Mineral resource distribution prediction method and system based on machine learning

Abstract

The invention discloses a mineral resource distribution prediction method and system based on machine learning, and the method comprises the steps: taking data obtained by a preset mineral resource database as to-be-analyzed data; extracting exploration features of the to-be-analyzed data, performing feature comparison analysis on the exploration features and the reference database to obtain a matching degree, outputting the exploration features with the matching degree higher than a matching degree threshold as first data, and otherwise, outputting the exploration features with the matching degree higher than the matching degree threshold as fuzzy data; calculating the fuzzy data and the reference data in a matching result to carry out uncertainty evaluation, and outputting the fuzzy data of which the dissimilarity is lower than a dissimilarity threshold value as second data; adding the second data into the first data to optimize the matching degree; and constructing a mineral resource distribution prediction model according to the first data, the matching degree and the environmental data, adding the second data to the reference database to obtain an updated reference database, and outputting a target prediction model.

Description

Technical Field

[0001] The present invention relates to the field of prediction, and particularly to a method and system for predicting the distribution of mineral resources based on machine learning. Background Art

[0002] The prediction of the distribution of mineral resources is a core link in geological exploration and resource development. Traditional methods mainly rely on expert experience or static statistical models, and have the following limitations: existing methods usually directly fuse multi-source heterogeneous data, but do not distinguish the certainty levels of the data, and the differential application of the reference database and the to-be-determined database is insufficient, resulting in low-quality data interfering with the modeling accuracy; the traditional deterministic database is updated laggingly and cannot timely incorporate new observation data generated during the exploration process, restricting the adaptability of the model to the dynamic geological environment; the processing of low-matching data is simply discarded without quantifying its difference characteristics from the reference data, causing waste of potential effective information; the mineral distribution is affected by the combined influence of geological structures and environmental factors, but existing models mostly focus on single geological features and lack collaborative modeling of multi-dimensional environmental data.

[0003] In recent years, machine learning technology has shown advantages in feature fusion and uncertainty modeling, but its application in mineral resource prediction still faces the following challenges: how to construct a dynamically updated reference database to balance the contributions of historical highly credible data and newly added low-credible data; how to design feature comparison and dissimilarity quantification rules to improve the reusability of fuzzy data; how to couple environmental data with geological features to construct a spatio-temporally sensitive distribution probability model. Therefore, there is an urgent need for a prediction method that can dynamically integrate multi-source data, hierarchically screen effective features, and quantify uncertainties. Summary of the Invention

[0004] The object of the present invention is to provide a method for predicting the distribution of mineral resources based on machine learning.

[0005] To achieve the above object, the present invention is implemented according to the following technical solutions:

[0006] The present invention includes the following steps:

[0007] Taking the data obtained from a preset mineral resource database within a specified time period as the data to be analyzed; wherein, the mineral resource database includes a reference database and a to-be-determined database; the reference database is a database with a higher certainty level than the to-be-determined database; the data includes multiple data types, and different data types are used to characterize the mineral distribution from different types; the data to be analyzed includes observation data and environmental data; the observation data is real-time exploration data of a preset area; the observation data includes geological data, geophysical data, remote sensing data, mineral exploration data, climate data, vegetation data, and hydrological data;

[0008] Extract the exploration features of the data to be analyzed, perform feature comparison and analysis on the exploration features and the reference database to obtain a matching degree, and output the exploration features with a matching degree higher than the matching degree threshold as the first data, and vice versa as fuzzy data;

[0009] Calculate the dissimilarity between the fuzzy data and the reference data in the matching result, perform uncertainty evaluation based on the dissimilarity, and output the fuzzy data with a dissimilarity lower than the dissimilarity threshold as the second data; Add the second data to the first data to optimize the matching degree;

[0010] Perform distribution probability prediction based on the first data, the matching degree, and the environmental data, construct a mineral resource distribution prediction model based on the distribution probability, add the second data to the reference database to obtain an updated reference database, and output the target prediction model.

[0011] Further, the data types include pictures, videos, and texts.

