproduce the results.

[0007] Finally, existing methods generally lack uncertainty representation and risk control mechanisms: whether it's discriminant diagrams, rule thresholds, or most classification models, they often aim to "give a unique category" as the output goal, lacking mechanisms such as posterior probability, conflict warnings, and rejection thresholds. In practical applications, when evidence is insufficient, data is missing, or different evidence sources point to inconsistent conclusions, forcibly outputting a single category will significantly increase the risk of misjudgment and cannot provide operational guidance for the next step (e.g., whether to supplement zircon or apatite testing, which key element groups should be added, whether there is mixed magma or multi-stage superimposed signals). Summary of the Invention

[0008] This invention provides a method and device for probabilistic identification of tectonic environments based on multi-source geochemical data, with the aim of improving the stability of tectonic environment identification.

[0009] To achieve the above objectives, this invention provides a method for probabilistic determination of tectonic environments based on multi-source geochemical data, comprising:

[0010] Step 1: Perform elemental analysis on the target igneous rock sample to obtain whole-rock elemental vectors, zircon location data, apatite location data, and metadata;

[0011] Step 2: Perform quality screening on the whole-rock element vector based on metadata to obtain the screened whole-rock element vector. Perform point-level screening on the zircon and apatite point data to obtain the screened zircon and apatite point data.

[0012] Step 3: Based on metadata, perform detection limit consistency and data transformation on the screened whole-rock element vector, screened zircon point data and screened apatite point data to obtain the transformed whole-rock element vector, transformed zircon point data and transformed apatite point data.

[0013] Step 4: Construct principal combination features and trace element ratio features and calculate rare earth curve morphology parameters for the transformed whole-rock element vector to obtain whole-rock features. Then, calculate the features of the transformed zircon point data and the transformed apatite point data to obtain zircon point features and apatite point features.

[0014] Step 5: Aggregate the zircon and apatite location features using the central statistic and dispersion index respectively to obtain zircon sample-level feature vectors and apatite sample-level feature vectors. Then, concatenate the whole-rock features with the zircon and apatite sample-level feature vectors to obtain the fused basic feature vector of the target igneous rock sample.

[0015] Step 6: Calculate the coupling characteristics of zircon-whole rock and apatite-whole rock using the rare earth curve morphology parameters of the whole rock element vector, zircon location characteristics, and apatite location characteristics, and calculate the consistency score based on the coupling characteristics of zircon-whole rock and apatite-whole rock.

[0016] Step 7: Incorporate the zircon-whole-rock coupling features, apatite-whole-rock coupling features, and consistency scores into the fusion basic feature vector to obtain the fusion feature vector of the target igneous rock. Then, input the fusion feature vector into the trained probability classification model for discrimination to obtain the posterior probability vectors of eight tectonic environments.

[0017] Furthermore, step 1 includes:

[0018] Whole-rock major and trace element analysis was performed on the target igneous rock sample to obtain whole-rock major and trace element data, and whole-rock element vectors were constructed based on the whole-rock major and trace element data.

[0019] In-situ trace element analysis of zircon and apatite in the target igneous rock sample was performed to obtain zircon position data and apatite position data.

[0020] Obtain metadata of the target igneous rock sample, including sample number, rock mass number, test batch, standard material, detection limits of each element, and analytical error.

[0021] Furthermore, a quality screening of the whole-rock element vector is performed to obtain the screened whole-rock element vector, including:

[0022] Find the analysis error corresponding to each element in the whole-rock element vector in the metadata;

[0023] An error threshold is set, and elements whose analysis error exceeds the error threshold are processed to obtain the whole rock element vector after initial screening.

[0024] Elements that exceed the preset range in the whole-rock element vector after initial screening are marked to obtain the marked elements;

[0025] Outlier removal is performed on the labeled elements in the whole-rock element vector after initial screening to obtain the screened whole-rock element vector.

[0026] Furthermore, the zircon and apatite point data are filtered at the point level to obtain filtered zircon and apatite point data, including:

[0027] Centered on the median, the absolute median difference between the zircon and apatite point data is calculated as the first and second dispersions, respectively.

[0028] The deviation of any point in the zircon point data from key features or key elements is evaluated based on the first degree of dispersion. If the deviation is greater than the first degree of dispersion, the point is identified as an outlier and is removed or downweighted to obtain the filtered zircon point data.

[0029] The deviation of any point in the apatite point data from key features or key elements is evaluated based on the second degree of dispersion. If the deviation is greater than the second degree of dispersion, the point is identified as an outlier and is removed or downweighted to obtain the filtered apatite point data.

[0030] Furthermore, step 3 includes:

[0031] Based on the detection limits of each element in the metadata, the elements in the screened whole-rock element vector, the screened zircon point data, and the screened apatite point data are screened to determine the elements below the detection limit.

[0032] Elements below the detection limit were uniformly substituted to obtain whole-rock element vectors, zircon location data, and apatite location data after element substitution.

