A semi-quantitative and neural network-based prediction method for ore prospecting based on airborne geophysical data

Abstract

The present invention relates to a semi-quantitative and neural network-based combined prediction prospecting method based on airborne geophysical data, comprising: a. organizing airborne geophysical data of a study area and obtaining airborne geophysical data conversion parameters; b. analyzing airborne geophysical anomaly characteristics of existing deposits and establishing a physical-mathematical deposit model; c. using the physical-mathematical deposit model to perform predictions in unknown areas using a semi-quantitative prediction method to generate semi-quantitative metallogenic favorableness evaluation parameters; d. using the semi-quantitative metallogenic favorableness parameters as constraints and using the physical-mathematical deposit model to perform neural network prediction to generate neural network predicted metallogenic favorableness evaluation parameters; e. overlaying the semi-quantitative metallogenic favorableness evaluation parameters and the neural network predicted metallogenic favorableness evaluation parameters with a geological map of the study area, analyzing the favorable metallogenic geological background, and delineating prospective metallogenic areas. The present invention can effectively utilize airborne geophysical data, combine semi-quantitative and neural network prediction methods, and accurately predict prospective metallogenic areas.

Description

Technical Field

[0001] The present invention relates to the technical field of mineral exploration, in particular to a semi-quantitative and neural network-based prediction method for prospecting based on aerial geophysical data. Background Art

[0002] With the development of instruments and equipment, more and more large-scale, high-precision airborne geophysical exploration projects have been carried out, covering most areas. Airborne geophysical exploration data contain rich mineralization geological information, which has advantages in studying the periphery of mineral deposits and deep prospecting, and can better reflect the special geochemical environment of mineral formation.

[0003] However, most of the current methods used are single mineralization prediction methods, which are not ideal, especially for the second prospecting space. In the search for deep minerals, the prospecting accuracy is low and the airborne geophysical data cannot be effectively utilized.

[0004] Further research on multi-method joint prediction based on airborne geophysical data can greatly improve the utilization efficiency and prospecting accuracy of airborne geophysical data, explore as much favorable mineralization information as possible, and provide important indicative information for further exploration of deep mineral deposits, especially for breakthroughs in prospecting in the second exploration space deeper than 500 m. Summary of the Invention

[0005] The purpose of the present invention is to provide a semi-quantitative and neural network combined prediction method based on airborne geophysical data to solve the problem that the current single mineralization prediction method is not ideal.

[0006] The present invention is achieved as follows: a semi-quantitative and neural network-based prediction prospecting method based on airborne geophysical data, characterized in that it includes the following steps.

[0007] a. Organize the aeromagnetic data and aerial radio data of the study area, and convert them into the aeromagnetic data and aerial radio data respectively to obtain the conversion parameters of the aeromagnetic data in the study area.

[0008] b. Analyze the anomalous characteristics of existing mineral deposits in the study area and establish a physical-mathematical mineral deposit model of typical mineral deposits.

[0009] c. In the unknown areas within the study area, based on the conversion parameters of the airborne geophysical data of the study area, the established physical-mathematical ore deposit model is used to adopt a semi-quantitative prediction method to generate semi-quantitative mineralization favorableness evaluation parameters.

[0010] d. Taking the semi-quantitative mineralization prediction favorableness parameters as the constraint conditions, the established physical-mathematical ore deposit model is used for neural network prediction. The conversion parameters of various airborne geophysical data in the study area are superimposed to generate neural network prediction mineralization favorableness evaluation parameters.

[0011] e. Overlay the semi-quantitative metallogenic favorableness evaluation parameters and the neural network predicted metallogenic favorableness evaluation parameters with the geological map of the study area, analyze the favorable metallogenic geological background, and delineate the prospective metallogenic areas.

[0012] In step a, the collated airborne geophysical data include aeromagnetic ΔT and basic parameters such as total airborne uranium content, airborne potassium content, and airborne thorium content.

[0013] In step a, the data conversion processing performed on the aeromagnetic data includes the average value, local anomaly, horizontal gradient modulus, vertical first (second) order derivative, entropy, skewness, kurtosis, and upward continuation; the data conversion processing performed on the aerial release data includes the average value, local anomaly, entropy, skewness, and kurtosis.