[0012] Further, the dissimilarity includes:

[0013] Extract the feature vectors of the reference data, and calculate the dissimilarity between the exploration features in the fuzzy data and the reference data in the matching result:

[0014]

[0015]

[0016] Among them, the a-th exploration feature vector of the k-th dimension at the key quantile p is The c-th reference data feature vector of the k-th dimension at the key quantile p is The dissimilarity between the fuzzy data and the reference data is The number of dimensions is N k , and the weight of the k-th dimension is The normalization factor of the k-th dimension is ρ, the bandwidth parameter is U, and the spatial distance between the environmental factor of the k-th dimension and the target area is d k , and the key quantile is p.

[0017] Further, the method also includes threshold adjustment for the matching degree threshold and the dissimilarity threshold:

[0018] Construct a graph attention network A based on the matching degree between the fuzzy data and the reference data, and perform confidence evaluation on the matching degree. The expression is:

[0019]

[0020] The matching degree between the a-th exploration feature vector and the c-th reference data feature vector is The confidence level of the matching degree is The amplitude modulation coefficient of the matching degree is β, and the number of exploration feature vectors is N u , and the attention weight between the a-th exploration feature vector and the c-th reference data feature vector is

[0021] Taking the exploration feature vector and the reference data feature vector as nodes, a graph attention network B is constructed according to the dissimilarity, and the confidence level of the dissimilarity is evaluated. The expression is:

[0022]

[0023] Among them, the dissimilarity between the a-th fuzzy data and the c-th reference data is The confidence level of the dissimilarity is The amplitude modulation coefficient of the dissimilarity is The feature distribution of the a-th node is E a , and the feature distribution of the c-th node is E c , the maximum value of the divergence is HX max , the feature distribution E a and the feature distribution E c The divergence of is HX(E a ‖E c );

[0024] Dynamically adjust the matching degree threshold and the dissimilarity threshold according to the confidence level. The expression is:

[0025]

[0026] Among them, the i-th target threshold after dynamic adjustment is D i , and the initial value of the i-th target threshold is The feature distribution density corresponding to the target threshold is V, and the average confidence level of the i-th target threshold is The adaptive adjustment function of the feature distribution density V and the average confidence level is

[0027] Furthermore, the method also includes updating the distribution probability:

[0028] Build a deep reinforcement neural network learning based on the exploration features of the first data, and give a probability update strategy. The expression is:

[0029]

[0030] Among them, the j-th probability update strategy is T j, the index variable is j, and the discount factor is ζ; update the distribution probability according to the probability update strategy.

[0031] Furthermore, the method for obtaining the distribution probability includes:

[0032] Taking the environmental data as the background, obtain the influence degree of the environmental data on the mine distribution through the time-varying influence algorithm; according to the first data dynamically adjusted by the influence degree, recalculate the matching degree through the adjusted first data, obtain the reference data with a matching degree higher than the matching threshold, calculate the regional matching degree of the reference data, take the region with the highest regional matching degree as the regional prediction result, and if the matching degree error is lower than 0.12, perform multimodal fusion on the region to obtain the regional prediction result; splice the regional prediction results and output them as the distribution probability.

[0033] Furthermore, the method for constructing a mineral resource distribution prediction model based on the distribution probability and environmental data includes:

[0034] Take the weighted sum of the variance and loss function of the distribution probability predicted value and the actual value of the distribution probability as the objective function of the mineral resource distribution prediction model;

[0035] The mineral resource distribution prediction model includes a random forest algorithm, a support vector machine algorithm, and a deep learning algorithm;

[0036] The random forest algorithm divides the input data into training data and test data according to 7:2;

[0037] The support vector machine algorithm separates the training data of different classes by finding an optimal hyperplane, and continuously iterates until the interval between the two classes is maximized to obtain the classification data;

[0038] The deep learning algorithm trains the classification data, automatically learns and extracts the features and patterns related to the existence of mineral resources, and takes the objective function as the optimization guidance to predict the distribution of mineral resources in unknown areas;

[0039] The test data is input into the mineral resource distribution prediction model to obtain the predicted mineral resource distribution, and the learning rate is continuously adjusted according to the mean square error between the actual mineral resource distribution and the predicted mineral resource distribution until the mean square error reaches the minimum.

