Mineral resource space intelligent prediction method and device based on multi-model ensemble learning
Through a multi-model integrated learning framework and convolutional neural network feature extraction, combined with multiple training and model evaluation, the uncertainty and complexity problems in mineral resource prediction are solved, and highly accurate and stable intelligent prediction is achieved.
Patent Information
- Application Number
- CN202510852187.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-24
- Publication Date
- 2025-10-03
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Existing mineral resource prediction methods rely on a single model and are unable to effectively process multi-source heterogeneous geological data, resulting in high uncertainty in prediction results. Traditional methods are also unable to cope with complexity and uncertainty problems.
A multi-model ensemble learning method is adopted, and the Stacking multi-model ensemble learning framework is used to combine RF, SVM, ANN primary learners and LR secondary learners. Feature extraction and multiple training are performed through convolutional neural networks, and model evaluation is performed using indicators such as ROC curves.
It improves the accuracy and stability of mineral resource prediction, reduces information redundancy, realizes automated and intelligent prediction process, and enhances the generalization ability of the model.
Smart Images
Figure CN120744868A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of mineral resource prediction, and in particular to a method and device for spatial intelligent prediction of mineral resources based on multi-model ensemble learning. Background Art
[0002] In the field of mineral resource prediction, traditional prediction methods often rely on a single model or algorithm, including spatial analysis methods based on GIS (geographic information system), geostatistical methods, etc.; however, these methods have many limitations in practical applications; for example, although GIS technology has powerful spatial information processing and analysis capabilities, when processing multi-source, heterogeneous geological data, it is often difficult to fully explore the deep-level, weak feature information in the data, resulting in high uncertainty in the prediction results; at the same time, traditional single-model prediction methods are often difficult to cope with the complexity and uncertainty problems faced in the mineral resource prediction process, and improper model selection may lead to deviations in the prediction results.
[0003] In recent years, with the rapid development of artificial intelligence and machine learning technologies, ensemble learning methods have gradually demonstrated their unique advantages in the field of mineral resource prediction. Ensemble learning combines multiple different classification or regression models and derives the final result through unified decision-making, which can overcome the shortcomings of each individual model and improve the accuracy and stability of prediction. However, the application of existing ensemble learning methods in the field of mineral resource prediction is still in its early stages, and a complete and creative set of prediction methods and devices has not yet been formed.
[0004] To this end, those skilled in the art have proposed a method and device for spatial intelligent prediction of mineral resources based on multi-model ensemble learning to solve the problems raised by the background technology. Summary of the Invention
[0005] In order to solve the above technical problems, the present invention provides a method and device for spatial intelligent prediction of mineral resources based on multi-model ensemble learning to solve the problems in the prior art.
[0006] The spatial intelligent prediction method of mineral resources based on multi-model ensemble learning includes:
[0007] S1. Acquire a feature dataset of a geological space to be predicted, wherein the feature dataset of the geological space to be predicted includes: a feature sub-dataset based on a knowledge perspective and a feature sub-dataset based on a data science perspective;
[0008] S2. Using the trained convolutional neural network to extract features from the feature dataset to obtain a new feature dataset;
[0009] S3. Perform mineral resource prediction based on the new feature data set and the trained Stacking multi-model integrated learning space intelligent modeling to output the final mineral resource prediction value, wherein the Stacking multi-model integrated learning framework includes multiple primary learners and secondary learners, and the outputs of multiple primary learners are connected to the secondary learners to serve as the input of the secondary learners. The primary learners include RF (random forest) primary learners, SVM (support vector machine) primary learners and ANN (artificial neural network) primary learners, and the secondary learners include LR (logistic regression) secondary learners.
[0010] Preferably, in step S2, the formula used for extracting features from the feature data set using the trained convolutional neural network includes:
[0011]
[0012] Among them, O j Represents the output of the j-th feature map, I i Represents the input feature map or original data, W ij represents the convolution kernel weight from the i-th input feature map to the j-th output feature map, b j represents the bias term, f represents the activation function (including ReLU, Sigmoid, etc.), M j Represents the set of input feature maps.
