A method for extracting data of geological mineral exploration

Abstract

The application relates to the technical field of mineral exploration, and discloses a geological mineral exploration data extraction method, which comprises the following steps: step one, overall data collection of a target region is carried out by a drone carrying a hyperspectral sensor, wherein the data comprises spectral information of multiple wave bands; step two, a ground sensor is arranged at a designated sampling point to collect aeromagnetic data and geological samples, and the working frequency of the ground sensor is set to 1 Hz; step three, wavelet transform technology is adopted to remove noise from the collected hyperspectral and aeromagnetic data, and signals lower than a set threshold are removed; and step four, characteristics are extracted by principal component analysis to generate a comprehensive characteristic matrix. By adopting the comprehensive data collection method of the unmanned aerial vehicle hyperspectral sensor and the ground sensor, the speed and coverage range of data collection are greatly improved, and compared with traditional manual sampling, the exploration of a large-area region can be completed in a shorter time.

Description

Technical Field

[0001] This invention relates to the field of mineral exploration technology, specifically a method for extracting data in geological and mineral exploration. Background Technology

[0002] In today's context of increasingly scarce global resources, the efficient exploration and development of mineral resources is of paramount importance. Traditional mineral exploration methods often rely on manual sampling and simple laboratory analysis. While these methods can yield some information, their inefficiency and lack of accuracy become increasingly prominent when dealing with large-scale mining areas. With technological advancements, adopting emerging data acquisition and analysis technologies has become a crucial way to improve exploration efficiency and accuracy.

[0003] Existing technologies mainly include ground sampling, geological surveys, and simple data analysis methods. These methods typically require lengthy field investigations and substantial human resources. Data processing and analysis often rely on manual judgment, making them susceptible to human influence. Furthermore, existing technologies lack effective integration of multiple data sources, failing to comprehensively reflect the actual conditions of the mining area. This results in a low rate of mineralization anomaly identification, impacting the scientific rigor and accuracy of decision-making.

[0004] The main problem with existing technologies is the lack of an efficient and accurate comprehensive data acquisition and analysis method, which makes it impossible to identify ore body characteristics and mineralization anomalies in real time and comprehensively. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a method for extracting data in geological and mineral exploration, solving the problem of the lack of efficient and accurate comprehensive data acquisition and analysis methods in existing technologies.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for extracting data in geological and mineral exploration, comprising the following steps:

[0007] Step 1: Collect comprehensive data on the target area using a drone equipped with a hyperspectral sensor. The data includes spectral information in multiple bands.

[0008] Step 2: Deploy ground sensors at designated sampling points to collect aeromagnetic data and geological samples. The operating frequency of the ground sensors is set to 1 Hz.

[0009] Step 3: Use wavelet transform technology to remove noise from the collected hyperspectral and aeromagnetic data, removing signals below a set threshold;

[0010] Step 4: Extract features through principal component analysis to generate a comprehensive feature matrix, and integrate different data sources through multidimensional data fusion methods to enhance data reliability;

[0011] Step 5: Analyze the comprehensive feature matrix using machine learning and deep learning algorithms to automatically identify mineralization anomalies and ore body characteristics, and generate an exploration report.

[0012] Preferably, the drone is set to fly at an altitude of 100 meters and a speed of 5 meters per second to ensure that the coverage rate during data collection is not less than 80%.

[0013] Preferably, the ground sensor includes a high-precision magnetometer and a laser scanner. The magnetometer can record changes in magnetic field strength in real time, and the laser scanner has a resolution of not less than 1 cm to acquire high-resolution terrain point cloud data.

[0014] Preferably, the sample collection step includes at least selecting 10 representative sampling points within the target area, extracting at least 500 grams of soil or rock samples from each sample using standard geological sampling tools, and recording the geographical location and characteristics of the samples on-site.

[0015] Preferably, in the noise elimination step, wavelet transform is used to perform multi-level signal decomposition, retaining high-frequency components related to geological features and removing noise below a set threshold.