[0033] Pseudo-counts are added to the zero or minimum values ​​in the whole-rock element vector, zircon point data and apatite point data after element substitution to obtain the whole-rock element vector, zircon point data and apatite point data with pseudo-counts added.

[0034] Select elements from the whole-rock element vector, zircon point data and apatite point data with added pseudo-counts and form vectors to obtain the element composition vector.

[0035] The elemental composition vector is transformed to obtain the transformed whole-rock elemental vector, the transformed zircon point data, and the transformed apatite point data.

[0036] Furthermore, after step 7, the following is also included:

[0037] Pratt scaling calibration or ordinal-preserving regression calibration is used to calibrate the posterior probability vectors of eight types of constructed environments.

[0038] The present invention also provides a tectonic environment probability discrimination device based on multi-source geochemical data, comprising:

[0039] The analysis module is used to perform elemental analysis on the target igneous rock sample to obtain whole-rock elemental vectors, zircon position data, apatite position data, and metadata.

[0040] The screening module is used to perform quality screening on the whole-rock element vector based on metadata, to obtain the screened whole-rock element vector, and to perform point-level screening on zircon point data and apatite point data, to obtain the screened zircon point data and screened apatite point data.

[0041] The transformation module is used to perform detection limit consistency and data transformation on the screened whole-rock element vector, screened zircon point data and screened apatite point data based on metadata, so as to obtain the transformed whole-rock element vector, transformed zircon point data and transformed apatite point data.

[0042] The calculation module is used to construct the principal combination features and trace element ratio features of the transformed whole-rock element vector and calculate the rare earth curve morphology parameters to obtain whole-rock features. It also calculates the features of the transformed zircon point data and the transformed apatite point data to obtain zircon point features and apatite point features.

[0043] The splicing module is used to aggregate zircon and apatite location features using central statistics and dispersion index respectively, to obtain zircon sample-level feature vectors and apatite sample-level feature vectors. The whole-rock features are then spliced ​​with the zircon sample-level feature vectors and apatite sample-level feature vectors to obtain the fused basic feature vector of the target igneous rock sample.

[0044] The coupling module is used to calculate the coupling characteristics of zircon-whole rock and apatite-whole rock using the rare earth curve morphology parameters of the whole rock element vector, zircon location characteristics and apatite location characteristics, and to calculate the consistency score based on the coupling characteristics of zircon-whole rock and apatite-whole rock.

[0045] The discrimination module is used to incorporate the coupling features of zircon-whole rock, apatite-whole rock and consistency scores into the fusion basic feature vector to obtain the fusion feature vector of the target igneous rock. The fusion feature vector is then input into the trained probability classification model for discrimination to obtain the posterior probability vectors of eight types of tectonic environments.

[0046] Furthermore, it also includes:

[0047] The calibration module is used to calibrate the posterior probability vectors of eight constructed environments using Pratt scaling calibration or ordinal regression calibration.

[0048] The present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements a method for probabilistic determination of tectonic environments based on multi-source geochemical data.

[0049] The present invention also provides a terminal device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements a structural environment probability discrimination method based on multi-source geochemical data.

[0050] The above-described solution of the present invention has the following beneficial effects:

[0051] Compared with existing technologies, this invention performs quality screening on whole-rock, zircon, and apatite data based on the metadata of the target igneous rock sample, reducing the impact of differences in the quality of the original data on the discrimination results from the source. By using metadata to ensure consistency of the detection limits of the screened whole-rock element vector, the screened zircon point data, and the screened apatite point data, the comparability across batches and laboratories is improved and spurious correlations are suppressed. Features are calculated and aggregated for the transformed whole-rock element vector, the transformed zircon point data, and the transformed apatite point data, respectively, to solve the problems of non-standard fusion of multi-point data and the inability to quantify cross-evidence consistency, thereby improving the discrimination stability of tectonic environment from multiple aspects.

[0052] Other beneficial effects of the present invention will be described in detail in the following detailed description section. Attached Figure Description

[0053] Figure 1 This is a flowchart illustrating an embodiment of the present invention;

[0054] Figure 2 This is a schematic diagram of the structure of the environmental probability discrimination device in an embodiment of the present invention;

[0055] Figure 3 This is a schematic diagram of the structure of the terminal device in an embodiment of the present invention. Detailed Implementation

[0056] To make the technical problems, solutions, and advantages of this invention clearer, a detailed description will be provided below with reference to the accompanying drawings and specific embodiments. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0057] In the description of this invention, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing the invention and for 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. Therefore, they should not be construed as limitations on the invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0058] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a locking connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0059] Furthermore, the technical features involved in the different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0060] This invention addresses existing problems by providing a method and device for probabilistic determination of structural environments based on multi-source geochemical data.

[0061] like Figure 1 As shown, embodiments of the present invention provide a method for probabilistic determination of tectonic environments based on multi-source geochemical data, including:

[0062] Step 1: Perform elemental analysis on the target igneous rock sample to obtain whole-rock elemental vectors, zircon location data, apatite location data, and metadata;

[0063] Step 2: Perform quality screening on the whole-rock element vector based on metadata to obtain the screened whole-rock element vector. Perform point-level screening on the zircon and apatite point data to obtain the screened zircon and apatite point data.