[0013] Preferably, in step S3, the training process of the Stacking multi-model integrated learning spatial intelligent modeling includes:
[0014] Dividing the new feature dataset into a training dataset and a test dataset;
[0015] The training data set is divided into 5 folds and trained 5 times respectively to obtain the first prediction result corresponding to the training data set. At the same time, one-fifth of the samples are retained each time for testing when training each primary learner;
[0016] Using each trained primary learner to predict the test data set, wherein each primary learner corresponds to 5 prediction results, and an average of the 5 prediction results is determined as the second prediction result of the primary learner;
[0017] splicing the first prediction results and the second prediction results of each primary learner to obtain an output prediction result;
[0018] The secondary learner is trained according to the prediction results output by each primary learner and the actual value of the training sample to obtain a trained secondary learner.
[0019] Preferably, the secondary learner is trained according to the prediction results output by each primary learner and the actual value of the training sample to obtain a trained secondary learner, including:
[0020] Based on the quantitative relationship between the predicted output results of each primary learner and the actual value of the training sample, the trained secondary learner is obtained.
[0021] Preferably, the method further comprises:
[0022] The test data set is used to perform model evaluation on the trained Stacking multi-model integrated spatial intelligent modeling, wherein the indicators used for model evaluation include: calculating the overall accuracy, recall rate, F-measure, Kappa coefficient and Matthews correlation coefficient of each machine learning model and the integrated model, and drawing the ROC curve.
[0023] A device for intelligent spatial prediction of mineral resources based on multi-model ensemble learning, using the above-mentioned method for intelligent spatial prediction of mineral resources based on multi-model ensemble learning, comprises:
[0024] An acquisition module is used to acquire a feature data set of a geological space to be predicted, wherein the feature data set of the geological space to be predicted includes: a feature sub-data set based on a knowledge perspective and a feature sub-data set based on a data science perspective;
[0025] A feature extraction module is used to extract features from the feature data set using a trained convolutional neural network to obtain a new feature data set;
[0026] A prediction module is used to predict mineral resources based on the new feature data set and the trained Stacking multi-model integrated learning spatial intelligent modeling to output a final mineral resource prediction value, wherein the Stacking multi-model integrated learning framework includes multiple primary learners and secondary learners, the outputs of multiple primary learners are connected to the secondary learners to serve as the input of the secondary learners, the primary learners include RF primary learners, SVM primary learners and ANN primary learners, and the secondary learners include LR secondary learners.
[0027] A processor is configured to execute the above-mentioned mineral resource spatial intelligent prediction method based on multi-model ensemble learning.
[0028] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the above-mentioned mineral resource spatial intelligent prediction method based on multi-model ensemble learning.
[0029] Compared with the prior art, the present invention has the following beneficial effects:
[0030] 1. The present invention adopts the Stacking multi-model integrated learning framework, uses the outputs of multiple primary learners (RF, SVM, ANN) as the input of the secondary learner (LR), and replaces the individual decision-making strategy of a single model with a collective decision-making strategy, effectively reducing the uncertainty of the final prediction result and improving the accuracy of the prediction.
[0031] 2. The present invention uses a convolutional neural network to extract features from the characteristic data set of geological space, which not only reduces the problem of information redundancy, but also solves the problem of difficulty in extracting deep and weak characteristic information in the process of mineral resource prediction.
[0032] 3. The present invention realizes an automated and intelligent prediction process by constructing a mineral resource spatial intelligent prediction device based on multi-model integrated learning, greatly improving the prediction efficiency and reducing the impact of human intervention and subjective judgment on the prediction results.
[0033] 4. The present invention uses a training set and a validation set to evaluate the model. The former reflects the model's ability to fit the data, and the latter reflects the model's generalization ability. The ROC curve is used to evaluate the performance of the prediction model. The AUC value can effectively indicate the prediction ability of different mineralization prediction models, thereby enhancing the generalization ability of the model. BRIEF DESCRIPTION OF THE DRAWINGS
[0034] Figure 1 This is a flow chart of the method for spatial intelligent prediction of mineral resources based on multi-model ensemble learning of the present invention;
[0035] Figure 2 This is a framework diagram of the mineral resource spatial intelligent prediction device based on multi-model integrated learning of the present invention. DETAILED DESCRIPTION
[0036] The following embodiments of the present invention are described in further detail with reference to the accompanying drawings and examples. The following examples are used to illustrate the present invention but are not intended to limit the scope of the present invention.
