A method for screening mineralization feature sets of sandstone-type uranium deposits based on maximum information coefficient

Abstract

This invention belongs to the interdisciplinary field of uranium geological exploration information technology and big data processing technology. Specifically, it relates to a method for screening mineralization feature sets of sandstone-type uranium deposits based on the maximum information coefficient (MIC). The method includes: Step S1: Extraction and organization of mineralization feature data; Step S2: Data preprocessing; Step S3: MIC calculation and preliminary ranking of feature values; Step S4: Screening feature thresholds based on inflection point analysis to determine key mineralization features. This method can identify key ore-controlling factors with high correlation to sandstone-type uranium mineralization from geoscientific big data, significantly improving the accuracy, efficiency, and interpretability of subsequent machine learning mineralization prediction models, and providing reliable data support for the selection of uranium exploration target areas.

Description

Technical Field

[0001] This invention belongs to the interdisciplinary field of uranium geological exploration information technology and big data processing technology, and specifically relates to a method for screening the mineralization feature set of sandstone-type uranium deposits based on the maximum information coefficient. Background Technology

[0002] The mineralization process of sandstone-type uranium deposits is controlled by multiple geological factors, such as tectonics, lithology, geophysical fields, and geochemical fields. These factors are coupled with complex nonlinear relationships. With the development of "digital geology," geologists can acquire and participate in modeling with a dramatic increase in the number of data dimensions (feature count), including a large amount of geological, geophysical, geochemical, and remote sensing data.

[0003] Faced with such high-dimensional, multi-source, and nonlinear big data, some feature selection methods (such as the Pearson correlation coefficient method) are mainly suitable for linear relationship selection and are not sensitive to nonlinear relationships. However, there are many complex nonlinear relationships between the ore-controlling factors and mineralization results of sandstone-type uranium deposits. Directly using these feature data for prediction will lead to high prediction model complexity, large computational workload, long training time, and may introduce noise and irrelevant features, reducing the model's generalization ability and prediction accuracy, resulting in the "curse of dimensionality" problem.

[0004] Therefore, there is an urgent need in this field for a screening method that can effectively capture nonlinear relationships and accurately identify the set of features most relevant to sandstone-type uranium mineralization from high-dimensional geoscience data, so as to provide high-quality input for subsequent high-precision mineralization prediction models. Summary of the Invention

[0005] The purpose of this invention is to provide a method for screening mineralization feature sets of sandstone-type uranium deposits based on the maximum information coefficient. This method can identify key ore-controlling factors with high correlation to the mineralization of sandstone-type uranium deposits from geoscientific big data, significantly improving the accuracy, efficiency and interpretability of subsequent machine learning mineralization prediction models, and providing reliable data support for the selection of uranium exploration target areas.

[0006] Technical solution to achieve the purpose of this invention: A method for screening mineralization feature sets of sandstone-type uranium deposits based on the maximum information coefficient includes: Step S1: Extraction and organization of mineralization characteristic data; Step S2: Data preprocessing; Step S3: MIC calculation and preliminary eigenvalue sorting; Step S4: Based on inflection point analysis, filter feature thresholds to determine key mineralization features.

[0007] Furthermore, step S1 includes: extracting mineralization characteristic data of the target area from the uranium ore database, performing preliminary data processing, and obtaining a dataset.

[0008] Furthermore, in step S1, the mineralization characteristic data includes: geophysical characteristics, geochemical characteristics, geological characteristics, formation characteristics, topographic characteristics, remote sensing characteristics, and hydrogeological characteristics.

[0009] Furthermore, step S2 includes: cleaning the extracted and organized dataset; after cleaning, normalizing or standardizing the continuous features and encoding the discrete features.

[0010] Furthermore, in step S2, data cleaning includes processing missing values ​​and outliers.

[0011] Furthermore, in step S2, numerical features are subjected to Z-score standardization, and textual features are subjected to one-hot encoding.

[0012] Further, step S3 includes: using the MIC algorithm to calculate the maximum information coefficient value (MIC) between each feature and the label; and sorting all features from largest to smallest according to the MIC values.

[0013] Furthermore, in step S3, the minepy library of Python is used to calculate the MIC value of each feature and label.