[0064] Step 3: Based on metadata, the detection limit is made consistent and data transformation is performed on the screened whole-rock element vector, the screened zircon point data and the screened apatite point data to obtain the transformed whole-rock element vector, the transformed zircon point data and the transformed apatite point data.

[0065] Step 4: Construct principal combination features and trace element ratio features and calculate rare earth curve morphology parameters for the transformed whole-rock element vector to obtain whole-rock features. Then, calculate the features of the transformed zircon point data and the transformed apatite point data to obtain zircon point features and apatite point features.

[0066] Step 5: Aggregate the zircon and apatite location features using the central statistic and dispersion index respectively to obtain zircon sample-level feature vectors and apatite sample-level feature vectors. Then, concatenate the whole-rock features with the zircon and apatite sample-level feature vectors to obtain the fused basic feature vector of the target igneous rock sample.

[0067] Step 6: Calculate the coupling characteristics of zircon-whole rock and apatite-whole rock using the rare earth curve morphology parameters of the whole rock element vector, zircon location characteristics, and apatite location characteristics, and calculate the consistency score based on the coupling characteristics of zircon-whole rock and apatite-whole rock.

[0068] Step 7: Incorporate the zircon-whole-rock coupling features, apatite-whole-rock coupling features, and consistency scores into the fusion basic feature vector to obtain the fusion feature vector of the target igneous rock. Then, input the fusion feature vector into the trained probability classification model for discrimination to obtain the posterior probability vectors of eight tectonic environments.

[0069] It should be noted that the eight types of tectonic environments mentioned in the embodiments of the present invention are continental arc, island arc, intraoceanic arc, back-arc basin, continental flood basalt, mid-ocean ridge, oceanic plateau, and ocean island.

[0070] Specifically, step 1 includes:

[0071] Whole-rock major and trace element analysis was performed on the target igneous rock sample to obtain whole-rock major and trace element data, and whole-rock element vectors were constructed based on the whole-rock major and trace element data.

[0072] In-situ trace element analysis of zircon and apatite in the target igneous rock sample was performed to obtain zircon position data and apatite position data.

[0073] Obtain metadata of the target igneous rock sample, including sample number, rock mass number, test batch, standard material, detection limits of each element, and analytical error.

[0074] Specifically, the process of analyzing the major and trace elements of the target igneous rock sample may include:

[0075] (1) Sample pretreatment: After removing the weathered surface and obvious altered parts, the sample is cleaned, dried, crushed and ground to obtain homogenized powder (e.g., less than 200 mesh), and the sample is divided by quartering or rotating sampler to form parallel samples for major and minor tests.

[0076] (2) Major element testing: Loss on ignition (LOI) was determined for powder samples, and the contents of major oxides such as SiO2, TiO2, Al2O3, Fe2O3 (or FeO / Fe2O3), MnO, MgO, CaO, Na2O, K2O, and P2O5 were determined by X-ray fluorescence (XRF) fused section method or acid digestion-inductively coupled plasma optical emission spectrometry (ICP-OES).

[0077] (3) Trace element and rare earth element test: The powder sample was subjected to high pressure closed acid digestion or alkali fusion-dissolution treatment using HF-HNO3 system, and trace elements and rare earth elements were determined by inductively coupled plasma mass spectrometry (ICP-MS).

[0078] Specifically, the process of in-situ trace element analysis of zircon and apatite in target igneous rock samples may include:

[0079] (1) Mineral sorting and sample preparation: The sample is crushed, sieved, separated by heavy liquid and magnetic separation to obtain zircon and apatite single mineral particles; under a stereomicroscope, particles with few cracks, few inclusions and good representativeness are selected, and epoxy resin targets or thin sheets are prepared and polished to expose the interior of the particles.

[0080] (2) Pretreatment characterization: The particles are subjected to cathodoluminescence (CL) or backscattering (BSE) imaging to identify zonal structures, cracks, inclusions and modified areas, and the point layout scheme is determined accordingly (at least 5 effective points are preferred for the same sample).

[0081] (3) In-situ trace element testing: Micro-area analysis techniques such as laser ablation-inductively coupled plasma mass spectrometry (LA-ICP-MS) or electron probe / secondary ion mass spectrometry are used to determine trace elements at selected sites; preferably, external standards (such as glass standard materials) and mineral standards (such as zircon / apatite reference materials) are used for joint calibration, and internal standard elements (such as Si or Zr for zircon, and Ca or P for apatite) are used for quantification;

[0082] (4) Data reduction and quality control: background subtraction, drift correction and isotope / element interference correction are performed on the signal to remove points that are obviously contaminated by inclusions / cracks or have abnormal signals; output the element concentration, detection limit, relative standard deviation / analysis error and other information of each measurement point, and establish a one-to-one correspondence with the sample number, particle number and point number to form a traceable point-level data table.