[0014] In step b, a known mineral deposit area is selected in the study area, and the mineral deposit is projected onto the conversion parameter information of each airborne geophysical data obtained in step a. The relationship between the known mineral deposit and the physical field is analyzed, and the correlation coefficient, favorable information area ratio and Fisher criterion of the known mineral deposit and the physical field are adjusted to determine the optimal field value combination relationship of the mineral deposit in the conversion field, thereby establishing a physical-mathematical mineral deposit model of a typical mineral deposit.

[0015] In step c, the data in each airborne geophysical data conversion parameter that conforms to the physical-mathematical ore deposit model is assigned 1, and the other data is assigned 0, where 1 represents favorable and 0 represents unfavorable. Then all parameter information is superimposed and analyzed to generate semi-quantitative mineralization favorableness evaluation parameters.

[0016] In step d, the semi-quantitative favorableness parameter obtained in step c is used as a constraint condition to constrain the parameters of each conversion field involved in the prediction. The data in each conversion field that conforms to the ore deposit model is assigned a certain weight according to the size of the correlation coefficient, representing favorable. Other data with a correlation coefficient less than the given value is assigned 0, representing unfavorable. Then, information superposition analysis is performed to generate a neural network to predict the evaluation parameters of mineralization favorableness.

[0017] In step e, the predicted favorableness information is projected onto the geological map, the metallogenic geological conditions of the highly favorable areas are analyzed, and areas with multiple composite information are selected to be delineated as prospective metallogenic areas.

[0018] The present invention uses known ore deposits in the study area to establish a physical-mathematical ore deposit model, uses the model to make predictions in unknown ore deposit areas, uses a semi-quantitative method to generate semi-quantitative metallogenic favorableness evaluation parameters, and uses the semi-quantitative metallogenic prediction favorableness parameters as constraints to perform neural network prediction to generate neural network predicted metallogenic favorableness evaluation parameters. Finally, the semi-quantitative metallogenic favorableness evaluation parameters and the neural network predicted metallogenic favorableness evaluation parameters are combined, the combined favorableness information is superimposed on the geological map, the favorable metallogenic geological background is analyzed, and the prospective metallogenic areas are delineated.

[0019] Airborne geophysical data include aeromagnetic and aerial radio data. After data conversion processing, a variety of airborne geophysical data conversion parameters can be obtained. Based on the analysis of a variety of airborne geophysical data conversion parameters, the physical-mathematical deposit model of a typical mineral deposit can be obtained. This model contains rich information and makes full use of airborne geophysical data.

[0020] The present invention can effectively utilize aerial geophysical data and, by combining semi-quantitative and neural network prediction methods, can accurately predict prospective mineralization areas, solving the problem of unsatisfactory results of a single mineralization prediction method. It provides an effective technical method for utilizing aerial geophysical data for mineral exploration, has broad application prospects, and is suitable for popularization and application. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 It is a flow chart of the present invention. DETAILED DESCRIPTION

[0022] The present invention is a semi-quantitative and neural network combined prediction prospecting method based on airborne geophysical data, which includes the following steps.

[0023] a. Organize the aeromagnetic data and aerial radio data of the study area, and convert them into the aeromagnetic data and aerial radio data respectively to obtain the conversion parameters of the aeromagnetic data in the study area.

[0024] b. Analyze the anomalous characteristics of existing mineral deposits in the study area and establish a physical-mathematical mineral deposit model of typical mineral deposits.

[0025] c. In the unknown areas within the study area, based on the conversion parameters of the airborne geophysical data of the study area, the established physical-mathematical ore deposit model is used to adopt a semi-quantitative prediction method to generate semi-quantitative mineralization favorableness evaluation parameters.

[0026] d. Taking the semi-quantitative mineralization prediction favorableness parameters as the constraint conditions, the established physical-mathematical ore deposit model is used for neural network prediction. The conversion parameters of various airborne geophysical data in the study area are superimposed to generate neural network prediction mineralization favorableness evaluation parameters.

[0027] e. Overlay the semi-quantitative metallogenic favorableness evaluation parameters and the neural network predicted metallogenic favorableness evaluation parameters with the geological map of the study area, analyze the favorable metallogenic geological background, and delineate the prospective metallogenic areas.

[0028] The detailed steps of the present invention are as follows.

