A metallogenic prediction method and system integrating fractal characterization engineering

Abstract

The present invention belongs to the technical field of metallogenic prediction and discloses a metallogenic prediction method integrating fractal characterization engineering. The purpose of the present invention is to provide a metallogenic prediction method integrating fractal characterization engineering, which is characterized in that the original data is processed through fractal analysis and feature selection to highlight the non-linear metallogenic information of the data set, screen out the optimal feature subset, and obtain the optimized data set. Predictions are made through this data set to improve the prediction accuracy of the artificial intelligence model, ultimately obtain a more accurate prediction result than traditional methods, reduce the labor and economic costs of exploration, reduce exploration risks, and further provide a more scientific, reliable and efficient metallogenic prediction method. It is applicable to the metallogenic prediction work of various artificial intelligence models. Further provided is an implementation method of the fractal characterization engineering.

Description

Technical Field

[0001] The present invention belongs to the technical field of mineralization prediction, and in particular relates to a mineralization prediction method and system integrating fractal characterization engineering. Background Art

[0002] Mineral resources are an important material basis for economic and social development. With the advancement of urbanization and industrialization, the demand for mineral resources is increasing, and it is particularly important to find new mineral prospecting areas. In addition, mineral exploration is characterized by high costs and high risks. How to obtain more accurate mineral prospecting targets has become a problem that needs to be solved urgently. Mineral resource potential evaluation maps can quantify and draw the probability of discovering mineral deposits through exploration in the study area. At present, artificial intelligence-driven mineral resource potential evaluation has become an increasingly accepted important tool for delineating mineral exploration targets. However, due to the inherent complexity and noise of the geological characteristics used in mineral resource potential evaluation, the application of artificial intelligence models in actual exploration is subject to certain limitations.

[0003] The mineralization prediction method technology integrated with fractal characterization engineering can mine potential nonlinear mineralization information from the original features, search for the optimal feature subset of the input data set, and meet the high-precision requirements in current mineral resource exploration.

[0004] Limitations of traditional algorithms: Existing AI-driven mineral resource potential evaluation usually only performs simple preprocessing on the original data set, which will largely miss the potential nonlinear mineralization information in the original features and make it difficult to search for the optimal feature subset from the input data set, resulting in deviations in the final prediction results, missing potential mineralization areas or exaggerating areas with low mineralization potential, causing inconvenience to actual exploration work.

[0005] Through the above analysis, the problems and defects of the prior art are as follows:

[0006] Existing AI-driven mineral resource potential evaluation usually only performs simple preprocessing on the original data set, which will largely miss the potential nonlinear mineralization information in the original features and make it difficult to search for the optimal feature subset from the input data set, resulting in deviations in the final prediction results, missing potential mineralization areas or exaggerating areas with low mineralization potential, causing inconvenience to actual exploration work. Summary of the invention

[0007] In view of the problems existing in the prior art, the present invention provides a mineralization prediction method integrating fractal characterization engineering.

[0008] The present invention is implemented in this way: a mineralization prediction method integrating fractal characterization engineering comprises:

[0009] S1, obtain the mineralization prediction data set and perform preprocessing;

[0010] Obtain a mineralization prediction dataset; the dataset contains labels, sample point coordinates, and sample features, covering various favorable mineralization factors;

[0011] S2, fractal analysis and feature selection of the data set;

[0012] Construct a mineralization prediction framework integrating fractal characterization engineering;

[0013] S3, model training and optimization;

[0014] After completing fractal analysis and feature selection, select the prediction model, train the model, and find the optimal parameter combination to further improve the model's adaptability and prediction accuracy;

[0015] S4, model predictions;

[0016] The trained and optimized mineralization prediction model can be used to predict the mineralization probability in the study area and assist in delineating the prospecting target area;

[0017] S5, result output and evaluation.

[0018] Further, the mineralization prediction data set is obtained and preprocessed:

[0019] (a) Denoising: Geological features are inherently complex and noisy, so the collected data often contain some noisy data, and it is necessary to process missing values, duplicate items, etc. in the data;

[0020] (b) Data normalization: To ensure the subsequent prediction effect, the data is normalized.