[0083] Specifically, the process of obtaining metadata for the target igneous rock sample is as follows:

[0084] Method blanks, parallel samples, and standard substances were inserted into the same batch to correct for instrument drift.

[0085] Record metadata such as test batch, instrument model, standard substance, limit of detection (LOD), measurement uncertainty / analytical error, etc., and unify the metadata to a consistent unit of measurement and field definition.

[0086] In this embodiment of the invention, the zircon and apatite point data preferably contain trace element data from at least 5 points; the analysis error is either relative or absolute.

[0087] Before step 2, this embodiment of the invention also includes unifying the fields and units of different data, establishing a missing policy code and a data lineage field, to ensure that subsequent processing can accurately trace the source and processing rules of each piece of data.

[0088] Specifically, a quality screening is performed on the whole-rock element vector to obtain the screened whole-rock element vector, including:

[0089] Find the analysis error corresponding to each element in the whole-rock element vector in the metadata;

[0090] An error threshold is set, and elements whose analysis error exceeds the error threshold are processed to obtain the whole rock element vector after initial screening.

[0091] Elements that exceed the preset range in the whole-rock element vector after initial screening are marked to obtain the marked elements;

[0092] Outlier removal is performed on the labeled elements in the whole-rock element vector after initial screening to obtain the screened whole-rock element vector.

[0093] In this embodiment of the invention, the error threshold can be selected within a preset range of 20% to 30%, for example, 25%. The preset range is set with upper and lower limits according to the reasonable range of the element in the corresponding medium, and elements exceeding 30% are marked.

[0094] Specifically, the zircon and apatite point data are filtered at the point level to obtain filtered zircon and apatite point data, including:

[0095] Centered on the median, the absolute median difference between the zircon and apatite point data is calculated as the first and second dispersions, respectively.

[0096] The deviation of any point in the zircon point data from key features or key elements is evaluated based on the first degree of dispersion. If the deviation is greater than the first degree of dispersion, the point is identified as an outlier and is removed or downweighted to obtain the filtered zircon point data.

[0097] The deviation of any point in the apatite point data from key features or key elements is evaluated based on the second degree of dispersion. If the deviation is greater than the second degree of dispersion, the point is identified as an outlier and is removed or downweighted to obtain the filtered apatite point data.

[0098] Specifically, using the median as the center, the absolute median difference between the zircon and apatite point data is calculated separately. The calculation process is as follows:

[0099] For a specific key element or key feature of the same sample, let... The observed values ​​at each point are First, calculate the median of the set of observations. Then calculate the absolute deviation of each point. And take the median of the absolute deviations. As the dispersion of this element / feature;

[0100] When it is necessary to align the dispersion with the standard deviation, you can let And calculate robustness deviation accordingly. ;

[0101] For multiple key elements / features, they can be calculated separately. The maximum value or weighted average is taken as the overall deviation of the points.

[0102] Since the detection limits may differ between different laboratories, instruments, and batches, the same element may frequently appear below the detection limit in different samples. If there are no instances of measurements below the detection limit, there are three common consequences: (1) Measurements below the detection limit will be forced to 0, resulting in deviations or even incomprehensibility in ratios, logarithms, and composition transformations; (2) Some elements will become undetectable in a large number of samples, causing inconsistencies in the available features of the samples and feature drift during model training and prediction; (3) Cross-data source incomparability will occur, and the model extrapolation will collapse. Therefore, step 3 of this embodiment specifically includes:

[0103] Based on the detection limits of each element in the metadata, the elements in the screened whole-rock element vector, the screened zircon point data, and the screened apatite point data are screened to determine the elements below the detection limit.

[0104] Elements below the detection limit are uniformly substituted to obtain whole-rock element vectors, zircon location data, and apatite location data after element substitution. For example, substitution can be performed using BDL=LOD×k, where k can be selected in the range of 0.3~0.7, for example, 0.5. LOD represents the lowest level that can be reliably distinguished. Since the actual sample measurement results are lower than LOD, they are usually between 0 and LOD. Therefore, using LOD×k gives a conservative positive value within this range, which is neither exaggerated nor does it avoid a value of 0.

[0105] In practical applications, even after substitution, some elements may still show 0 in certain samples, and logarithmic operations are required, which require positive inputs (x > 0). Therefore, to avoid situations where logarithms are absent, values ​​diverge, or transformations fail, this embodiment of the invention adds pseudo-counts to the zero or minimum values ​​in the whole-rock element vectors, zircon point data, and apatite point data after element substitution. This yields whole-rock elemental vectors, zircon point data, and apatite point data with added pseudo-counts. Available in 10 -6 ~10 -3 Select from a range, for example, select 10. -5 ;

[0106] Select elements from the whole-rock element vector, zircon point data and apatite point data with added pseudo-counts and form vectors to obtain the element composition vector.

[0107] The elemental composition vector is transformed to obtain the transformed whole-rock elemental vector, the transformed zircon point data, and the transformed apatite point data.