[0016] Preferably, in the feature extraction step, principal component analysis is used to construct a covariance matrix to identify principal components. This step includes calculating eigenvalues ​​and eigenvectors, and selecting principal components with a cumulative contribution rate of 95% for analysis.

[0017] Preferably, the data fusion step employs a weighted fusion technique, combining hyperspectral data and aeromagnetic data, as shown in the formula:

[0018] F combined =αF yperspectral +βF magnetic

[0019] Here, α and β are weighting coefficients set according to the importance of each data source.

[0020] Preferably, the machine learning algorithm includes random forest and convolutional neural network for classifying and identifying mineralization anomalies, wherein the training dataset includes at least 1,000 samples.

[0021] Preferably, the analysis results are presented through visualization tools to generate easy-to-understand geological maps and exploration reports, assisting exploration personnel in making decisions.

[0022] Preferably, the method further includes real-time monitoring and feedback of the exploration data, and using a cloud computing platform to store and process the data for access and analysis.

[0023] This invention provides a method for extracting data in geological and mineral exploration. It has the following beneficial effects:

[0024] 1. This invention significantly improves the speed and coverage of data acquisition by using a comprehensive data acquisition method that combines UAV hyperspectral sensors and ground sensors. Compared with traditional manual sampling, it can complete the survey of large areas in a shorter time.

[0025] 2. This invention improves the accuracy of identifying mineralization anomalies and ore body characteristics by integrating multiple data sources. Comprehensive analysis of hyperspectral data and ground sensor data makes the analysis of geological features more comprehensive and reduces errors caused by a single data source.

[0026] 3. By applying machine learning and deep learning algorithms, this invention can automatically process and analyze large amounts of complex data, identify potential mineralization areas and features, significantly improve identification accuracy and efficiency, and reduce reliance on manual judgment.

[0027] 4. The data acquisition and analysis process of this invention achieves real-time processing, enabling the immediate generation of exploration reports, which facilitates decision-making and subsequent mineral development. Simultaneously, by meticulously recording the environmental conditions and equipment status at sampling points, the traceability and reliability of the data are ensured.

[0028] 5. This invention is applicable to various geological environments and mineral types, and can be widely used in the exploration of different types of mineral resources, thus improving the applicability and flexibility of the technology. Attached Figure Description

[0029] Figure 1 This is a flowchart of the present invention. Detailed Implementation

[0030] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0031] Example 1:

[0032] Please see the appendix Figure 1 This invention provides a method for extracting data in geological and mineral exploration. This embodiment is carried out in a metal mining area where known mineralization phenomena exist, with the goal of improving the understanding of the distribution and characteristics of ore bodies.

[0033] Specifically, the steps include:

[0034] Step 1: Data Collection

[0035] 1.1 Application of UAVs Equipped with Hyperspectral Sensors

[0036] Equipment selection and preparation:

[0037] In this embodiment, a drone with good stability, such as the DJI Matrice 600, is selected to ensure that it can carry a hyperspectral sensor, such as the Headwall Photonics Nano-Hyperspec.

[0038] Preparations include a thorough check of the drone's battery status to ensure it is fully charged, and equipment calibration to ensure sensor accuracy.

[0039] Flight path design:

[0040] Using flight planning software, such as Mission Planner, design the flight path, set the flight altitude to 100 meters, and the flight speed to 5 meters per second, ensuring that the overlap of the target area reaches more than 80%.

[0041] Pay special attention to avoiding buildings and high-voltage power lines to ensure flight safety and reduce the risk of collision.

[0042] Data acquisition process:

[0043] During actual flight, the drone is activated, its flight status is monitored in real time, and image data is received. As the drone flies over the target area, the sensors capture the reflection spectrum of each band, recording high-resolution spectral images.

[0044] Data is stored in the drone's built-in memory and backed up to the cloud regularly to prevent data loss.

[0045] 1.2 Deployment and Data Collection of Ground Sensors

[0046] Selection of key sampling points:

[0047] Before the flight, select at least 10 key sampling points (such as known mineralized areas or anomalous regions) and accurately locate them using GPS equipment. Ensure that these points are reasonably distributed and representative of the geological features of the entire study area.