[0021] Further, the data set is subjected to fractal analysis and feature selection:

[0022] (1) Fractal and multifractal analysis: The box dimension method is used to describe the fractal characteristics of the mineral points, and then the Fry analysis is used to interpret the fractal results to enhance the subtle morphology of the mineral occurrence; the spatial correlation between these occurrences and geological characteristics is characterized to mine the mineralization distribution pattern; the sliding window technology is used to process the evidence layer of the original features to assist in capturing the local features within the massive prediction units, and multifractal analysis is performed to obtain the fractal representation of the original features;

[0023] (2) Feature selection: Based on fractal processing, a variety of feature selection methods are used for analysis. The analysis results of various methods are combined to obtain the best combination of fractal indicator features and construct a fractal index feature data set for model prediction. The specific feature selection methods used are: prediction-area chart, K-means clustering, information gain, chi-square test and Pearson correlation coefficient.

[0024] Furthermore, the model training and optimization are as follows:

[0025] (a) Select a prediction model: With the rapid development of artificial intelligence algorithms, many excellent artificial intelligence algorithms have emerged. The models commonly used for metallogenic prediction include: artificial neural network, random forest, support vector machine, and convolutional neural network;

[0026] (b) Model training and optimization: By using the random grid search method and ten-fold cross-validation method to find the optimal parameter combination, further improve the prediction ability of the model; In addition, in order to minimize the influence of randomness, the data set should also be randomly selected multiple times to form different training sets for training respectively, and finally take the average value as the result.

[0027] Furthermore, the model prediction is as follows:

[0028] (a) Process the prediction data set: The prediction data set should contain feature columns, coordinates, and ore point label information; By preprocessing the prediction data, perform fractal characterization on the prediction data set;

[0029] (b) Input the prediction data: Input the processed data to be predicted into the prediction models trained by different training sets for prediction respectively. The model will make predictions based on the sample features, and the final prediction result takes the average value;

[0030] (c) Quantitative analysis: The model will output the metallogenic probability of each sample point. Combining the success rate curve and the uncertainty scatter plot, screen out the high-potential metallogenic areas and low-risk exploration areas.

[0031] Furthermore, the result output and evaluation are as follows:

[0032] (a) Draw a map of the potential evaluation of mineral resources: Combining the prediction results of the model with the delineated high-potential metallogenic areas and low-risk exploration areas, obtain a map of the potential evaluation of mineral resources and delineate the prospecting target areas;

[0033] (b) Result evaluation: Output the evaluation indicators and feature importance of the model, and combine the existing exploration experience and metallogenic research to evaluate the prediction results of the model, which is convenient for carrying out subsequent geological research and mineral exploration work.

[0034] Another object of the present invention is to provide a metallogenic prediction system integrating fractal characterization engineering, including:

[0035] A preprocessing module, which is used to obtain and preprocess the metallogenic prediction data set; Obtain the metallogenic prediction data set; This data set contains labels, sample point coordinates, and sample features, covering various favorable factors for metallogenesis;

[0036] A feature selection module for performing fractal analysis and feature selection on a dataset; constructing a metallogenic prediction framework integrating fractal characterization engineering;

[0037] An optimization module for model training and optimization; after completing fractal analysis and feature selection, a prediction model is selected, the model is trained to find the optimal parameter combination, and further improve the adaptability and prediction accuracy of the model;

[0038] A prediction module for model prediction; the trained and optimized metallogenic prediction model can be used to predict the metallogenic probability in the study area and assist in delineating ore prospecting target areas;

[0039] An evaluation module for result output and evaluation.

[0040] Another object of the present invention is to provide a computer device, which includes a memory and a processor, the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the metallogenic prediction method integrating fractal characterization engineering.

[0041] Another object of the present invention is to provide a computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, the processor executes the steps of the metallogenic prediction method integrating fractal characterization engineering.

[0042] Another object of the present invention is to provide an information data processing terminal for implementing the metallogenic prediction system integrating fractal characterization engineering.

[0043] Combined with the above technical solutions and solved technical problems, the advantages and positive effects of the technical solution to be protected by the present invention are as follows:

[0044] First, the object of the present invention is to provide a metallogenic prediction method integrating fractal characterization engineering, which is characterized in that the original data is processed through fractal analysis and feature selection, highlighting the non-linear metallogenic information of the dataset, screening out the optimal feature subset, and obtaining an optimized dataset. Predicting through this dataset can improve the prediction accuracy of the artificial intelligence model, and finally obtain a more accurate prediction result than the traditional method, reduce the human and economic costs of exploration, reduce the exploration risk, and thus provide a more scientific, reliable and efficient metallogenic prediction method. It is applicable to the metallogenic prediction work of various artificial intelligence models. Further provide an implementation method of this fractal characterization engineering.