[0108] In this embodiment of the invention, central logarithmic transformation is performed on the whole-rock element vector, zircon point data and apatite point data with added pseudo-counts, respectively. In cases where it is necessary to maintain geometric properties or reduce dimensional correlation, equidistant logarithmic ratio transformation can be selected.

[0109] In this embodiment of the invention, the consistency score is used to quantify whether multiple pieces of evidence corroborate each other. It can be normalized to the 0-1 range, and its calculation can be based on multidimensional distance, correlation, or likelihood contribution.

[0110] It should be noted that the probabilistic classification model used in the embodiments of the present invention is not limited to a specific algorithm; the key is that its output is a probability distribution rather than a hard classification.

[0111] Specifically, the process of constructing principal component combination features and trace element ratio features and calculating rare earth curve morphology parameters for the transformed whole-rock element vector, as well as the feature calculation for the transformed zircon and apatite point data, may include:

[0112] (1) Construction of whole-rock features:

[0113] 1) Principal component combination characteristics: Calculation of Mg# based on principal component oxide content. The combined characteristics of K2O / Na2O, A / NK, A / CNK, TiO2 content levels, etc., among which Mg# can be expressed as a molar ratio of Mg / (Mg+Fe) 2+ )calculate;

[0114] 2) Trace element ratio characteristics: The ratio or logarithmic ratio characteristics that are sensitive to the structural environment and relatively robust in different media, such as Th / Nb, Th / Yb, Nb / Yb, Zr / Y, Ti / Y, La / Nb, Ba / Th, Sr / Y, La / Yb, Dy / Yb, Nb / U, etc.

[0115] 3) Rare Earth Curve Morphology Parameters: The rare earth elements are chondrite normalized to calculate the LREE / HREE slope and curvature indices (e.g., (La / Yb)N, (Gd / Yb)N), as well as anomaly parameters (e.g., Eu / Eu and Ce / Ce). Eu / Eu can be calculated as EuN / sqrt(SmN×GdN), and Ce / Ce can be calculated as CeN / sqrt(LaN×PrN). If necessary, parameters such as ΣREE, (La / Sm)N, and (Dy / Yb)N can be further calculated to characterize the overall shape of the curve.

[0116] (2) Zircon point feature calculation: Trace element calculation of zircon points is performed to determine the features related to magma oxygen fugacity, differentiation degree and source region properties, such as Ce anomaly, Eu anomaly, U / Yb, Th / U, Zr / Hf, Hf content level, (Dy / Yb)N, etc. When the point data contains Ti, the Ti-in-zircon temperature index can be further calculated as an optional feature.

[0117] (3) Calculation of apatite point characteristics: characteristics sensitive to the calculation of trace elements and volatile matter, differentiation stage and arc-related processes at apatite points, such as Sr, Y, ΣREE, (La / Yb)N, (Gd / Yb)N, Eu anomaly, Ce anomaly, Sr / Y, (Sm / Nd)N, etc.; when the test includes volatile matter indicators such as F and Cl, F / Cl, Cl content level, etc. can be further calculated as optional characteristics.

[0118] (4) Explanation of calculation method: If the logarithm or logarithmic ratio is involved in the calculation, the positive data after processing in step 3 shall be used; if it is necessary to take into account the closure constraints of the constituent data, the features can be supplemented and constructed in the clr / ilr transformation space, and together with the above interpretable features, they form a feature set.

[0119] Specifically, the process of calculating the coupling characteristics of zircon-whole rock and apatite-whole rock using the rare earth curve morphology parameters of whole-rock element vectors, zircon location characteristics, and apatite location characteristics may include:

[0120] (1) Feature alignment and standardization: Select a set of features that can be explained or mapped by both whole rock and mineral (e.g., Eu anomaly, Ce anomaly, LREE / HREE slope index, differentiation correlation ratio, etc.), and perform the same caliber transformation (e.g., logarithmic transformation, standardization or quantile normalization) on the corresponding features of whole rock and mineral side to eliminate dimensional differences;

[0121] (2) Zircon-whole-rock coupling characteristics: The differences and similarities between the zircon sample-level feature vector and the whole-rock feature vector are calculated in the alignment feature dimension to form coupling characteristics, such as differences in each dimension. Weighted Euclidean distance Cosine similarity wait;

[0122] (3) Apatite-whole-rock coupling characteristics: Similarly, the difference and similarity between the apatite sample-level feature vector and the whole-rock feature vector are calculated on the alignment feature dimension to obtain... , and Equivalent coupling characteristics;

[0123] (4) Summary of coupling features: The above-mentioned difference indicators are summarized. Combined with similarity indices (such as cos), zircon-whole-rock coupling feature vectors and apatite-whole-rock coupling feature vectors are formed to characterize the degree of consistency and direction of deviation between the mineral record and the whole-rock record on the same tectonic environment indication dimension.