[0048] Sensor deployment:

[0049] At each sampling point, a high-precision geomagnetic instrument, such as a GSM-19, is deployed and set to a recording frequency of 1 Hz for real-time monitoring to obtain data on changes in magnetic field strength.

[0050] At the same time, a laser scanner, such as the Leica BLK360, is used to acquire ground 3D point cloud data with a resolution of 1 cm to ensure high data accuracy.

[0051] Data recording and environmental monitoring:

[0052] During data acquisition, record the status of each sensor (e.g., battery level, connection status) and environmental conditions (e.g., temperature, humidity) to ensure data traceability and reliability. Regularly check the sensor's operating status to ensure continuous data acquisition.

[0053] Step 2: Data Preprocessing

[0054] 2.1 Noise Reduction and Data Cleaning

[0055] Applications of wavelet transform:

[0056] Wavelet transform is applied to the acquired hyperspectral data for noise reduction. A suitable mother wavelet function is selected to decompose the signal and remove noise components below a set threshold.

[0057] Ensure that the processed data retains its key features for subsequent analysis.

[0058] 2.2 Data Normalization

[0059] Normalization process:

[0060] Data from different sources, including hyperspectral and ground sensor data, are normalized to eliminate dimensional differences. The specific method is as follows:

[0061]

[0062] Normalized data facilitates subsequent feature extraction and model training.

[0063] Step 3: Feature Extraction and Analysis

[0064] 3.1 Feature Extraction

[0065] Principal Component Analysis (PCA):

[0066] Purpose and Principle:

[0067] In this embodiment, Principal Component Analysis (PCA) is used to extract features from the processed hyperspectral and ground sensor data to reduce the dimensionality of the data while retaining the most important features. The basic principle of PCA is to project high-dimensional data into a new coordinate system such that the dimensions (principal components) of the new coordinate system are as independent as possible, and the variance of each principal component is maximized.

[0068] Specific implementation steps:

[0069] Data standardization: The collected data is standardized to ensure that the mean of each feature is 0 and the standard deviation is 1, so as to eliminate the influence of different units on the analysis results.

[0070] Covariance matrix calculation: The covariance matrix of the standardized data is calculated using the following formula:

[0071]

[0072] Where X is the standardized data matrix and n is the number of samples.

[0073] Eigenvalue and eigenvector decomposition: The covariance matrix is ​​decomposed into eigenvalues ​​to extract eigenvalues ​​and their corresponding eigenvectors. Larger eigenvalues ​​indicate a greater proportion of the variance in the data, and therefore greater importance.

[0074] Principal component selection: Select the top k principal components based on the eigenvalues. Typically, select principal components that explain 85%-95% of the cumulative variance to ensure that sufficient information is retained.

[0075] Feature matrix construction: The selected feature vectors are combined into a feature matrix, and the original data is projected onto this feature space to form a new comprehensive feature matrix.

[0076] 3.2 Data Fusion

[0077] Multidimensional data fusion:

[0078] Purpose and Principle:

[0079] In this embodiment, multidimensional data fusion technology is employed to integrate hyperspectral data with ground sensor data to improve the reliability and effectiveness of the data. Multidimensional data fusion can extract complementary information from different data sources, enhancing the depth and breadth of data analysis.

[0080] Specific implementation steps:

[0081] Data preprocessing: Ensure all data formats are consistent and perform necessary calibration and synchronization, such as aligning timestamps to ensure time consistency across different data sources.

[0082] Choosing a fusion algorithm: A suitable fusion algorithm should be selected based on the characteristics of the data. Commonly used methods include weighted average, principal component analysis, and Kalman filtering. In this embodiment, the weighted average method is used for fusion.

[0083]

[0084] Among them, D fused For the merged data, D i For the i-th data source, w i The corresponding weights can be set based on factors such as the signal-to-noise ratio and reliability of the data.