[0040] In a second aspect, a mineral resource distribution prediction system based on machine learning includes:

[0041] Data acquisition module: It is used to take the data obtained from the preset mineral resource database within a specified time period as the data to be analyzed. Among them, the mineral resource database includes a reference database and a pending database. The reference database is a database with a higher certainty level than the pending database. The data contains multiple data types, and different data types are used to characterize the mineral distribution from different types. The data to be analyzed includes observation data and environmental data. The observation data is real-time exploration data of a preset area. The observation data includes geological data, geophysical data, remote sensing data, mineral exploration data, climate data, vegetation data, and hydrological data.

[0042] Match degree comparison and analysis module: It is used to extract the exploration features of the data to be analyzed, perform feature comparison and analysis on the exploration features and the reference database to obtain the match degree, and output the exploration features with a match degree higher than the match degree threshold as the first data, and vice versa as fuzzy data.

[0043] Uncertainty evaluation and optimization module: It is used to calculate the dissimilarity between the fuzzy data and the reference data in the matching result, perform uncertainty evaluation based on the dissimilarity, and output the fuzzy data with a dissimilarity lower than the dissimilarity threshold as the second data; add the second data to the first data to optimize the match degree.

[0044] Model construction module: It is used to perform distribution probability prediction based on the first data, the match degree, and the environmental data, construct a mineral resource distribution prediction model according to the distribution probability, add the second data to the reference database to obtain an updated reference database, and output the target prediction model.

[0045] In a third aspect, an embodiment of the present application further provides an electronic device, including:

[0046] A processor; and a memory arranged to store computer-executable instructions, and the executable instructions, when executed, cause the processor to execute the method steps described in the first aspect.

[0047] In a fourth aspect, an embodiment of the present application further provides a computer-readable storage medium, and the computer-readable storage medium stores one or more programs, and when the one or more programs are executed by an electronic device including a plurality of application programs, the electronic device is caused to execute the method steps described in the first aspect.

[0048] The beneficial effects of the present invention are:

[0049] The present invention is a method and system for predicting the distribution of mineral resources based on machine learning. Compared with the prior art, the present invention has the following technical effects:

[0050] Through steps of preprocessing, obtaining first data and fuzzy data, obtaining dissimilarity, obtaining second data, updating distribution prediction, and model construction, the present invention can improve the accuracy of mineral resource distribution prediction based on machine learning, thereby enhancing the precision of mineral resource distribution prediction based on machine learning, optimizing the mineral resource distribution prediction based on machine learning, greatly saving resources, improving work efficiency, enabling intelligent prediction and identification of mineral resource distribution, correcting and updating data for mineral resource distribution prediction in real time, which is of great significance for mineral resource distribution prediction, and can adapt to different standards and requirements of mineral resource distribution prediction, having a certain universality. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 It is a flowchart of the steps of a method for predicting the distribution of mineral resources based on machine learning according to the present invention;

[0052] Figure 2 It is a schematic structural diagram of an electronic device in an embodiment of this specification. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0053] The present invention will be further described below through specific embodiments. The illustrative embodiments and explanations of this invention are used to explain the present invention, but do not limit the present invention.

[0054] A method and system for predicting the distribution of mineral resources based on machine learning according to the present invention include the following steps:

[0055] As Figure 1 shown, in this embodiment, it includes the following steps:

[0056] The data obtained from a preset mineral resource database within a specified time period is used as the data to be analyzed; wherein, the mineral resource database includes a reference database and a pending database; the reference database is a database with a higher certainty level than the pending database; the data to be analyzed includes multiple data types, and different data types are used to characterize mineral distribution from different types; the data to be analyzed includes observation data and environmental data; the observation data is real-time exploration data of a preset area; the observation data includes geological data, geophysical data, remote sensing data, mineral exploration data, climate data, vegetation data, and hydrological data;