[0045] The present invention provides a metallogenic prediction method integrating fractal characterization engineering, which is characterized in that the original data is processed through fractal analysis and feature selection, highlighting the non-linear metallogenic information of the dataset, screening out the optimal feature subset, and obtaining an optimized dataset.

[0046] Second, the technical solution of the present invention solves the technical problems that people have been eager to solve but have never been successful in: many metallogenic geological information are indirect information, that is, it is necessary to effectively extract and interpret the original data to obtain information beneficial to metallogenic prediction. The effectiveness of information extraction and interpretation will directly affect the accuracy of metallogenic prediction. The geological information related to mineralization is the product of the coupling of multiple fields and multiple processes such as mechanical deformation, thermodynamics, fluid flow, temperature change, etc., and has experienced a long geological evolution with a million years as the basic unit. Therefore, it has significant complexity and nonlinearity. Cracking the complexity and nonlinearity of metallogenic geological information and improving the effectiveness of information extraction and interpretation have always been difficult problems in the field of metallogenic prediction and are also the prerequisite for efficient metallogenic prediction. In view of this problem, the present invention uses fractal measurement as a tool to characterize the complexity and nonlinearity of metallogenic geological information, accurately tracks and captures the trace feature information left by the metallogenic process, and improves the interpretation accuracy of metallogenic geological information and the accuracy of metallogenic prediction.

[0047] Third, the metallogenic prediction method of the present invention significantly improves the accuracy and efficiency of metallogenic prediction in industrial applications by integrating the fractal characterization engineering and the metallogenic prediction model, solves the limitations of the prediction methods in the prior art, and has made significant technological progress.

[0048] 1. Solved the problem of insufficient accuracy of metallogenic prediction

[0049] In traditional metallogenic prediction methods, it is often difficult to accurately describe the multi-scale characteristics of geological data, resulting in insufficient prediction accuracy. The present invention adopts fractal characterization engineering, which can identify the multi-scale characteristics of data and extract important metallogenic favorable factors. Through fractal analysis, the expression ability of data characteristics is improved, the input data of the model is optimized, the accuracy of metallogenic prediction is effectively enhanced, and the delineation of metallogenic target areas is more accurate.

[0050] 2. Improved the efficiency of data processing and feature selection

[0051] The present invention introduces feature selection technology in the data preprocessing stage, removes redundant information through fractal analysis, and extracts key metallogenic factors. Compared with the prior art methods, this technical means simplifies the data processing flow, significantly reduces the computational complexity, improves the efficiency of the model, makes the prediction process faster and more efficient, and meets the processing requirements of larger-scale metallogenic data.

[0052] 3. Enhanced the adaptability and stability of the model

[0053] Through multiple trainings and parameter optimizations of the model, the present invention constructs a metallogenic prediction model with high stability and adaptability. Compared with the prediction models based on traditional single algorithms, the method of the present invention can maintain a high prediction accuracy in various geological environments, providing support for the diversity of mining areas and significantly improving adaptability and stability.

[0054] 4. Reduced prospecting costs and exploration risks

[0055] Traditional prospecting usually requires a large amount of manpower and material resources, and due to high uncertainty, it is easy to cause waste of costs in blind exploration. Through the efficient prediction of metallogenic probability, the present invention helps geological teams effectively delineate metallogenic target areas, avoid unnecessary resource waste, thereby significantly reducing exploration costs and risks, and providing efficient auxiliary decision-making support for the mining industry.

[0056] 5. Provided visual metallogenic prediction results

[0057] The present invention outputs the prediction results as a metallogenic probability map, which, combined with prediction evaluation indicators, provides users with intuitive metallogenic distribution information. This visual way not only facilitates the understanding and application of the results, but also provides a reliable reference for decision-making levels, further enhancing the practicality of the prediction method in actual industrial applications.

[0058] In summary, the metallogenic prediction method of the present invention solves the problems of low accuracy, complexity and high cost in traditional metallogenic prediction. Through innovative means such as data feature extraction, model optimization and fractal analysis, it has achieved significant technological progress and provided efficient, accurate and practical technical support for mining exploration. Description of the Drawings

[0059] Figure 1 is a flowchart of the metallogenic prediction method integrating fractal characterization engineering provided by an embodiment of the present invention.

[0060] Figure 2 is a flowchart of the fractal analysis and feature selection method for the dataset provided by an embodiment of the present invention.

[0061] Figure 3 is a structural block diagram of the metallogenic prediction system integrating fractal characterization engineering provided by an embodiment of the present invention.