[0124] Specifically, the calculation process for the consistency score based on the coupling characteristics of zircon-whole rock and apatite-whole rock may include:

[0125] (1) Distance Aggregation: The difference / distance indices in the zircon-whole-rock coupling characteristics are aggregated into a comprehensive deviation Dzr, and the difference / distance indices in the apatite-whole-rock coupling characteristics are aggregated into a comprehensive deviation Dap. The aggregation method can be weighted sum or weighted norm, for example... or , Similarly;

[0126] (2) Distance to score mapping: Map the deviation to a consistency score (CS) in the range of 0-1, for example, using an exponential mapping. ,in For scale parameters, These are weighting coefficients, which can be set based on the statistical distribution of the training set; or a normalized mapping can be used. Where D0 is the reference scale;

[0127] (3) Reliability and Conflict Determination: When CS is lower than the second preset threshold, the consistency of multiple pieces of evidence is deemed insufficient and "rejection / requirement of supplementary evidence" is triggered; when and When the deviation direction is significantly opposite or contradicts the key anomaly parameters of the whole rock side, a "conflict warning / mixed signal risk" flag is output, and the key dimensions that are triggered (such as Eu anomaly inconsistency, LREE / HREE slope inconsistency, etc.) are interpreted as output.

[0128] Specifically, after step 7, the following is also included:

[0129] The posterior probability vectors of eight constructed environments are calibrated using Pratt scaling or ordinal-preserving regression. Perform calibration.

[0130] To avoid forced classification despite insufficient evidence, this invention employs a rejection and conflict warning mechanism. When the maximum posterior probability falls below a first preset threshold or the consistency score falls below a second preset threshold, the system outputs "Rejection / Supplementary Evidence Required" and returns the Top-k candidate categories (k can be 3) and their corresponding probabilities. When whole-rock sub-evidence and mineral sub-evidence are inconsistent in the Top-1 category and the probability difference exceeds a set threshold, a "Conflict Warning / Risk of Mixed Signals" flag is output. To enhance usability in cases of missing data, this invention introduces a robust mechanism for missing data: when zircon or apatite is missing data, a mask or sub-model switching method is used to maintain output interface consistency, and a data sufficiency index DS (0-1 range) is output to indicate the reliability of the results and guide subsequent supplementary testing.

[0131] Compared with existing technologies, this invention performs quality screening on whole-rock, zircon, and apatite data based on the metadata of the target igneous rock sample. This reduces the impact of differences in the quality of the original data on the discrimination results from the source of the data. The detection limits of the screened whole-rock element vector, the screened zircon point data, and the screened apatite point data are made consistent through metadata to improve comparability across batches and laboratories and suppress spurious correlations. Features are calculated and aggregated for the transformed whole-rock element vector, the transformed zircon point data, and the transformed apatite point data to solve the problems of non-standard fusion of multi-point data and the inability to quantify cross-evidence consistency. This improves the discrimination stability of tectonic environments from multiple aspects.

[0132] Regarding the tectonic environment probability discrimination method based on multi-source geochemical data described in the above embodiments, such as... Figure 2As shown, this embodiment of the invention also provides a tectonic environment probability discrimination device 100 based on multi-source geochemical data, the tectonic environment probability discrimination device 100 comprising:

[0133] Analysis module 101 is used to perform elemental analysis on the target igneous rock sample to obtain whole-rock element vectors, zircon position data, apatite position data and metadata;

[0134] The screening module 102 is used to perform quality screening on the whole-rock element vector based on metadata to obtain the screened whole-rock element vector, and to perform point-level screening on zircon point data and apatite point data to obtain the screened zircon point data and screened apatite point data.

[0135] The transformation module 103 is used to perform detection limit consistency and data transformation on the screened whole-rock element vector, the screened zircon point data and the screened apatite point data based on metadata, so as to obtain the transformed whole-rock element vector, the transformed zircon point data and the transformed apatite point data.

[0136] The calculation module 104 is used to construct the principal combination features and trace element ratio features of the transformed whole rock element vector and calculate the rare earth curve morphology parameters to obtain the whole rock features. It also calculates the features of the transformed zircon point data and the transformed apatite point data to obtain the zircon point features and apatite point features.

[0137] The splicing module 105 is used to aggregate the zircon location features and apatite location features using the central statistic and dispersion index respectively, to obtain the zircon sample-level feature vector and the apatite sample-level feature vector, and splice the whole rock features with the zircon sample-level feature vector and the apatite sample-level feature vector to obtain the fused basic feature vector of the target igneous rock sample.

[0138] The coupling module 106 is used to calculate the coupling characteristics of zircon-whole rock and apatite-whole rock using the rare earth curve morphology parameters of the whole rock element vector, zircon location characteristics and apatite location characteristics, and to calculate the consistency score based on the coupling characteristics of zircon-whole rock and apatite-whole rock.

[0139] The discrimination module 107 is used to incorporate the coupling features of zircon-whole rock, the coupling features of apatite-whole rock, and the consistency score into the fusion basic feature vector to obtain the fusion feature vector of the target igneous rock. The fusion feature vector is then input into the trained probability classification model for discrimination to obtain the posterior probability vectors of eight types of tectonic environments.