[0057] In the actual evaluation, XXX Copper Mine is taken as the research object. The reference database stores a large amount of highly reliable data that has been verified over a long period, including detailed stratigraphic distribution and fault strike data in the target area over the past 30 years; the copper ore reserve is 5.36 million tons, and the grade reaches 2%; the to-be-determined database is the preliminary geological survey data of the newly discovered suspected mineralized area, and a weak metal anomaly signal has been preliminarily detected in the rock samples of the target area; the preliminary geological survey data of the newly discovered suspected mineralized area, and a weak metal anomaly signal has been preliminarily detected in the rock samples of the target area;

[0058] The observed data in the data to be analyzed are as follows: the lithologic distribution of the strata in a certain area obtained from the latest on-site surveying and mapping. The main strata at a depth of 500 - 1000 meters in the target area are interbedded granite and sandstone; the gravity anomaly data shows that the gravity anomaly range in a specific area is between -50 and +30 milligals; the remote sensing data shows that the surface vegetation coverage in the target area is about 40%, and there are linear texture features in some areas; at a drilling depth of 200 meters at a certain location, a small amount of lead-zinc mineralization signs have been detected in the collected samples; the average temperature in the target area in the past 10 years is 15°C, and the annual precipitation is 800 mm; the target area is mainly temperate broad-leaved forest and grassland, with coverage rates of 31% and 9.81% respectively; the flow rate of a certain river during the dry season is 5 cubic meters per second, and the flow rate during the flood season is 20 cubic meters per second;

[0059] The environmental data in the data to be analyzed are the topographic and geomorphic data of the selected area. It is simulated that the altitude in the target area is between 500 - 1500 meters, mainly consisting of mountains and hills; the data related to plate movement, and it is simulated that the target area is located at the plate edge, and the relative plate movement speed is 2 - 3 centimeters per year;

[0060] Extract the exploration features of the data to be analyzed, conduct a feature comparison and analysis of the exploration features and the reference database to obtain the matching degree, and output the exploration features with a matching degree higher than the matching degree threshold as the first data, and output the rest as fuzzy data;

[0061] In the actual evaluation, the exploration features are stratigraphic lithology, geological structure, surface linear texture, number of exploration drill holes, temperature, extreme weather events, vegetation coverage, tonal anomaly, precipitation, river flow rate, water level change, groundwater level and water quality, preliminary features;

[0062] The fuzzy data includes gravity anomaly, magnetic anomaly, underground electrical structure, location of mineralization points;

[0063] The matching degree threshold is 0.713; the matching degrees of the target area with Copper Mine C1, Copper Mine C3, and Copper Mine C7 are 0.715, 0.809, and 0.785 respectively;

[0064] Calculate the dissimilarity between the fuzzy data and the reference data in the matching results, perform uncertainty evaluation according to the dissimilarity, and output the fuzzy data with a dissimilarity lower than the dissimilarity threshold as the second data; add the second data to the first data to optimize the matching degree;

[0065] In the actual evaluation, the dissimilarity threshold is 0.257, and the second data is gravity anomaly and mineralization point location;

[0066] Perform distribution probability prediction according to the first data, the matching degree, and the environmental data, construct a mineral resource distribution prediction model according to the distribution probability, add the second data to the reference database to obtain an updated reference database, and output the target prediction model;

[0067] In this embodiment, the data types include pictures, videos, and texts.

[0068] In this embodiment, the dissimilarity includes:

[0069] Extract the feature vectors of the reference data, and calculate the dissimilarity between the exploration features in the fuzzy data and the reference data in the matching results:

[0070]

[0071] Where the a-th exploration feature vector at the k-th dimension at the key quantile p is The c-th reference data feature vector at the k-th dimension at the key quantile p is The dissimilarity between the fuzzy data and the reference data is The number of dimensions is N k , and the weight of the k-th dimension is The normalization factor of the k-th dimension is ρ, the bandwidth parameter is U, and the spatial distance between the environmental factor of the k-th dimension and the target area is d k , and the key quantile is p;

[0072] In the actual evaluation, the key quantiles are 0.12, 0.49, and 0.93 respectively.