[0062] Figure 4 is a comparison chart of the comprehensive performance of the model integrating fractal characterization engineering and traditional methods provided by an embodiment of the present invention.

[0063] Figure 5 is a comparison chart of the metallogenic prospects integrating fractal characterization engineering and traditional methods provided by an embodiment of the present invention. Detailed Embodiments

[0064] In order to make the objectives, technical solutions and advantages of the present invention more clear and understandable, the present invention will be further described in detail below in conjunction with embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0065] As Figure 1 shown, a metallogenic prediction method integrating fractal characterization engineering provided by an embodiment of the present invention includes the following steps:

[0066] S1. Obtain a metallogenic prediction data set and perform preprocessing;

[0067] Obtain a metallogenic prediction data set; this data set contains labels, sample point coordinates, and sample features, covering various metallogenic favorable factors;

[0068] S2. Perform fractal analysis and feature selection on the data set;

[0069] Construct a metallogenic prediction framework integrating fractal characterization engineering;

[0070] S3. Model training and optimization;

[0071] After completing fractal analysis and feature selection, select a prediction model, train the model, find the optimal parameter combination, and further improve the adaptability and prediction accuracy of the model;

[0072] S4. Model prediction;

[0073] The trained and optimized metallogenic prediction model can be used to predict the metallogenic probability in the study area and assist in delineating ore prospecting target areas;

[0074] S5. Result output and evaluation.

[0075] In the process of metallogenic prediction, the acquisition and preprocessing of data are important basic steps. The metallogenic prediction data set includes labels, sample point coordinates, and sample features of various metallogenic favorable factors. First, collect relevant data through means such as geological exploration and remote sensing to form a comprehensive data set containing metallogenic conditions and features. The preprocessing steps include data cleaning, normalization processing, and processing of missing values, etc., so as to ensure the integrity and consistency of the data and provide high-quality data input for subsequent steps.

[0076] Fractal analysis is an effective method for describing complex geological phenomena and can identify multi-scale features in the data set. By constructing a fractal characterization engineering, important geological features in the sample data are identified, and redundant information is removed through feature selection, and the most favorable features for metallogenic prediction are extracted. The combination of fractal analysis and feature selection enables the metallogenic prediction framework to not only handle the complexity of the data, but also improve the efficiency and accuracy of the prediction model.

[0077] After completing the fractal analysis and feature selection, the next step is to construct and train the model. Using the selected important features, select a suitable metallogenic prediction model (such as neural network, random forest, etc.), and optimize the model parameters. Through multiple trainings and tests, continuously adjust the model parameters to find the optimal parameter combination, so that the model can better adapt to the data characteristics and further improve the adaptability and prediction accuracy of the model.

[0078] The trained and optimized model is used to predict the metallogenic probability of the study area. Input the new regional geological data into the model, and the model outputs the metallogenic possibility of this area. This step helps geologists effectively delineate the prospecting target area, reduces the blindness and cost of exploration to a certain extent, and provides a more accurate metallogenic probability distribution map.

[0079] The prediction results are output as a metallogenic probability map through a visualization tool, and the results are evaluated. Compare the actual prospecting data of the metallogenic target area with the model prediction results to evaluate the prediction effect of the model. Evaluate the performance of the model through indicators such as accuracy, recall rate, and F1 score, analyze the degree of agreement between the results and the actual situation, and verify the prediction reliability and stability of the model.

[0080] According to the evaluation feedback of the prediction results, further improve the model, adjust the details of the fractal representation and feature selection, or introduce more complex algorithms to improve the adaptability of the model. Such a cyclic feedback mechanism makes the model closer to the real geological metallogenic environment during continuous optimization, enhancing the practical application effect of this metallogenic prediction method.

[0081] The embodiment of the present invention provides obtaining a metallogenic prediction data set and performing preprocessing:

[0082] Obtain a metallogenic prediction data set; this data set contains labels, sample point coordinates, and sample features (such as: geological features, geophysical features, geochemical features, etc.), covering various metallogenic favorable factors; to improve the prediction accuracy of the model, it is necessary to perform preprocessing on the obtained metallogenic prediction data; the preprocessing operations include the following:

[0083] (a) Denoising processing: Geological features have inherent complexity and noise, so the collected data often contains some noisy data, and it is necessary to process the missing values, duplicate items, etc. in the data to ensure the integrity and reliability of the data;

[0084] (b) Data normalization: To ensure the subsequent prediction effect, perform data normalization processing, which can improve the prediction efficiency and also improve the prediction accuracy of the model to a certain extent;

[0085] Obtain a processed metallogenic prediction data set through preprocessing, providing data guarantee for the subsequent training and prediction of the model.