[0140] Specifically, the construction environment probability discrimination device 100 also includes:

[0141] The calibration module 108 is used to calibrate the posterior probability vectors of eight types of constructed environments using Pratt scaling calibration or ordinal regression calibration.

[0142] It should be noted that the information interaction and execution process between the above-mentioned devices / units are based on the same concept as the method embodiments of this application. For details on their specific functions and technical effects, please refer to the method embodiments section, and they will not be repeated here.

[0143] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0144] This invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements a structural environment probability discrimination method based on multi-source geochemical data.

[0145] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of this application can be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include at least: any entity or device capable of carrying the computer program code to a building device / terminal device, a recording medium, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium. Examples include USB flash drives, portable hard drives, magnetic disks, or optical disks.

[0146] This invention also provides a terminal device, such as... Figure 3 As shown, the terminal device D10 of this embodiment includes: at least one processor D100 ( Figure 3 The diagram shows only one processor, a memory D101, and a computer program D102 stored in the memory D101 and executable on the at least one processor D100. When the processor D100 executes the computer program D102, it implements the above-described method for determining the probabilistic structural environment based on multi-source geochemical data.

[0147] The terminal device D10 can be a desktop computer, laptop, handheld computer, server, server cluster, or cloud server, etc. This terminal device may include, but is not limited to, a processor D100 and a memory D101. Those skilled in the art will understand that... Figure 3 This is merely an example of terminal device D10 and does not constitute a limitation on terminal device D10. It may include more or fewer components than shown in the figure, or combine certain components, or different components, such as input / output devices, network access devices, etc.

[0148] The processor D100 can be a central processing unit (CPU), or it can be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor.

[0149] In some embodiments, the memory D101 may be an internal storage unit of the terminal device D10, such as a hard disk or memory of the terminal device D10. In other embodiments, the memory D101 may be an external storage device of the terminal device D10, such as a plug-in hard disk, smart media card (SMC), secure digital card (SD), flash card, etc., equipped on the terminal device D10. Furthermore, the memory D101 may include both internal and external storage units of the terminal device D10. The memory D101 is used to store the operating system, applications, bootloader, data, and other programs, such as the program code of the computer program. The memory D101 can also be used to temporarily store data that has been output or will be output.

[0150] It should be noted that the information interaction and execution process between the above-mentioned devices / units are based on the same concept as the method embodiments of this application. For details on their specific functions and technical effects, please refer to the method embodiments section, and they will not be repeated here.

[0151] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0152] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for probabilistic determination of tectonic environments based on multi-source geochemical data, characterized in that, include: Step 1: Perform elemental analysis on the target igneous rock sample to obtain whole-rock elemental vectors, zircon location data, apatite location data, and metadata; Step 2: Perform quality screening on the whole-rock element vector based on the metadata to obtain the screened whole-rock element vector; perform point-level screening on the zircon point data and the apatite point data to obtain the screened zircon point data and the screened apatite point data. Step 3: Based on the metadata, perform detection limit consistency and data transformation on the screened whole-rock element vector, the screened zircon point data and the screened apatite point data to obtain the transformed whole-rock element vector, the transformed zircon point data and the transformed apatite point data. Step 4: Construct principal combination features and trace element ratio features and calculate rare earth curve morphology parameters for the transformed whole-rock element vector to obtain whole-rock features. Then, calculate the features of the transformed zircon point data and the transformed apatite point data to obtain zircon point features and apatite point features. Step 5: Aggregate the zircon location features and the apatite location features using the central statistic and dispersion index respectively to obtain zircon sample-level feature vectors and apatite sample-level feature vectors. Then, concatenate the whole-rock features with the zircon sample-level feature vectors and the apatite sample-level feature vectors to obtain the fused basic feature vector of the target igneous rock sample. Step 6: Calculate the coupling characteristics of zircon-whole rock and apatite-whole rock using the rare earth curve morphology parameters of the whole rock element vector, the zircon location characteristics, and the apatite location characteristics, and calculate the consistency score based on the zircon-whole rock and apatite-whole rock coupling characteristics. Step 7: Incorporate the zircon-whole-rock coupling features, apatite-whole-rock coupling features, and consistency score into the fusion basic feature vector to obtain the fusion feature vector of the target igneous rock. Then, input the fusion feature vector into the trained probability classification model for discrimination to obtain the posterior probability vectors of eight tectonic environments.

2. The method for probabilistic determination of tectonic environments based on multi-source geochemical data according to claim 1, characterized in that, Step 1 includes: Whole-rock major and trace element analysis was performed on the target igneous rock sample to obtain whole-rock major and trace element data, and a whole-rock element vector was constructed based on the whole-rock major and trace element data. In-situ trace element analysis of zircon and apatite in the target igneous rock sample was performed to obtain zircon position data and apatite position data. Obtain metadata of the target igneous rock sample, including sample number, rock mass number, test batch, standard material, detection limit of each element, and analytical error.