[0073] In this embodiment, the method further includes threshold adjustment for the matching degree threshold and the dissimilarity threshold:

[0074] Construct a graph attention network A according to the matching degree between the fuzzy data and the reference data, and perform confidence evaluation on the matching degree. The expression is:

[0075]

[0076] Where the matching degree between the a-th exploration feature vector and the c-th reference data feature vector is The confidence level of the matching degree is The amplitude modulation coefficient of the matching degree is β, and the number of exploration feature vectors is N u , and the attention weight between the a-th exploration feature vector and the c-th reference data feature vector is

[0077] Taking the exploration feature vector and the reference data feature vector as nodes, construct a graph attention network B according to the dissimilarity, and evaluate the confidence level of the dissimilarity. The expression is:

[0078]

[0079] Among them, the dissimilarity between the a-th fuzzy data and the c-th reference data is The confidence level of the dissimilarity is The amplitude modulation coefficient of the dissimilarity is The feature distribution of the a-th node is E a , and the feature distribution of the c-th node is E c , the maximum value of the divergence is HX max , the feature distribution E a and the feature distribution E c The divergence of is HX(E a ‖E c );

[0080] Dynamically adjust the matching degree threshold and the dissimilarity threshold according to the confidence level. The expression is:

[0081]

[0082] Among them, the i-th target threshold after dynamic adjustment is D i , and the initial value of the i-th target threshold is The feature distribution density corresponding to the target threshold is V, and the average confidence level of the i-th target threshold is The adaptive adjustment function of the feature distribution density V and the average confidence level is

[0083] In this embodiment, the method further includes updating the distribution probability:

[0084] Build a deep reinforcement neural network learning based on the exploration features of the first data, and give a probability update strategy. The expression is:

[0085]

[0086] Among them, the j-th probability update strategy is T j , the index variable is j, and the discount factor is ζ; update the distribution probability according to the probability update strategy.

[0087] In this embodiment, the method for obtaining the distribution probability includes:

[0088] Taking the environmental data as the background, obtaining the influence degree of the environmental data on the mine distribution through a time-varying influence algorithm; according to the first data dynamically adjusted by the influence degree, recalculating the matching degree through the adjusted first data, obtaining reference data with a matching degree higher than the matching threshold, calculating the regional matching degree of the reference data, taking the region with the highest regional matching degree as the regional prediction result, and if the matching degree error is lower than 0.12, performing multi-modal fusion on the region to obtain the regional prediction result; splicing the regional prediction results and outputting them as the distribution probability.

[0089] In this embodiment, the method for constructing a mineral resource distribution prediction model according to the distribution probability and environmental data includes:

[0090] Taking the weighted sum of the variance and loss function of the distribution probability predicted value and the actual value of the distribution probability as the objective function of the mineral resource distribution prediction model;

[0091] The mineral resource distribution prediction model includes a random forest algorithm, a support vector machine algorithm, and a deep learning algorithm;

[0092] The random forest algorithm divides the input data into training data and test data according to 7:2;

[0093] The support vector machine algorithm separates the training data of different classes by finding an optimal hyperplane, and continuously iterates until the interval between the two classes is maximized to obtain classification data;

[0094] The deep learning algorithm trains the classification data, automatically learns and extracts features and patterns related to the existence of mineral resources, and takes the objective function as the optimization guide to predict the distribution of mineral resources in unknown regions;

[0095] The test data is input into the mineral resource distribution prediction model to obtain the predicted mineral resource distribution, and the learning rate is continuously adjusted until the mean square error reaches the minimum according to the mean square error between the actual mineral resource distribution and the predicted mineral resource distribution;

[0096] In actual evaluation, the learning rate is 0.001.

[0097] In a second aspect, a mineral resource distribution prediction system based on machine learning includes:

[0098] Data acquisition module: It is used to take the data obtained from the preset mineral resource database within a specified time period as the data to be analyzed. Among them, the mineral resource database includes a reference database and a pending database; the reference database is a database with a higher certainty level than the pending database; the data contains multiple data types, and different data types are used to characterize the mineral distribution from different types; the data to be analyzed includes observation data and environmental data; the observation data is real-time exploration data of a preset area; the observation data includes geological data, geophysical data, remote sensing data, mineral exploration data, climate data, vegetation data, and hydrological data.