[0086] As Figure 2 shown, the present invention provides a fractal analysis and feature selection for a data set:

[0087] The core of the present invention lies in constructing a metallogenic prediction framework integrating fractal characterization engineering; the specific steps are as follows:

[0088] S201, Fractal and multifractal analysis:

[0089] The box dimension method is used to describe the fractal features of ore points, and then the Fry analysis is used to interpret the fractal results to enhance the subtle morphology of the mineral occurrence; the spatial correlation between these occurrences and geological features is characterized, and the mineralization distribution pattern is mined; the sliding window technique is used to process the evidence layer of the original features to assist in capturing the local features within a large number of prediction units for multifractal analysis to obtain the fractal representation of the original features;

[0090] S202, Feature selection:

[0091] Based on the fractal processing, multiple feature selection methods are used for analysis, and the analysis results of multiple methods are combined to obtain the best combination of fractal index features, and a fractal index feature data set is constructed for model prediction; the specific feature selection methods used are: prediction-area diagram, K-means clustering, information gain, chi-square test, Pearson correlation coefficient, etc.;

[0092] Combining fractal analysis and feature selection, the potential non-linear metallogenic information in the original features of the data set is further mined, and the optimal feature subset of the input data set is constructed; through the method of fractal characterization, the present invention can effectively improve the accuracy of metallogenic prediction.

[0093] Model training and optimization provided by the embodiments of the present invention:

[0094] After completing the fractal analysis and feature selection, a prediction model is selected, the model is trained to find the optimal parameter combination, and the adaptability and prediction accuracy of the model are further improved; the specific process is as follows:

[0095] (a) Selecting a prediction model: With the rapid development of artificial intelligence algorithms, many excellent artificial intelligence algorithms have emerged. The models commonly used for metallogenic prediction are: artificial neural network, random forest, support vector machine, convolutional neural network, etc.;

[0096] (b) Model training and optimization: The optimal parameter combination is found by using the random grid search method and the ten-fold cross-validation method to further improve the prediction ability of the model; in addition, in order to minimize the influence of randomness, the data set should also be randomly selected multiple times to form different training sets for training respectively, and finally the average value is taken as the result to improve the credibility of the final prediction;

[0097] Through the above model training and optimization steps, a well-trained metallogenic prediction model is obtained and adapted to the metallogenic prediction task of the given fractal characterization dataset.

[0098] The model prediction provided by the embodiments of the present invention is as follows:

[0099] The metallogenic prediction model trained and optimized can be used to predict the metallogenic probability in the study area and assist in delineating the prospecting target areas. The specific steps are as follows:

[0100] (a) Process the prediction dataset: The prediction dataset should contain information such as feature columns, coordinates, and ore point labels (optional); preprocess the prediction data by the method in Step 1 above, and perform fractal characterization on the prediction dataset by the method in (1) of Step 2.

[0101] (b) Input the prediction data: Input the processed data to be predicted into the prediction models trained with different training sets for prediction respectively. The model will make predictions based on the sample features, and the final prediction result is the average value.

[0102] (c) Quantitative analysis: The model will output the metallogenic probability of each sample point. Combining the success rate curve and the uncertainty scatter plot, screen out the high-potential metallogenic areas and low-risk exploration areas.

[0103] Through the above steps, the present technology can achieve the metallogenic prediction work for a certain area and finally output the metallogenic probability distribution map of the area.

[0104] The result output and evaluation provided by the embodiments of the present invention are as follows:

[0105] (a) Draw the mineral resource potential evaluation map: Combine the prediction results of the model with the delineated high-potential metallogenic areas and low-risk exploration areas to obtain the mineral resource potential evaluation map and delineate the prospecting target areas.

[0106] (b) Result evaluation: Output the evaluation indicators and feature importance of the model, and combine the existing exploration experience and metallogenic research to evaluate the prediction results of the model, which is convenient for carrying out subsequent geological research and mineral exploration work.

[0107] As Figure 3 shown, a metallogenic prediction system integrating fractal characterization engineering provided by the embodiments of the present invention includes:

[0108] A preprocessing module, configured to obtain a metallogenic prediction dataset and perform preprocessing; obtain a metallogenic prediction dataset; the dataset contains labels, sample point coordinates, and sample features, covering various metallogenic favorable factors.

