CN117238405B - Geochemical data analysis method and device based on deep learning - Google Patents
Geochemical data analysis method and device based on deep learning Download PDFInfo
- Publication number
- CN117238405B CN117238405B CN202311219804.6A CN202311219804A CN117238405B CN 117238405 B CN117238405 B CN 117238405B CN 202311219804 A CN202311219804 A CN 202311219804A CN 117238405 B CN117238405 B CN 117238405B
- Authority
- CN
- China
- Prior art keywords
- data
- ore
- deep learning
- calculated
- depth
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 28
- 238000013135 deep learning Methods 0.000 title claims abstract description 23
- 238000007405 data analysis Methods 0.000 title claims abstract description 20
- 239000000454 talc Substances 0.000 claims description 115
- 229910052623 talc Inorganic materials 0.000 claims description 115
- 235000012222 talc Nutrition 0.000 claims description 115
- 238000005070 sampling Methods 0.000 claims description 69
- XEEYBQQBJWHFJM-UHFFFAOYSA-N Iron Chemical compound [Fe] XEEYBQQBJWHFJM-UHFFFAOYSA-N 0.000 claims description 61
- 238000013136 deep learning model Methods 0.000 claims description 51
- 230000008021 deposition Effects 0.000 claims description 38
- 238000004364 calculation method Methods 0.000 claims description 32
- XUIMIQQOPSSXEZ-UHFFFAOYSA-N Silicon Chemical compound [Si] XUIMIQQOPSSXEZ-UHFFFAOYSA-N 0.000 claims description 31
- 229910052710 silicon Inorganic materials 0.000 claims description 31
- 239000010703 silicon Substances 0.000 claims description 31
- NINIDFKCEFEMDL-UHFFFAOYSA-N Sulfur Chemical compound [S] NINIDFKCEFEMDL-UHFFFAOYSA-N 0.000 claims description 30
- 229910052742 iron Inorganic materials 0.000 claims description 30
- 229910052717 sulfur Inorganic materials 0.000 claims description 30
- 239000011593 sulfur Substances 0.000 claims description 30
- 239000011435 rock Substances 0.000 claims description 21
- 238000000547 structure data Methods 0.000 claims description 16
- 229910052947 chalcocite Inorganic materials 0.000 claims description 14
- GWEVSGVZZGPLCZ-UHFFFAOYSA-N titanium dioxide Inorganic materials O=[Ti]=O GWEVSGVZZGPLCZ-UHFFFAOYSA-N 0.000 claims description 14
- 239000010453 quartz Substances 0.000 claims description 9
- VYPSYNLAJGMNEJ-UHFFFAOYSA-N silicon dioxide Inorganic materials O=[Si]=O VYPSYNLAJGMNEJ-UHFFFAOYSA-N 0.000 claims description 9
- 238000013527 convolutional neural network Methods 0.000 claims description 7
- 239000013049 sediment Substances 0.000 claims description 6
- 238000004422 calculation algorithm Methods 0.000 claims description 5
- 239000011044 quartzite Substances 0.000 claims description 5
- 238000010586 diagram Methods 0.000 claims description 4
- 238000005457 optimization Methods 0.000 claims description 4
- 238000010276 construction Methods 0.000 claims description 3
- FPAFDBFIGPHWGO-UHFFFAOYSA-N dioxosilane;oxomagnesium;hydrate Chemical compound O.[Mg]=O.[Mg]=O.[Mg]=O.O=[Si]=O.O=[Si]=O.O=[Si]=O.O=[Si]=O FPAFDBFIGPHWGO-UHFFFAOYSA-N 0.000 claims description 3
- 230000006870 function Effects 0.000 claims description 3
- 239000004575 stone Substances 0.000 claims 3
- 229910052500 inorganic mineral Inorganic materials 0.000 description 6
- 239000011707 mineral Substances 0.000 description 6
- 235000010755 mineral Nutrition 0.000 description 6
- 230000015572 biosynthetic process Effects 0.000 description 5
- 239000000203 mixture Substances 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 230000000694 effects Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 239000012530 fluid Substances 0.000 description 2
- 239000011573 trace mineral Substances 0.000 description 2
- 235000013619 trace mineral Nutrition 0.000 description 2
- 230000002776 aggregation Effects 0.000 description 1
- 238000004220 aggregation Methods 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 239000013078 crystal Substances 0.000 description 1
- 125000004122 cyclic group Chemical group 0.000 description 1
- -1 deposit type Substances 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 239000003673 groundwater Substances 0.000 description 1
- 239000010423 industrial mineral Substances 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000000155 isotopic effect Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 210000003462 vein Anatomy 0.000 description 1
Landscapes
- Geophysics And Detection Of Objects (AREA)
Abstract
The invention provides a geochemical data analysis method and a device based on deep learning, which relate to the technical field of geochemical data analysis methods.