[0085] Analysis of merged results: The merged data is analyzed to assess its accuracy and reliability. The validity of the merged results can be verified through cross-validation or comparison with known geological features.

[0086] Generate a comprehensive report: Integrate the analysis results to form a detailed exploration report, including the distribution of mineralization anomalies, ore body characteristics, and other important information, providing decision support for subsequent exploration work.

[0087] Step 4: Intelligent Analysis and Recognition

[0088] 4.1 Applications of Machine Learning and Deep Learning

[0089] Model training:

[0090] Objectives and methods:

[0091] In this embodiment, machine learning algorithms (such as random forests and support vector machines) and deep learning algorithms (such as convolutional neural networks, CNNs) are used to perform in-depth analysis of the comprehensive feature matrix to identify mineralization anomalies and ore body characteristics. Through model training, patterns and regularities in the data can be effectively captured.

[0092] Specific implementation steps:

[0093] Data preparation:

[0094] Collect labeled training samples, including data from known mineralized and unmineralized regions. These samples will serve as the basis for model training, ensuring that the model learns effective features.

[0095] Data partitioning:

[0096] The dataset is divided into training and test sets, typically in an 80 / 20 ratio. The training set is used to train the model, while the test set is used to validate the model's generalization ability.

[0097] Model selection and training:

[0098] Machine learning models:

[0099] For random forests and support vector machines, first select appropriate hyperparameters (such as the number of trees, depth, etc.), and then use cross-validation to evaluate the model performance.

[0100] During training, the model parameters are continuously optimized, for example, by using grid search to find the optimal hyperparameter configuration.

[0101] Deep learning models:

[0102] The architecture of a convolutional neural network includes an input layer, multiple convolutional layers, pooling layers, and fully connected layers. Convolutional layers extract spatial features, while pooling layers reduce data dimensionality.

[0103] The network is trained using the backpropagation algorithm, and the weights are gradually adjusted to minimize the loss function (e.g., cross-entropy loss).

[0104] Monitor performance on the validation set during training to prevent overfitting.

[0105] Optimization and Evaluation:

[0106] After model training is complete, the model performance is evaluated using a test set, and metrics such as accuracy, precision, recall, and F1 score are calculated. This ensures that the model has high recognition accuracy in real-world applications.

[0107] 4.2 Results Output and Report Generation

[0108] Generate exploration report:

[0109] Report content and structure:

[0110] Based on the analysis results, a detailed exploration report will be generated, which should include the following parts:

[0111] Introduction: This section outlines the exploration objectives, methods, and background information of the study area.

[0112] Data sources and processing methods: Describe in detail the steps of data collection, processing, and analysis to ensure the transparency of the results.

[0113] Mineralization area distribution map: Use GIS tools to visualize the identified mineralization areas, generate a map, and mark the location and extent of the ore bodies.

[0114] Feature description: Provides a detailed description of the features of the mineralized area, including information such as the depth, shape, and material composition of the ore body.

[0115] Potential value assessment: Based on the identified mineralization characteristics, a potential economic value assessment is conducted to provide a basis for subsequent mineral development.

[0116] Output and Publishing:

[0117] Use professional report generation software (such as LaTeX or Microsoft Word) to format the report, ensuring the content is clear and easy to understand. The report should ultimately be submitted to relevant decision-making bodies via email or online platforms to ensure timely dissemination and use of information.

[0118] Example 2:

[0119] This embodiment uses traditional data analysis methods to explore a certain metal mining area. Similar to Embodiment 1, the exploration is conducted in a certain metal mining area where known mineralization phenomena exist. The goal is to identify the distribution and characteristics of ore bodies.

[0120] Step 1: Data Collection

[0121] Data collection methods:

[0122] In this embodiment, a traditional ground sampling method was used. Five fixed sampling points were set up within the mining area, and soil and rock samples were collected manually at each point. Each sample weighed 500 grams.

[0123] Data Records:

[0124] Record the GPS coordinates and environmental conditions (such as temperature and humidity) of each sampling point, and number each sample to ensure traceability.