3. The method for probabilistic determination of tectonic environments based on multi-source geochemical data according to claim 1, characterized in that, The whole-rock element vector is subjected to quality screening to obtain the screened whole-rock element vector, including: Find the analysis error corresponding to each element in the whole-rock element vector in the metadata; An error threshold is set, and elements whose analysis error is greater than the error threshold are processed to obtain the whole rock element vector after initial screening; Elements that exceed a preset range in the initial screened whole-rock element vector are marked to obtain the marked elements; Outlier removal is performed on the labeled elements in the whole-rock element vector after initial screening to obtain the screened whole-rock element vector.

4. The method for probabilistic determination of tectonic environments based on multi-source geochemical data according to claim 1, characterized in that, The zircon and apatite point data are filtered at the point level to obtain filtered zircon and apatite point data, including: Centered on the median, the absolute median difference between the zircon point data and the apatite point data is calculated as the first dispersion and the second dispersion, respectively. The deviation of any point in the zircon point data from key features or key elements is evaluated based on the first dispersion. If the deviation is greater than the first dispersion, the point is determined to be an outlier, and the outlier is removed or reduced in weight to obtain the filtered zircon point data. The deviation of any point in the apatite point data from key features or key elements is evaluated based on the second dispersion. If the deviation is greater than the second dispersion, the point is identified as an outlier and is removed or downweighted to obtain the filtered apatite point data.

5. The method for probabilistic determination of tectonic environments based on multi-source geochemical data according to claim 1, characterized in that, Step 3 includes: Based on the detection limits of each element in the metadata, the elements in the screened whole-rock element vector, the screened zircon point data, and the screened apatite point data are screened to determine the elements below the detection limit. Elements below the detection limit were uniformly substituted to obtain whole-rock element vectors, zircon location data, and apatite location data after element substitution. Pseudo-counts are added to the zero or minimum values ​​in the whole-rock element vector, zircon point data and apatite point data after element substitution to obtain the whole-rock element vector, zircon point data and apatite point data with pseudo-counts added. Select elements from the whole-rock element vector, zircon point data and apatite point data with added pseudo-counts and form vectors to obtain the element composition vector. The elemental composition vector is transformed to obtain the transformed whole-rock elemental vector, the transformed zircon point data, and the transformed apatite point data.

6. The method for probabilistic determination of tectonic environments based on multi-source geochemical data according to claim 1, characterized in that, Step 7 is followed by: Pratt scaling calibration or ordinal-preserving regression calibration is used to calibrate the posterior probability vectors of eight types of constructed environments.

7. A tectonic environment probability discrimination device based on multi-source geochemical data, characterized in that, include: The analysis module is used to perform elemental analysis on the target igneous rock sample to obtain whole-rock elemental vectors, zircon position data, apatite position data, and metadata. The screening module is used to perform quality screening on the whole-rock element vector based on the metadata to obtain the screened whole-rock element vector, and to perform point-level screening on the zircon point data and the apatite point data to obtain the screened zircon point data and the screened apatite point data. The transformation module is used to perform detection limit consistency and data transformation on the screened whole-rock element vector, the screened zircon point data and the screened apatite point data based on the metadata, so as to obtain the transformed whole-rock element vector, the transformed zircon point data and the transformed apatite point data. The calculation module is used to construct the principal combination features and trace element ratio features of the transformed whole-rock element vector and calculate the rare earth curve morphology parameters to obtain whole-rock features. It also calculates the features of the transformed zircon point data and the transformed apatite point data to obtain zircon point features and apatite point features. The splicing module is used to aggregate the zircon location features and the apatite location features using the central statistic and dispersion index respectively, to obtain zircon sample-level feature vectors and apatite sample-level feature vectors, and splices the whole-rock features with the zircon sample-level feature vectors and apatite sample-level feature vectors to obtain the fused basic feature vector of the target igneous rock sample. The coupling module is used to calculate the coupling characteristics of zircon-whole rock and apatite-whole rock using the rare earth curve morphology parameters of the whole rock element vector, the zircon point features and the apatite point features, and to calculate the consistency score based on the zircon-whole rock coupling characteristics and apatite-whole rock coupling characteristics. The discrimination module is used to incorporate the coupling features of zircon-whole rock, the coupling features of apatite-whole rock, and the consistency score into the fusion basic feature vector to obtain the fusion feature vector of the target igneous rock, and input the fusion feature vector into the trained probability classification model for discrimination to obtain the posterior probability vectors of eight types of tectonic environments.

8. The tectonic environment probability discrimination device based on multi-source geochemical data according to claim 7, characterized in that, Also includes: The calibration module is used to calibrate the posterior probability vectors of eight constructed environments using Pratt scaling calibration or ordinal regression calibration.

9. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the structural environment probability discrimination method based on multi-source geochemical data as described in any one of claims 1 to 6.

10. A terminal device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the structural environment probability discrimination method based on multi-source geochemical data as described in any one of claims 1 to 6.

Images