[0109] A feature selection module for performing fractal analysis and feature selection on a dataset; constructing a metallogenic prediction framework integrating fractal characterization engineering;

[0110] An optimization module for model training and optimization; after completing fractal analysis and feature selection, select a prediction model, train the model, find the optimal parameter combination, and further improve the adaptability and prediction accuracy of the model;

[0111] A prediction module for model prediction; the trained and optimized metallogenic prediction model can be used to predict the metallogenic probability in the study area and assist in delineating ore prospecting target areas;

[0112] An evaluation module for result output and evaluation.

[0113] Another object of the present invention is to provide a computer device, the computer device includes a memory and a processor, the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the metallogenic prediction method integrating fractal characterization engineering.

[0114] Another object of the present invention is to provide a computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, the processor executes the steps of the metallogenic prediction method integrating fractal characterization engineering.

[0115] Another object of the present invention is to provide an information data processing terminal for implementing the metallogenic prediction system integrating fractal characterization engineering.

[0116] Specific implementation of the present invention:

[0117] Example 1

[0118] (1) Obtain a metallogenic prediction dataset.

[0119] Obtain a metallogenic prediction dataset, and the dataset can include various metallogenic favorable factors, such as: geological features, geophysical features, and geochemical features favorable for mineralization, etc. The samples include three types: ore points, non-ore points, and samples to be predicted.

[0120] Preprocessing steps: (a) Denoising processing: First, perform data cleaning on the dataset, process the noise data, and ensure the accuracy of model training and prediction. (b) Data normalization: Perform normalization processing on the data, accelerate the learning process of the model, improve the accuracy, and weaken the influence of outliers.

[0121] (2) Perform fractal characterization on the dataset.

[0122] The box dimension method is used to describe the fractal features of ore points and non-ore points in the training set. The Fry analysis is used to interpret the fractal results, enhance the subtle morphology of the mineral occurrence, and depict the spatial correlation between these occurrences and geological features. The sliding window is used to process the original evidence layer to better capture the local features within the prediction unit.

[0123] The specific process of obtaining data by the sliding window is as follows: (a) The feature to be studied is covered by a grid with side length m. A window with side length m slides from left to right and from bottom to top in turn. (b) The lower left corner of the feature map is set as the starting position, and the window slides until it reaches the key position in the upper right corner of the image.

[0124] After obtaining the data by the sliding window, the multi-fractal spectrum is estimated by combining the moment method, and then parameters such as the fractal dimension and singularity index are obtained to perform fractal characterization on the original features.

[0125] Methods such as the prediction-area graph, K-means clustering, information gain, chi-square test, and Pearson correlation coefficient are used in turn for feature selection. The results obtained by different methods are jointly ranked, and finally the best combination of fractal index features is obtained to complete the fractal characterization of the data set.

[0126] (3) Training and optimization of the model.

[0127] (a) Training process:

[0128] The artificial intelligence model is trained using the fractal characterization data set to find the optimal parameters to improve its accuracy and reliability in ore-forming prediction. The specific training steps are as follows:

[0129] Randomly select negative samples with the same number as the positive samples in the data set characterized by fractals, and divide them into a training set and a test set in a ratio of 7:3, and ensure that the positive and negative samples are evenly distributed in the training set and the test set in a ratio of 1:1.

[0130] Set the corresponding hyperparameters according to the selected artificial intelligence model.

[0131] The random grid search method is combined with ten-fold cross-validation to search for the optimal parameter combination to improve the prediction accuracy of the model.

[0132] (b) Model optimization:

[0133] To minimize the influence of randomness, the process of randomly splitting the data set is carried out 10 times to generate ten training data sets. The model is trained on the ten data sets respectively to improve the reliability of the model.

[0134] (4) Model prediction.

[0135] For model prediction, the prediction data should be input into different models trained by ten training data sets for prediction respectively. Finally, the average value of the results of 5 models is taken for the prediction results and model evaluation indicators. The average metallogenic probability distribution map in the study area is obtained, and the metallogenic areas are briefly divided.

[0136] (5) Mineral resource potential evaluation.

[0137] (a) Divide different metallogenic potential areas: Draw the success rate curve of the prediction results, and divide the high-potential, medium-potential, low-potential and background areas of mineralization according to the curve slope.

[0138] (b) Define the prospecting target areas: Combine the high-potential metallogenic areas divided in (a) through the uncertainty scatter plot, and divide the exploration areas with high potential and low risk as the prospecting target areas.