[0037] Embodiment: The present invention provides a method for spatial intelligent prediction of mineral resources based on multi-model ensemble learning, such as Figure 1 Shown, including:
[0038] S1. Acquire a feature dataset of a geological space to be predicted, wherein the feature dataset of the geological space to be predicted includes: a feature sub-dataset based on a knowledge perspective and a feature sub-dataset based on a data science perspective;
[0039] S2. Using the trained convolutional neural network to extract features from the feature dataset to obtain a new feature dataset;
[0040] S3. Perform mineral resource prediction based on the new feature data set and the trained Stacking multi-model integrated learning space intelligent modeling to output the final mineral resource prediction value, wherein the Stacking multi-model integrated learning framework includes multiple primary learners and secondary learners, and the outputs of multiple primary learners are connected to the secondary learners to serve as the input of the secondary learners. The primary learners include RF (random forest) primary learners, SVM (support vector machine) primary learners and ANN (artificial neural network) primary learners, and the secondary learners include LR (logistic regression) secondary learners.
[0041] From the above, it can be seen that this method adopts the Stacking multi-model integrated learning framework and uses the outputs of multiple primary learners such as RF, SVM, and ANN as the input of the LR secondary learner, which effectively reduces the uncertainty of the final prediction results and improves the accuracy of the prediction. At the same time, the convolutional neural network is used to extract features from the characteristic data set of the geological space, which solves the problem that deep and weak feature information is difficult to extract. Through the automated and intelligent prediction device, the prediction efficiency is greatly improved, human intervention is reduced, and the generalization ability of the model is enhanced.
[0042] Furthermore, the formula used to extract features from the feature data set using the trained convolutional neural network includes:
[0043]
[0044] Among them, O j Represents the output of the j-th feature map, I i Represents the input feature map or original data, W ij represents the convolution kernel weight from the i-th input feature map to the j-th output feature map, b j represents the bias term, f represents the activation function (including ReLU, Sigmoid, etc.), M j Represents the set of input feature maps.
[0045] From the above, we can see that through precise mathematical operations and activation function processing, the key information in the input feature map can be efficiently extracted, and a feature map output with stronger representation ability can be generated. This method not only reduces information redundancy, but also solves the problem of deep and weak feature information being difficult to extract in the mineral resource prediction process, thereby improving the accuracy and effectiveness of feature extraction and providing a higher-quality feature data foundation for subsequent multi-model integrated learning.
[0046] Furthermore, in step S3, the training process of the Stacking multi-model integrated learning spatial intelligent modeling includes:
[0047] Dividing the new feature dataset into a training dataset and a test dataset;
[0048] The training data set is divided into 5 folds and trained 5 times respectively to obtain the first prediction result corresponding to the training data set. At the same time, one-fifth of the samples are retained each time for testing when training each primary learner;
[0049] Using each trained primary learner to predict the test data set, wherein each primary learner corresponds to 5 prediction results, and an average of the 5 prediction results is determined as the second prediction result of the primary learner;
[0050] splicing the first prediction results and the second prediction results of each primary learner to obtain an output prediction result;
[0051] The secondary learner is trained according to the prediction results output by each primary learner and the actual value of the training sample to obtain a trained secondary learner.
[0052] As can be seen from the above, through fine data division and multiple training, the training data set and test data set are effectively utilized; this method not only ensures the stability and accuracy of the primary learner through five-fold cross-validation, but also further enhances the reliability of the prediction results by calculating the average prediction results of the primary learner on the test data set; the prediction results of each primary learner are spliced together, and rich input information is provided to the secondary learner, enabling it to learn a more comprehensive and accurate prediction model. This training process greatly improves the prediction performance of Stacking multi-model ensemble learning, making the final mineral resource prediction results more accurate and stable.
[0053] Furthermore, the secondary learner is trained according to the prediction results output by each primary learner and the actual value of the training sample to obtain a trained secondary learner, including:
[0054] Based on the quantitative relationship between the predicted output results of each primary learner and the actual value of the training sample, the trained secondary learner is obtained, and the formula is:
[0055] Y=β1X1+β2X2+β3X3+α,
[0056] Among them, β1, β2, and β3 represent regression coefficients, X1 represents the prediction result output by the RF primary learner, X2 represents the prediction result output by the SVM primary learner, X3 represents the prediction result output by the ANN primary learner, Y represents the final mineral resource prediction value obtained by the Stacking multi-model integrated learning space intelligent modeling, and α represents the regression constant.
[0057] From the above, we can see that by using a specific formula to train the secondary learner (LR) based on the quantitative relationship between the prediction results output by each primary learner (RF, SVM, ANN) and the actual value of the training sample, the intrinsic connection between the prediction results of the primary learner and the actual value can be accurately captured; this method ensures that the secondary learner can learn the optimal weight combination, that is, the regression coefficient, so as to more accurately integrate the output of the primary learner to obtain the trained secondary learner; it not only improves the prediction accuracy of the secondary learner, but also enhances the stability and generalization ability of the entire Stacking multi-model integrated learning system, making the final mineral resource prediction results more reliable and accurate.