[0099] Match degree comparison and analysis module: It is used to extract the exploration features of the data to be analyzed, conduct feature comparison and analysis on the exploration features and the reference database to obtain the match degree, and output the exploration features with a match degree higher than the match degree threshold as the first data, and vice versa as fuzzy data.

[0100] Uncertainty evaluation and optimization module: It is used to calculate the dissimilarity between the fuzzy data and the reference data in the matching result, conduct uncertainty evaluation according to the dissimilarity, and output the fuzzy data with a dissimilarity lower than the dissimilarity threshold as the second data; add the second data to the first data to optimize the match degree.

[0101] Model construction module: It is used to conduct distribution probability prediction according to the first data, the match degree, and the environmental data, construct a mineral resource distribution prediction model according to the distribution probability, add the second data to the reference database to obtain an updated reference database, and output the target prediction model.

[0102] Figure 2 It is a schematic structural diagram of an electronic device according to an embodiment of the present application. Please refer to Figure 2 , at the hardware level, the electronic device includes a processor, and optionally also includes an internal bus, a network interface, and a memory. Among them, the memory may include internal memory, such as high-speed random access memory (Random-Access Memory, RAM), and may also include non-volatile memory, such as at least one disk memory, etc. Of course, the electronic device may also include other hardware required for other services.

[0103] The processor, network interface, and memory can be interconnected through an internal bus, which can be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus, an EISA (Extended Industry Standard Architecture) bus, or the like. The bus can be divided into an address bus, a data bus, a control bus, etc. For the sake of convenience of representation, Figure 2 only a bidirectional arrow is used in the figure, but it does not mean that there is only one bus or one type of bus.

[0104] Memory, which is used to store programs. Specifically, the program can include program code, and the program code includes computer operation instructions. The memory can include a memory and a non-volatile memory, and provides instructions and data to the processor.

[0105] The processor reads the corresponding computer program from the non-volatile memory into the memory and then runs it, forming a machine learning-based mineral resource distribution prediction device at the logical level. The processor executes the program stored in the memory and is specifically used to execute any one of the foregoing machine learning-based mineral resource distribution prediction methods.

[0106] As described above in this application Figure 1A method for predicting the distribution of mineral resources based on machine learning disclosed in the illustrated embodiment can be applied to a processor or implemented by a processor. The processor may be an integrated circuit chip with signal processing capabilities. During implementation, the steps of the above method can be completed by the integrated logic circuit of the hardware in the processor or instructions in software form. The above-mentioned processor may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components. It can implement or execute the various methods, steps, and logic block diagrams disclosed in the embodiments of the present application. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. The steps of the method disclosed in combination with the embodiments of the present application can be directly embodied as being executed and completed by a hardware decoding processor, or executed and completed by a combination of hardware and software modules in the decoding processor. The software module may be located in a mature storage medium in the art such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. This storage medium is located in the memory, and the processor reads the information in the memory and combines its hardware to complete the steps of the above method.

[0107] The electronic device can also execute Figure 1 a method for predicting the distribution of mineral resources based on machine learning, and implement Figure 1 the functions of the illustrated embodiment, which will not be elaborated herein in the embodiments of the present application.

[0108] The embodiments of the present application also propose a computer-readable storage medium that stores one or more programs. The one or more programs include instructions that, when executed by an electronic device including multiple application programs, execute any of the foregoing methods for predicting the distribution of mineral resources based on machine learning.

[0109] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can take the form of an all-hardware embodiment, an all-software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memory, CD-ROM, optical memory, etc.) that contain computer-usable program code.

[0110] The present application is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowchart and / or block diagram, and the combination of flows and / or blocks in the flowchart and / or block diagram, can be realized by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate means for realizing the functions specified in one Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0111] These computer program instructions can 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 generate a manufactured article including instruction means, and the instruction means realizes the functions specified in one Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0112] These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for realizing the functions specified in one Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0113] In a typical configuration, a computing device includes one or more processors (CPUs), an input / output interface, a network interface, and a memory.