[0139] (c) Result evaluation: Combine the existing metallogenic theories and exploration experiences to evaluate the results of model prediction and various evaluation indicators, and judge the reliability of the prediction results.

[0140] The generated results include the metallogenic prediction results of the target area and the potential metallogenic areas, and show the metallogenic prediction distribution map of the target area. This technology is widely applicable to multiple fields such as mineral resource exploration and ore deposit evaluation, and provides a more reliable, scientific and efficient metallogenic prediction method. The above is only a specific embodiment of the present invention, but the technical features of the present invention are not limited thereto.

[0141] The present invention is specifically applied to the technical field of metallogenic prediction, and particularly relates to a metallogenic prediction method and system integrating fractal characterization engineering.

[0142] The following are the relevant results obtained in the embodiments of the present invention.

[0143] The comprehensive performance of the models predicted by the fractal characterization data set and the original data set is as Figure 4 shown, Figure 4 (a), (b), and (c) The evaluation indicators are the prediction efficiency, accuracy rate, and kappa coefficient of the model respectively. The results are evaluated from the perspective of actual exploration through the prediction efficiency, and the classification accuracy is measured from the algorithm perspective through the accuracy rate and kappa coefficient. It can be intuitively seen that for the prediction efficiency, prediction accuracy rate, and kappa coefficient of all artificial intelligence models, in the same model, the models trained by the fractal characterization data set are higher than those trained by the original data set.

[0144] The final metallogenic prospect maps generated by each artificial intelligence model are shown in Figure 5 , where Figure 5 (a), (c) 、 (e) 、(g) are the prediction results of the fractal characterization dataset in different artificial intelligence models (in turn: artificial neural network, random forest, decision tree, logistic regression), and figures (b), (d), (f), (h) are the prediction results of the original dataset in different artificial intelligence models. It can intuitively indicate the high-potential metallogenic areas in the study area.

[0145] Combined with Figure 4 and Figure 5 , taking the prediction efficiency of the artificial neural network model as an example for analysis. The prediction efficiency of the artificial neural network model trained with the original dataset is 8.6723, and 70.34% of the known ore points are captured through 9% of the high metallogenic potential areas, while the prediction efficiency of the artificial neural network model trained with fractal characterization is 9.3785, and 78.81% of the known ore points are successfully captured within 9% of the high metallogenic potential areas, reaching the highest prediction efficiency, which is significantly higher than the training effect of the original dataset. Through Figure 4 (b) and (c), it can be seen that in terms of accuracy and kappa coefficient, the artificial neural network model trained with the fractal characterization dataset is also higher than the artificial neural network model trained with the original dataset. It intuitively reflects the superiority of the fractal characterization project in improving exploration efficiency and model classification accuracy.

[0146] It should be noted that the embodiments of the present invention can be implemented through hardware, software, or a combination of software and hardware. The hardware part can be implemented using dedicated logic; the software part can be stored in a memory and executed by an appropriate instruction execution system, such as a microprocessor or dedicated designed hardware. Those of ordinary skill in the art can understand that the above devices and methods can be implemented using computer-executable instructions and / or included in processor control code, for example, such code is provided on a carrier medium such as a disk, CD, or DVD-ROM, a programmable memory such as a read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. The devices and modules of the present invention can be implemented by hardware circuits of programmable hardware devices such as very large scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, etc., or field programmable gate arrays, programmable logic devices, etc., can also be implemented by software executed by various types of processors, or can be implemented by a combination of the above hardware circuits and software, such as firmware.

[0147] The above is only the specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any modification, equivalent replacement, and improvement made within the spirit and principle of the present invention by those skilled in the art in the technical scope disclosed by the present invention should be covered by the protection scope of the present invention.