[0125] Step 2: Data Processing

[0126] Data analysis methods:

[0127] Conventional laboratory analytical methods, such as X-ray fluorescence spectroscopy (XRF) and flame photometry, were used to assess the composition and content of the mineral samples.

[0128] For XRF analysis, the sample needs to be pulverized, prepared, and placed in the analyzer for detection. The analysis time for each sample is approximately 30 minutes.

[0129] Step 3: Results Analysis

[0130] Data statistics:

[0131] All samples were summarized and statistically analyzed to calculate the average content, standard deviation, and other indicators of each element, and corresponding bar charts were drawn.

[0132] Mineralized areas are identified manually, mineralization distribution maps are drawn manually, and their characteristics and potential value are recorded.

[0133] In Example 1, comprehensive mineralization information can be quickly acquired and distribution maps can be automatically generated through UAV hyperspectral data acquisition and deep learning algorithms. In Example 2, however, data acquisition and analysis are performed manually, which is time-consuming.

[0134] The results of Example 1 show that the accuracy rate of identifying mineralized areas reached over 90%, while that of Example 2 was only around 70%, and the analysis cycle was significantly extended.

[0135] By comparing the efficiency and accuracy of the two implementation schemes, the advantages of Implementation Scheme 1 in data acquisition and processing are clearly identified, further proving that it is the optimal solution.

[0136] Table 1 Comparison of Experimental Data from Examples (Table 1)

[0137]

[0138]

[0139] By comparing the tables, it can be seen that Example 1 is superior to Example 2 in terms of data acquisition efficiency, analysis accuracy, and time, proving that the integrated data acquisition and intelligent analysis method of Example 1 is the optimal solution.

[0140] Example 3:

[0141] This embodiment uses a ground sensor for single data acquisition, aiming to further verify the superiority of the integrated data acquisition and intelligent analysis method in Embodiment 1.

[0142] Similar to Example 1, this study was conducted in a metal mining area where known mineralization was present, with the goal of identifying the distribution and characteristics of ore bodies.

[0143] Step 1: Data Collection

[0144] Data collection methods:

[0145] In this embodiment, only a high-precision geomagnetic instrument (such as GSM-19) is deployed in the mining area, and five key sampling points are set up for data collection.

[0146] Each sampling point continuously records geomagnetic data at a fixed location, with a recording frequency set to 1 Hz and a data acquisition time of 1 hour.

[0147] Data Records:

[0148] The GPS coordinates, environmental conditions (such as temperature and humidity), and recorded geomagnetic data of each sampling point must be recorded in detail to ensure data traceability.

[0149] Step 2: Data Processing

[0150] Data analysis methods:

[0151] The collected geomagnetic data were processed using conventional methods, and time-domain analysis was employed to calculate the geomagnetic intensity variation curve for each sampling point and identify abnormal peak values.

[0152] The identification of abnormal peaks relies on manually set thresholds, requiring manual analysis of each dataset.

[0153] Step 3: Results Analysis

[0154] Mineralization anomaly identification:

[0155] The presence of mineralization anomalies is determined by manually analyzing the geomagnetic data at each sampling point. Identification of mineralization areas relies primarily on the experience and judgment of technical personnel.

[0156] Table 2 Comparison of Experimental Data from Examples (Table 2)

[0157]

[0158]

[0159] By comparing the tables, Example 1 is significantly better than Example 3 in terms of data acquisition efficiency, analysis accuracy, and recognition effect, further verifying that the comprehensive data acquisition and intelligent analysis method of Example 1 is the optimal solution.

[0160] Comparative Example 1:

[0161] This embodiment employs traditional mineral exploration methods, primarily relying on manual observation and simple laboratory analysis. It is conducted within the same metal mining area where known mineralization phenomena exist, with the goal of identifying the distribution and characteristics of ore bodies.

[0162] Step 1: Data Collection

[0163] Data collection methods:

[0164] In this embodiment, a traditional field survey method is adopted. First, the exploration team randomly selects 10 sampling points in the mining area and uses hand tools (such as shovels and picks) to collect soil and rock samples.