[0014] Further, step S4 includes: drawing a MIC value sorting line graph based on the feature sorting results, identifying inflection points in the line graph, and selecting all features before the inflection point as the mineralization key feature set.

[0015] The beneficial technical effects of this invention are as follows: 1. This invention provides a method for screening mineralization feature sets of sandstone-type uranium deposits based on the maximum information coefficient (MIC). Employing the MIC effectively captures the complex nonlinear and non-functional dependencies widely present in sandstone-type uranium deposit mineralization systems, resulting in screening results that better conform to mineralization regularities. This method not only avoids the subjectivity and experience-based dependence of traditional manual selection but also compensates for the shortcomings of other data processing methods in handling nonlinear data, achieving rapid and efficient processing of high-dimensional data.

[0016] 2. The present invention provides a method for screening mineralization feature sets of sandstone-type uranium deposits based on the maximum information coefficient (MIC). The method uses the magnitude of the MIC value for quantitative analysis, which reflects the correlation strength between the features and mineralization. This method facilitates geologists' understanding and verification of the screening results, and helps to discover potential, overlooked mineralization-controlling factors, thus having clear geological interpretation significance.

[0017] 3. The present invention provides a method for screening the mineralization feature set of sandstone-type uranium deposit data based on the maximum information coefficient, which is applicable to the processing of large data of sandstone-type uranium deposits that are multi-sourced, complex and have a large number of nonlinear relationships. Attached Figure Description

[0018] Figure 1 The flowchart of a method for screening mineralization feature sets of sandstone-type uranium deposits based on the maximum information coefficient provided by the present invention is shown below. Figure 2 This is a line graph showing the sorting of MIC values ​​of mineralization features in an embodiment of the present invention. Detailed Implementation

[0019] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments.

[0020] like Figure 1 As shown, this invention provides a method for screening the metallogenic feature set of sandstone-type uranium deposit data based on the maximum information coefficient, specifically including the following steps: Step S1: Extraction and processing of mineralization characteristic data Mineralization characteristic data of the target area are extracted from the uranium database, and the data is preliminarily processed according to the needs of subsequent work to obtain a dataset.

[0021] Mineralization characteristics data include: geophysical characteristics (such as gravity anomalies, aeromagnetic anomalies, and aeromagnetic U, Tu, and K anomalies), geochemical characteristics (such as the content and ratio of elements such as uranium, thorium, and potassium, and molybdenum content), geological characteristics (granite bodies, fault uplift structures, and thickness of single-layer sand bodies), formation characteristics (fan deltaic lobe sand bodies, mud-sand-mudstone lithology, and sedimentary facies), topographic characteristics (such as elevation and slope), remote sensing characteristics (such as replenishment-drainage mineralization systems and alteration anomaly intensity), and hydrogeological characteristics (such as uranium anomalies and radon anomalies in water).

[0022] Step S2: Data Preprocessing The extracted and organized dataset is cleaned, including handling missing and outlier values. After cleaning, continuous features are normalized or standardized to eliminate the influence of units. Discrete features are encoded to convert non-numerical categorical information into numerical vectors.

[0023] In a specific implementation, numerical features are Z-score standardized, and textual features are one-hot encoded.

[0024] Step S3: MIC calculation and preliminary sorting of eigenvalues.

[0025] The MIC algorithm is used to calculate the maximum information coefficient (MIC) between each feature and the label; all features are then sorted from largest to smallest based on their MIC values.

[0026] In a specific implementation, the minepy library in Python is used to calculate the MIC value of each feature and label.

[0027] Step S4: Based on inflection point analysis, filter feature thresholds to determine key mineralization features. Based on the feature ranking results, a line graph of MIC value ranking is drawn, the "inflection point" in the line graph is identified, and all features before the inflection point are selected as the key feature set for mineralization.

[0028] Example 1 Taking the sandstone-type uranium deposit in the Chagandelesu Depression of the Yin'e Basin as an example, this embodiment provides a method for screening the metallogenic feature set of sandstone-type uranium deposit data based on the maximum information coefficient, specifically including the following steps: Step S1: Extract multi-source data from the uranium mine big data platform. Data types and contents include, but are not limited to: geophysical characteristics (e.g., gravity anomalies, aeromagnetic anomalies, aeromagnetic U, Tu, K anomalies), geochemical characteristics (e.g., uranium, thorium, potassium and other elemental contents and ratios, molybdenum content), geological characteristics (granite bodies, fault uplift structures, single-layer sand body thickness), formation characteristics (fan deltaic lobe sand bodies, mud-sand-mudstone lithology, sedimentary facies), topographic characteristics (e.g., elevation, slope), remote sensing characteristics (e.g., replenishment-drainage-discharge mineralization system, alteration anomaly intensity), and hydrogeological characteristics (e.g., uranium anomalies in water, radon anomalies in water), etc.