[0029] Airborne geophysical data from the study area were compiled and prepared. These data included aeromagnetic ΔT and basic parameters such as total airborne uranium, potassium, and thorium concentrations, and were converted. The aeromagnetic ΔT data were converted using parameters such as mean, local anomaly, horizontal gradient modulus, vertical first and second derivatives, entropy, skewness, kurtosis, and upward continuation. The airborne geophysical data (total airborne uranium, potassium, and thorium concentrations) were converted using parameters such as mean, local anomaly, entropy, skewness, and kurtosis. This resulted in the conversion parameters for the airborne geophysical data in the study area.

[0030] A known mineral deposit area (horizontal projection range of the ore body) is selected in the study area, and the mineral deposit is projected onto the conversion parameter information of each airborne geophysical data. The relationship between the known mineral deposit and the physical field is analyzed, and the correlation coefficient, favorable information area ratio and Fisher criterion of the known mineral deposit and the physical field are adjusted to determine the optimal field value combination relationship of the field where the mineral deposit is located in the conversion field, thereby establishing a physical-mathematical mineral deposit model of a typical mineral deposit.

[0031] The formation of mineral deposits has its own special geochemical environment, which is reflected in the geophysical fields (such as magnetic field, gravity field and gamma field, etc.). There is a specific combination relationship between the geophysical field and its related conversion field values. The physical-mathematical model of the mineral deposit can be established, and the optimal combination relationship of the conversion field values ​​can be determined by adjusting the correlation coefficient parameters between the mineral deposit and the physical field.

[0032] Mineral deposits have specific geophysical fields and their related conversion field value combinations. During the modeling process, corresponding parameters are selected to ensure that the physical field characteristics can be best described. These conversion parameters are obtained by sliding the window and performing corresponding statistics and calculations on the data in the window.

[0033] Apply existing ore deposit models to areas with unknown deposits to predict mineralization prospects. The choice of ore deposit model depends on whether there are known deposits in the study area. In areas where existing deposits are present, the model is preferred, with appropriate reference to ore deposit models from other areas or abroad. For areas where the research is less advanced and no deposits have yet been discovered, only existing ore deposit models from other areas can be used.

[0034] First, a semi-quantitative mineralization prediction is carried out. The data in the conversion parameters of each airborne geophysical data that conform to the previously established physical-mathematical mineral deposit model are assigned 1, and the other data that do not conform to the physical-mathematical mineral deposit model are assigned 0, where 1 represents favorable and 0 represents unfavorable. Then, all parameters are superimposed and analyzed to generate semi-quantitative mineralization favorableness evaluation parameters.

[0035] Then, neural network mineralization prediction is carried out. The semi-quantitative favorableness parameters obtained above are used as constraints. Various conversion parameters are assigned in favorable areas (unfavorable areas are assigned to 0). The parameters of each conversion field involved in the prediction are constrained. The data that conforms to the ore deposit model in each conversion field is assigned a certain weight according to the size of the correlation coefficient, representing favorable. For other data with correlation coefficients less than the given value, it is assigned 0, representing unfavorable. Then, information superposition analysis is carried out to generate neural network prediction mineralization favorableness evaluation parameters.

[0036] When performing neural network mineralization prediction, the prediction data for the study area is first input, converted, and extracted into individual prediction subgrids. The ore deposit model is then input, the ore deposit subgrid files are extracted, and the correlation coefficients between each prediction subgrid and the ore deposit subgrid are calculated. The prediction information is then selected based on the correlation coefficients and standardized. The weights for each prediction information are then determined based on the correlation coefficients, and non-geological structural mineralization-favorable information is calculated. This non-geological structural mineralization-favorable information is then combined with the geological and structural information to form the final mineralization prediction.

[0037] Finally, based on the favorableness parameters obtained by the joint prediction of semi-quantitative and neural networks, the predicted favorableness information is projected onto the geological map, the metallogenic geological conditions of the high favorableness areas are analyzed, and the areas with favorable metallogenic geological conditions and good predicted favorableness (multiple information composite areas) are selected as prospective mineralization areas.

[0038] The present invention uses known ore deposits in the study area to establish a physical-mathematical ore deposit model, uses the model to make predictions in unknown ore deposit areas, uses a semi-quantitative method to generate semi-quantitative metallogenic favorableness evaluation parameters, and uses the semi-quantitative metallogenic prediction favorableness parameters as constraints to perform neural network prediction to generate neural network predicted metallogenic favorableness evaluation parameters. Finally, the semi-quantitative metallogenic favorableness evaluation parameters and the neural network predicted metallogenic favorableness evaluation parameters are combined, the combined favorableness information is superimposed on the geological map, the favorable metallogenic geological background is analyzed, and the prospective metallogenic areas are delineated.