[0058] Furthermore, the method further comprises:
[0059] The test data set is used to perform model evaluation on the trained Stacking multi-model integrated spatial intelligent modeling, wherein the indicators used for model evaluation include: calculating the overall accuracy, recall rate, F-measure, Kappa coefficient and Matthews correlation coefficient of each machine learning model and the integrated model, and drawing the ROC curve.
[0060] From the above, we can see that by using the test data set to conduct a comprehensive model evaluation of the trained Stacking multi-model integrated spatial intelligent modeling, covering multiple key indicators such as overall accuracy, recall rate, F-measure, Kappa coefficient and Matthews correlation coefficient, and drawing an ROC curve to intuitively display the performance of the model; this multi-dimensional evaluation method not only comprehensively reflects the predictive ability of the model, but also can accurately identify the performance of the model in different prediction scenarios; it not only helps to optimize model parameters and improve prediction accuracy, but also can enhance the interpretability and credibility of the model, providing solid technical support for the scientific prediction and rational development of mineral resources.
[0061] Furthermore, the results of the mineral resource spatial intelligent prediction method based on multi-model ensemble learning of the embodiment are compared with the current traditional prediction methods (traditional GIS spatial analysis method, geostatistical method), and the following table is obtained:
[0062]
[0063] As can be seen from the above table, compared with traditional GIS spatial analysis methods and geostatistical methods, the mineral resource spatial intelligent prediction method based on multi-model ensemble learning shows significant advantages in key indicators such as overall accuracy, recall rate, F-measure, Kappa coefficient and Matthews correlation coefficient; this proves the effectiveness and accuracy of this method in the field of mineral resource prediction, and provides strong technical support for the scientific prediction and rational development of mineral resources.
[0064] Mineral resource spatial intelligent prediction device based on multi-model ensemble learning, such as Figure 2 As shown, using the above-mentioned mineral resource spatial intelligent prediction method based on multi-model ensemble learning, the device includes:
[0065] An acquisition module is used to acquire a feature data set of a geological space to be predicted, wherein the feature data set of the geological space to be predicted includes: a feature sub-data set based on a knowledge perspective and a feature sub-data set based on a data science perspective;
[0066] A feature extraction module is used to extract features from the feature data set using a trained convolutional neural network to obtain a new feature data set;
[0067] A prediction module is used to predict mineral resources based on the new feature data set and the trained Stacking multi-model integrated learning spatial intelligent modeling to output a final mineral resource prediction value, wherein the Stacking multi-model integrated learning framework includes multiple primary learners and secondary learners, the outputs of multiple primary learners are connected to the secondary learners to serve as the input of the secondary learners, the primary learners include RF primary learners, SVM primary learners and ANN primary learners, and the secondary learners include LR secondary learners.
[0068] Working principle: First, a feature dataset of the geological space to be predicted is obtained, which integrates features based on knowledge perspective and data science perspective; then, the trained convolutional neural network is used to perform deep feature extraction on the feature dataset to obtain a more representative new feature dataset; then, the Stacking multi-model integrated learning framework is adopted, and the outputs of multiple primary learners (including random forest, support vector machine, artificial neural network) are used as the input of secondary learners (including logistic regression). Through multiple training and cross-validation, the parameters of the primary and secondary learners are continuously optimized; finally, the trained model is evaluated using a test dataset. By calculating multiple performance indicators (including overall accuracy, recall rate, etc.) and drawing ROC curves, the predictive ability and generalization of the model are fully verified, thereby realizing intelligent spatial prediction of mineral resources.
[0069] The embodiment of the present application provides an electronic device applicable to the above-mentioned method for intelligent spatial prediction of mineral resources based on multi-model ensemble learning, including:
[0070] Memory, used to protect computer programs and data;
[0071] Processor, used to run system programs.
[0072] An embodiment of the present application provides a computer storage medium, which is suitable for the above-mentioned mineral resource spatial intelligent prediction method based on multi-model integrated learning, and performs hierarchical confidentiality management on the above-mentioned system and data in accordance with confidentiality management requirements.
[0073] Those skilled in the art will appreciate that the embodiments of the present application can be provided as a system or a computer program product. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0074] The present application is described with reference to the flowcharts and / or block diagrams of the devices (systems) and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0075] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.