[0029] Step S2: Clean the extracted and organized dataset. After cleaning, perform Z-score standardization on numerical features and one-hot encoding on text features.

[0030] Step S3: Calculate the MIC value of each feature and label using the minepy library in Python.

[0031] Step S4: Draw a line chart sorted by MIC values, as shown below. Figure 2 As shown, inflection points were identified based on the line graph (or a MIC threshold was set to be no less than a certain value, and then features were filtered). A clear inflection point was found at the 11th ranked feature. Therefore, the top 11 features were selected as the key mineralization feature set. These 11 features include: uranium content in sand bodies (strata), uranium element anomalies in water, redox alteration zones, fault-uplift tectonic combinations, upper Bayangobi Formation strata, fan deltaic lobe sand bodies, U anomalies from aerial spraying, groundwater recharge-drainage-discharge mineralization conditions, radon anomalies in water, U / Th ratio, and molybdenum-tin-rhenium element combination anomalies, which are highly consistent with geological understanding.

[0032] The present invention has been described in detail above with reference to the accompanying drawings and embodiments. However, the present invention is not limited to the above embodiments, and various changes can be made within the scope of knowledge possessed by those skilled in the art without departing from the spirit of the present invention. All contents not described in detail in the present invention can be derived from existing technologies.

Claims

1. A method for screening mineralization feature sets of sandstone-type uranium deposits based on the maximum information coefficient, characterized in that, include: Step S1: Extraction and organization of mineralization characteristic data; Step S2: Data preprocessing; Step S3: MIC calculation and preliminary eigenvalue sorting; Step S4: Based on inflection point analysis, filter feature thresholds to determine key mineralization features.

2. The method for screening the metallogenic feature set of sandstone-type uranium deposits based on the maximum information coefficient according to claim 1, characterized in that, Step S1 includes: extracting mineralization characteristic data of the target area from the uranium ore database, performing preliminary data processing, and obtaining a dataset.

3. The method for screening the metallogenic feature set of sandstone-type uranium deposits based on the maximum information coefficient according to claim 2, characterized in that, In step S1, the mineralization characteristic data includes: geophysical characteristics, geochemical characteristics, geological characteristics, structural characteristics, topographic characteristics, remote sensing characteristics, and hydrogeological characteristics.

4. The method for screening the metallogenic feature set of sandstone-type uranium deposits based on the maximum information coefficient according to claim 2, characterized in that, Step S2 includes: cleaning the extracted and organized dataset; after cleaning, normalizing or standardizing continuous features and encoding discrete features.

5. The method for screening the metallogenic feature set of sandstone-type uranium deposits based on the maximum information coefficient according to claim 4, characterized in that, In step S2, data cleaning includes processing missing values ​​and outliers.

6. The method for screening the metallogenic feature set of sandstone-type uranium deposits based on the maximum information coefficient according to claim 4, characterized in that, In step S2, numerical features are subjected to Z-score standardization, and textual features are subjected to one-hot encoding.

7. The method for screening the metallogenic feature set of sandstone-type uranium deposits based on the maximum information coefficient according to claim 4, characterized in that, Step S3 includes: using the MIC algorithm to calculate the maximum information coefficient value (MIC) between each feature and the label; and sorting all features from largest to smallest according to the MIC values.

8. The method for screening the metallogenic feature set of sandstone-type uranium deposits based on the maximum information coefficient according to claim 7, characterized in that, In step S3, the minepy library of Python is used to calculate the MIC value of each feature and label.

9. The method for screening the metallogenic feature set of sandstone-type uranium deposits based on the maximum information coefficient according to claim 7, characterized in that, Step S4 includes: drawing a MIC value sorting line chart based on the feature sorting results, identifying inflection points in the line chart, and selecting all features before the inflection point as the mineralization key feature set.

Images