Claims

1. A metallogenic prediction method integrating fractal characterization engineering, characterized in that, The following steps are involved: S1, obtain the mineralization prediction data set and perform preprocessing; Obtain a mineralization prediction dataset; the dataset contains labels, sample point coordinates, and sample features, covering various favorable mineralization factors; S2, fractal analysis and feature selection of the data set; Construct a mineralization prediction framework integrating fractal characterization engineering; S3, model training and optimization; After completing fractal analysis and feature selection, select the prediction model, train the model, and find the optimal parameter combination to further improve the model's adaptability and prediction accuracy; S4, model predictions; The trained and optimized mineralization prediction model can be used to predict the mineralization probability in the study area and assist in delineating the prospecting target area; S5, result output and evaluation; The mineralization prediction data set is obtained and preprocessed: (a) Denoising: Geological features are inherently complex and noisy, so the collected data often contain some noisy data, and it is necessary to process missing values ​​and duplicates in the data; (b) Data normalization: To ensure the subsequent prediction effect, the data is normalized; The fractal analysis and feature selection of the data set are as follows: (1) Fractal and multifractal analysis: The box dimension method is used to describe the fractal characteristics of the mineral points, and then the Fry analysis is used to interpret the fractal results to enhance the subtle morphology of the mineral occurrence; the spatial correlation between these occurrences and geological characteristics is characterized to mine the mineralization distribution pattern; the sliding window technology is used to process the evidence layer of the original features to assist in capturing the local features within the massive prediction units, and multifractal analysis is performed to obtain the fractal representation of the original features; (2) Feature selection: Based on fractal processing, a variety of feature selection methods are used for analysis. The analysis results of various methods are combined to obtain the best combination of fractal indicator features and construct a fractal index feature data set for model prediction. The specific feature selection methods used are: prediction-area chart, K-means clustering, information gain, chi-square test and Pearson correlation coefficient.

2. The metallogenic prediction method integrating fractal characterization engineering according to claim 1, wherein The model training and optimization: (a) Select prediction model: With the rapid development of artificial intelligence algorithms, many excellent artificial intelligence algorithms have emerged. The models commonly used for mineralization prediction include artificial neural networks, random forests, support vector machines, and convolutional neural networks; (b) Model training and optimization: The optimal parameter combination is found by using the random grid search method and the ten-fold cross-validation method to further improve the prediction ability of the model. In addition, in order to minimize the impact of randomness, the data set should be randomly selected multiple times to form different training sets for separate training, and the final average value is taken as the result.

3. The metallogenic prediction method integrating fractal characterization engineering according to claim 1, wherein The model predicts: (a) Processing the prediction data set: The prediction data set should contain feature columns, coordinates and mineral point label information; by preprocessing the prediction data, the prediction data set is fractally characterized; (b) Input prediction data: Input the processed data to be predicted into the prediction model trained with different training sets, and make predictions respectively. The model will make predictions based on the sample characteristics, and the final prediction results will be averaged; (c) Quantitative analysis: The model outputs the mineralization probability of each sample point. By combining the success rate curve and the uncertainty scatter plot, high-potential mineralization areas and low-risk exploration areas are screened out.

4. The metallogenic prediction method integrating fractal characterization engineering according to claim 1, wherein The result output and evaluation are as follows: (a) Drawing the mineral resource potential evaluation map: Combining the prediction results of the model with the delineated high-potential mineralization areas and low-risk exploration areas, the mineral resource potential evaluation map is obtained, and the prospecting target areas are delineated. (b) Result evaluation: The evaluation indicators and feature importance of the model are output. Combining the existing exploration experience and mineralization research, the prediction results of the model are evaluated to facilitate the subsequent geological research and mineral exploration work.

5. A metallogenic prediction system for a fusion fractal characterization engineering implementing the metallogenic prediction method of the fusion fractal characterization engineering as described in any one of claims 1-4, characterized in that, (4) The mineralization prediction system integrating the fractal characterization project includes: A preprocessing module, which is used to obtain and preprocess the mineralization prediction data set; obtain the mineralization prediction data set; this data set contains labels, sample point coordinates and sample features, covering various favorable factors for mineralization. A feature selection module, which is used to perform fractal analysis and feature selection on the data set; construct a mineralization prediction framework integrating the fractal characterization project. An optimization module, which is used for model training and optimization; after completing the fractal analysis and feature selection, a prediction model is selected, the model is trained, and the optimal parameter combination is found to further improve the adaptability and prediction accuracy of the model. A prediction module, which is used for model prediction; the trained and optimized mineralization prediction model can be used to predict the mineralization probability in the study area and assist in delineating the prospecting target areas. An evaluation module, which is used for result output and evaluation.

6. A computer device, characterized in that, The computer device includes a memory and a processor. When the computer program stored in the memory is executed by the processor, the processor executes the steps of the mineralization prediction method of the fractal characterization project integration according to any one of claims 1-4.

7. A computer-readable storage medium stores a computer program. When the computer program is executed by a processor, the processor executes the steps of the mineralization prediction method of the fractal characterization project integration according to any one of claims 1-4.

8. An information data processing terminal, characterized in that, The information data processing terminal is used to implement the mineralization prediction system of the fractal characterization project integration according to claim 5.

Images