[0165] Each sampling point requires the collection of approximately 1 kilogram of sample. During the sampling process, the GPS coordinates of the sampling point and the observed mineral characteristics (such as color, texture, etc.) must be recorded.

[0166] Step 2: Data Processing

[0167] Data analysis methods:

[0168] The collected samples will be sent to the laboratory for basic chemical analysis, using flame photometry and gravity separation to preliminarily assess the metal content. This process typically takes several days, and the sample analysis is quite time-consuming.

[0169] The analysis results only provide basic information on mineral composition and content, lacking in-depth data mining.

[0170] Table 3 Comparative Experiment Data Table

[0171]

[0172] As can be seen from the comparison table, Example 1 is significantly superior to traditional mineral exploration methods in all aspects, further proving the feasibility of the method.

[0173] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for extracting data in geological and mineral exploration, characterized in that, Includes the following steps: Step 1: Collect comprehensive data on the target area using a drone equipped with a hyperspectral sensor. The data includes spectral information in multiple bands. Step 2: Deploy ground sensors at designated sampling points to collect aeromagnetic data and geological samples. The operating frequency of the ground sensors is set to 1 Hz. Step 3: Use wavelet transform technology to remove noise from the collected hyperspectral and aeromagnetic data, removing signals below a set threshold; Step 4: Extract features through principal component analysis to generate a comprehensive feature matrix, and integrate different data sources through multidimensional data fusion methods to enhance data reliability; Step 5: Analyze the comprehensive feature matrix using machine learning and deep learning algorithms to automatically identify mineralization anomalies and ore body characteristics, and generate an exploration report.

2. The method for extracting geological and mineral exploration data according to claim 1, characterized in that, The drone is set to fly at an altitude of 100 meters and a speed of 5 meters per second to ensure that the coverage rate during data collection is not less than 80%.

3. The method for extracting geological and mineral exploration data according to claim 1, characterized in that, The ground sensors include a high-precision magnetometer and a laser scanner. The magnetometer can record changes in magnetic field strength in real time, and the laser scanner has a resolution of no less than 1 cm to acquire high-resolution terrain point cloud data.

4. The method for extracting geological and mineral exploration data according to claim 1, characterized in that, The sample collection steps include at least selecting 10 representative sampling points within the target area, extracting at least 500 grams of soil or rock samples from each sample using standard geological sampling tools, and recording the geographical location and characteristics of the samples on-site.

5. The method for extracting data in geological and mineral exploration according to claim 1, characterized in that, In the noise elimination step, wavelet transform is used to perform multi-level signal decomposition, retaining high-frequency components related to geological features and removing noise below a set threshold.

6. The method for extracting data in geological and mineral exploration according to claim 1, characterized in that, In the feature extraction step, principal component analysis is used to construct a covariance matrix to identify principal components. This step includes calculating eigenvalues ​​and eigenvectors, and selecting principal components with a cumulative contribution rate of 95% for analysis.

7. The method for extracting data in geological and mineral exploration according to claim 1, characterized in that, The data fusion step employs a weighted fusion technique, combining hyperspectral data and aeromagnetic data, as shown in the formula: F combined =αF yperspectral +βF magnetic Here, α and β are weighting coefficients set according to the importance of each data source.

8. The method for extracting data in geological and mineral exploration according to claim 1, characterized in that, The machine learning algorithms include random forests and convolutional neural networks, used to classify and identify mineralization anomalies, wherein the training dataset includes at least 1,000 samples.

9. The method for extracting data in geological and mineral exploration according to claim 1, characterized in that, The analysis results are presented through visualization tools to generate easy-to-understand geological maps and exploration reports, assisting exploration personnel in making decisions.

10. The method for extracting data in geological and mineral exploration according to claim 1, characterized in that, The method also includes real-time monitoring and feedback of exploration data, and using a cloud computing platform to store and process the data for access and analysis.

Images