[0076] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.
[0077] In a typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.
[0078] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.
[0079] Computer-readable media includes permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology for information storage. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices, or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include transitory computer-readable media such as modulated data signals and carrier waves.
[0080] It should also be noted that the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, commodity, or apparatus comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, commodity, or apparatus. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, commodity, or apparatus comprising the element.
[0081] The embodiments of the present invention are provided for the purpose of illustration and description. Although the embodiments of the present invention have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limitations on the present invention. Ordinary technicians in this field can change, modify, replace and modify the above embodiments within the scope of the present invention.
Claims
1. A method for spatial intelligent prediction of mineral resources based on multi-model ensemble learning, characterized by: include: S1. Acquire a feature dataset of a geological space to be predicted, wherein the feature dataset of the geological space to be predicted includes: a feature sub-dataset based on a knowledge perspective and a feature sub-dataset based on a data science perspective; S2. Using the trained convolutional neural network to extract features from the feature dataset to obtain a new feature dataset; S3. Mineral resource prediction is performed based on the new feature data set and the trained Stacking multi-model integrated learning spatial intelligent modeling, wherein the Stacking multi-model integrated learning framework includes multiple primary learners and secondary learners, and the outputs of the multiple primary learners are connected to the secondary learners.
2. The method and device for spatial intelligent prediction of mineral resources based on multi-model ensemble learning according to claim 1, characterized in that: In step S2, the formula used for extracting features from the feature data set using the trained convolutional neural network includes: Among them, O j Represents the output of the j-th feature map, I i Represents the input feature map or original data, W ij represents the convolution kernel weight from the i-th input feature map to the j-th output feature map, b j represents the bias term, f represents the activation function, M j Represents the set of input feature maps.
3. The method for spatial intelligent prediction of mineral resources based on multi-model ensemble learning according to claim 1, characterized in that: In step S3, the training process of the stacking multi-model integrated learning spatial intelligent modeling includes: Dividing the new feature dataset into a training dataset and a test dataset; The training data set is divided into 5 folds and trained 5 times respectively to obtain the first prediction result corresponding to the training data set. At the same time, one-fifth of the samples are retained each time for testing when training each primary learner; Using each trained primary learner to predict the test data set, wherein each primary learner corresponds to 5 prediction results, and an average of the 5 prediction results is determined as the second prediction result of the primary learner; splicing the first prediction results and the second prediction results of each primary learner; The secondary learners are trained according to the prediction results output by each primary learner and the actual values of the training samples.
4. The method for spatial intelligent prediction of mineral resources based on multi-model ensemble learning as claimed in claim 3 is characterized by: The secondary learner is trained according to the prediction results output by each primary learner and the actual value of the training sample, including: Based on the quantitative relationship between the predicted output results of each primary learner and the actual value of the training sample, the trained secondary learner is obtained.
5. The method for spatial intelligent prediction of mineral resources based on multi-model ensemble learning as claimed in claim 3, characterized in that: The method further comprises: The test data set is used to perform model evaluation on the trained Stacking multi-model integrated spatial intelligent modeling, wherein the indicators used for model evaluation include: calculating the overall accuracy, recall rate, F-measure, Kappa coefficient and Matthews correlation coefficient of each machine learning model and the integrated model.
6. A spatial intelligent prediction device for mineral resources based on multi-model ensemble learning, characterized by: Using the method for spatial intelligent prediction of mineral resources based on multi-model ensemble learning according to any one of claims 1 to 5, the device comprises: An acquisition module is used to acquire a feature data set of a geological space to be predicted, wherein the feature data set of the geological space to be predicted includes: a feature sub-data set based on a knowledge perspective and a feature sub-data set based on a data science perspective; A feature extraction module is used to extract features from the feature data set using a trained convolutional neural network to obtain a new feature data set; A prediction module is used to predict mineral resources based on the new feature data set and the trained Stacking multi-model integrated learning spatial intelligent modeling to output a final mineral resource prediction value, wherein the Stacking multi-model integrated learning framework includes multiple primary learners and secondary learners, and the outputs of the multiple primary learners are connected to the secondary learners.
7. A processor, characterized in that: It is configured to execute the mineral resource spatial intelligent prediction method based on multi-model ensemble learning according to any one of claims 1 to 5.
8. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed by a processor, the method for spatial intelligent prediction of mineral resources based on multi-model integrated learning as described in any one of claims 1 to 5 is implemented.