[0114] The memory may include non-permanent memory in the computer-readable medium, random access memory (RAM), and / or non-volatile memory in the form of, for example, read-only memory (ROM) or flash memory (flash RAM). The memory is an example of a computer-readable medium.

[0115] A computer-readable medium includes both permanent and non-permanent, removable and non-removable media and can implement information storage by any method or technology. The 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 technologies, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transitory media that can be used to store information that can be accessed by a computing device. As defined herein, a computer-readable medium does not include transitory computer-readable media such as modulated data signals and carrier waves.

[0116] It should also be noted that the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, such that a process, method, article or device comprising a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "comprising an..." does not exclude the presence of additional identical elements in the process, method, article or device comprising the element.

[0117] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system or a computer program product. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0118] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for predicting mineral resource distribution based on machine learning, characterized in that: The following steps are involved: The data obtained from the preset mineral resource database within the specified time period is used as the data to be analyzed; wherein the mineral resource database includes a reference database and a pending database; the reference database is a database with a higher certainty level than the pending database; the data to be analyzed includes multiple data types, and different data types are used to characterize the mineral distribution from different types; the data to be analyzed includes observation data and environmental data; the observation data is real-time exploration data of a preset area; the observation data includes geological data, geophysical data, remote sensing data, mineral exploration data, climate data, vegetation data and hydrological data; Extracting the exploration features of the data to be analyzed, performing feature comparison analysis on the exploration features and the reference database to obtain a matching degree, and outputting the exploration features with a matching degree higher than a matching degree threshold as first data, otherwise outputting the exploration features as fuzzy data; Calculating the dissimilarity between the fuzzy data and the reference data in the matching result, performing uncertainty assessment according to the dissimilarity, outputting the fuzzy data whose dissimilarity is lower than a dissimilarity threshold as second data; and adding the second data to the first data to optimize the matching degree; A distribution probability prediction is performed based on the first data, the matching degree and the environmental data, a mineral resource distribution prediction model is constructed based on the distribution probability, the second data is added to the reference database to obtain an updated reference database, and a target prediction model is output.

2. The method for predicting mineral resource distribution based on machine learning according to claim 1, characterized in that: The data types include pictures, videos, and texts.

3. The method for predicting mineral resource distribution based on machine learning according to claim 1, characterized in that: The dissimilarity includes: Extract the feature vector of the reference data and calculate the dissimilarity between the exploration features in the fuzzy data and the reference data in the matching results: The a-th exploration feature vector of the k-th dimension at the key quantile p is The cth reference data feature vector of the kth dimension at the key quantile p is The dissimilarity between the fuzzy data and the reference data is The number of dimensions is N k , the weight of the kth dimension is The normalization factor of the kth dimension is ρ, the bandwidth parameter is U, and the spatial distance between the kth dimension environmental factor and the target area is d k , the key quantile is p.

4. The method for predicting mineral resource distribution based on machine learning according to claim 1, characterized in that: The method further includes threshold adjustment of the matching degree threshold and the dissimilarity threshold: According to the matching degree between the fuzzy data and the reference data, the graph attention network A is constructed, and the confidence of the matching degree is evaluated. The expression is: The matching degree between the ath exploration feature vector and the cth reference data feature vector is The confidence level of the match is The modulation coefficient of the matching degree is β, and the number of exploration feature vectors is N u , the attention weight of the a-th exploration feature vector and the c-th reference data feature vector is θ a,u ; The exploration feature vector and the reference data feature vector are used as nodes, and the graph attention network B is constructed according to the dissimilarity. The confidence evaluation of the dissimilarity is performed, and the expression is: The dissimilarity between the ath fuzzy data and the cth reference data is The confidence level of dissimilarity is The modulation coefficient of dissimilarity is χ, and the feature distribution of the ath node is E a , the feature distribution of the cth node is E c , the maximum divergence is HX max , characteristic distribution E a and characteristic distribution E c The divergence is HX(E a ‖E c ); The matching threshold and dissimilarity threshold are dynamically adjusted according to the confidence level. The expression is: The dynamically adjusted i-th target threshold is D i , the initial value of the i-th target threshold is The feature distribution density corresponding to the target threshold is V, and the average confidence value of the i-th target threshold is The adaptive adjustment function of feature distribution density V and confidence average is:

5. The method for predicting mineral resource distribution based on machine learning according to claim 1, characterized in that: The method further includes updating the distribution probability: According to the exploration characteristics of the first data, a deep reinforcement neural network learning is built and a probability update strategy is given. The expression is: The jth probability update strategy is T j , the index variable is j, the discount factor is ζ; the distribution probability is updated according to the probability update strategy.