[0039] Airborne geophysical data include aeromagnetic and aerial radio data. After data conversion processing, a variety of airborne geophysical data conversion parameters can be obtained. Based on the analysis of a variety of airborne geophysical data conversion parameters, the physical-mathematical deposit model of a typical mineral deposit can be obtained. This model contains rich information and makes full use of airborne geophysical data.

[0040] The present invention can effectively utilize aerial geophysical data and, by combining semi-quantitative and neural network prediction methods, can accurately predict prospective mineralization areas, solving the problem of unsatisfactory results of a single mineralization prediction method. It provides an effective technical method for utilizing aerial geophysical data for mineral exploration, has broad application prospects, and is suitable for popularization and application.

Claims

1. A semi-quantitative and neural network-based prediction method for mineral exploration based on airborne geophysical data, characterized in that: The following steps are involved: a. Organize the aeromagnetic data and aerial radio data of the study area, and convert them into the aeromagnetic data and aerial radio data respectively, and obtain the conversion parameters of the aeromagnetic data of the study area; b. Analyze the anomalous characteristics of existing mineral deposits in the study area through airborne geophysical exploration and establish a physical-mathematical mineral deposit model for typical mineral deposits; Select a known mineral deposit area in the study area and project the mineral deposit onto the conversion parameter information of each airborne geophysical data obtained in step a. Analyze the relationship between the known mineral deposit and the physical field. Adjust the correlation coefficient, favorable information area ratio, and Fisher criterion between the known mineral deposit and the physical field to determine the optimal field value combination relationship of the mineral deposit in the conversion field, thereby establishing a physical-mathematical mineral deposit model for the typical mineral deposit. c. In unknown areas within the study area, based on the parameters converted from the airborne geophysical data of the study area, using the established physical-mathematical ore deposit model, a semi-quantitative prediction method is used to generate semi-quantitative mineralization favorableness evaluation parameters; d. Using the semi-quantitative mineralization favorableness evaluation parameters as constraints, the established physical-mathematical ore deposit model is used for neural network prediction. The parameters converted from various airborne geophysical data in the study area are superimposed to generate neural network prediction parameters for mineralization favorableness evaluation; Using the semi-quantitative favorableness parameter obtained in step c as a constraint condition, the parameters of each conversion field involved in the prediction are constrained. The data that conforms to the ore deposit model in each conversion field is assigned a certain weight based on the size of the correlation coefficient, representing favorable. Other data with a correlation coefficient less than the given value is assigned 0, representing unfavorable. Then, information superposition analysis is performed to generate a neural network prediction of the mineralization favorableness evaluation parameters; e. Overlay the semi-quantitative metallogenic favorableness evaluation parameters and the neural network predicted metallogenic favorableness evaluation parameters with the geological map of the study area, analyze the favorable metallogenic geological background, and delineate the prospective metallogenic areas.

2. The semi-quantitative and neural network combined prediction method for prospecting based on airborne geophysical data according to claim 1 is characterized in that: In step a, the collated airborne geophysical data include aeromagnetic ΔT and basic parameters such as total airborne uranium content, airborne potassium content, and airborne thorium content.

3. The semi-quantitative and neural network combined prediction method for prospecting based on airborne geophysical data according to claim 2 is characterized in that: In step a, the data conversion processing performed on the aeromagnetic data includes the average value, local anomaly, horizontal gradient modulus, vertical first-order derivative or vertical second-order derivative, entropy, skewness, kurtosis, and upward continuation; the data conversion processing performed on the aerial release data includes the average value, local anomaly, entropy, skewness, and kurtosis.

4. The semi-quantitative and neural network combined prediction method for prospecting based on airborne geophysical data according to claim 1 is characterized in that: In step c, the data in each airborne geophysical data conversion parameter that conforms to the physical-mathematical ore deposit model is assigned 1, and the other data is assigned 0, where 1 represents favorable and 0 represents unfavorable. Then all parameter information is superimposed and analyzed to generate semi-quantitative mineralization favorableness evaluation parameters.

5. The semi-quantitative and neural network combined prediction method for prospecting based on airborne geophysical data according to claim 1 is characterized in that: In step e, the predicted favorableness information is projected onto the geological map, the metallogenic geological conditions of the highly favorable areas are analyzed, and areas with multiple composite information are selected to be delineated as prospective metallogenic areas.

Images