6. The method for predicting mineral resource distribution based on machine learning according to claim 1, characterized in that: The method for obtaining the distribution probability comprises: Taking environmental data as the background, the influence of environmental data on the distribution of mines is obtained through the time-varying impact algorithm; the first data is dynamically adjusted according to the influence degree, the matching degree is recalculated through the adjusted first data, and the reference data with a matching degree higher than the matching threshold is obtained. The regional matching degree of the reference data is calculated, and the area with the highest regional matching degree is taken as the regional prediction result. If the matching degree error is lower than 0.12, multimodal fusion is performed on the region to obtain the regional prediction result; the spliced ​​regional prediction results are output as distribution probabilities.

7. The method for predicting mineral resource distribution based on machine learning according to claim 1, characterized in that: The method for constructing a mineral resource distribution prediction model based on the distribution probability and environmental data comprises: The weighted sum of the variance and loss function of the distribution probability prediction value and the actual distribution probability value is used as the objective function of the mineral resource distribution prediction model; Mineral resource distribution prediction models include random forest algorithm, support vector machine algorithm, and deep learning algorithm; The random forest algorithm divides the input data into training data and test data in a ratio of 7:2; The support vector machine algorithm separates the training data of different categories by finding an optimal hyperplane, and iterates continuously until the interval between two categories is maximized to obtain classified data; The deep learning algorithm automatically learns and extracts the features and patterns related to the existence of mineral resources by training classified data, and predicts the distribution of mineral resources in unknown areas with the objective function as the optimization guide; The test data is input into the mineral resource distribution prediction model to obtain the predicted mineral resource distribution. According to the mean square error between the actual mineral resource distribution and the predicted mineral resource distribution, the learning rate is continuously adjusted until the mean square error is minimized.

8. A mineral resource distribution prediction system based on machine learning, used to execute the method according to any one of claims 1 to 7, characterized in that: include: Data acquisition module: used to use the data acquired from the preset mineral resource database within a specified time period as the data to be analyzed; wherein the mineral resource database includes a reference database and a pending database; the reference database is a database with a higher certainty level than the pending database; the data includes multiple data types, and different data types are used to characterize the mineral distribution from different types; the data to be analyzed includes observation data and environmental data; the observation data is real-time exploration data of a preset area; the observation data includes geological data, geophysical data, remote sensing data, mineral exploration data, climate data, vegetation data and hydrological data; Matching degree comparison and analysis module: used for extracting the exploration features of the data to be analyzed, performing feature comparison and analysis on the exploration features and the reference database to obtain the matching degree, and outputting the exploration features with a matching degree higher than a matching degree threshold as first data, otherwise outputting them as fuzzy data; Uncertainty evaluation and optimization module: used for calculating the dissimilarity between the fuzzy data and the reference data in the matching result, performing uncertainty evaluation according to the dissimilarity, outputting the fuzzy data whose dissimilarity is lower than the dissimilarity threshold as second data; and adding the second data to the first data to optimize the matching degree; Model building module: used to predict the distribution probability based on the first data, the matching degree and the environmental data, build a mineral resource distribution prediction model based on the distribution probability, add the second data to the reference database to obtain an updated reference database, and output a target prediction model.

9. An electronic device, comprising: processor; as well as A memory arranged to store computer executable instructions, which when executed cause the processor to perform the method according to any one of claims 1 to 7.

10. A computer-readable storage medium storing one or more programs, which, when executed by an electronic device including a plurality of application programs, enables the electronic device to execute the method according to any one of claims 1 to 7.

Images