Description
Technical Field
The invention relates to the technical field of geochemical data analysis methods, in particular to a geochemical data analysis method and device based on deep learning.
Background
Geochemistry is the subject of studying geochemical composition, cyclic processes and interactions, and is widely used in geological exploration, which can provide information about geologic structures, mineral composition, deposit type, groundwater resources, etc., helping users to predict various mineral resources. With the rise of big data and artificial intelligence technology, machine learning methods such as vector machines and convolutional neural networks are also used for calculation in geochemical data analysis. Compared with the traditional manual calculation mode, the machine deep learning training device has the advantages that the machine deep learning training device can be used for analyzing and processing large-batch and high-latitude data more quickly, and when input variables are more complex, the machine deep learning training device has more obvious advantages.
Talc is a very important industrial mineral, the study of which has a great significance for geochemistry, the presence of which can provide important clues to the environment in which the deposit is formed, the formation of which generally requires specific temperature, pressure and chemical environmental conditions. By studying the trace elements, isotopic composition and crystal structure in talc, it is possible to infer the depth of crust, temperature, fluid composition and magma and deposit formation processes at deposit formation, and the formation and stable presence of talc in connection with geothermal activity and trace elements can be used as indicators to help study the characteristics and evolution processes of geothermal systems, so key information on deposit formation, geothermal activity, geological processes and fluid circulation can be obtained by studying the location, depth etc. of talc.
In the prior art, various geological conditions, geophysics, geochemistry, remote sensing data and the like are generally used as input variables to predict the position of talc ores, but the prediction results are generally only plane coordinates of predicted mineral points, the plane coordinates are converted into longitudes and latitudes to confirm the positions, the depth of the predicted mineral points is not outputted, the depth positions of the predicted mineral points in a stratum are difficult to distinguish, and the types of the talc ores are not distinguished, so that erroneous judgment on the prediction results is easy to occur in the prior art, so that it is necessary to provide an analysis method capable of predicting the depth positions of the mineral points by using deep learning and analyzing the talc ore types in an unexplored mine according to the prediction depth.
Disclosure of Invention
The invention aims to provide a geochemical data analysis method and device based on deep learning, which are used for solving the problems in the background technology.
In order to achieve the above purpose, the present invention provides the following technical solutions:
a geochemical data analysis method based on deep learning comprises the following steps:
acquiring data of existing talcum ores, including position information, geological structure data, lithology data, stratum data, landform topographic data and age data of the areas, forming a data set of each area, marking on geological map data, numbering the existing talcum ores, forming sample tags, and enabling the data set of each area to correspond to the sample tags of the area one by one;
a deep learning model based on a convolutional neural network is constructed, data of the existing talc ore is input into the deep learning model, the average deposition rate of the position of the existing talc ore is generated, and the depth of the talc ore is calculated according to the age data;
According to the actual depth of the existing talcum ore, generating an ideal confidence interval of the talcum ore depth calculated by a deep learning model, then according to the talcum ore depth calculated by the deep learning model, generating a calculated confidence interval, and comparing the ideal confidence interval with the calculated confidence interval to evaluate the deep learning model;
logic for calculating the depth of the talc ore based on the age data is:
Calibrating the data acquired by the sampling points, numbering the data from the ground surface to the sedimentary layers, calibrating the thickness of each sedimentary layer to be D i, and calibrating the age of each sampling point to be The depth of each sampling point relative to the surface is calibrated as/>I represents the number of layers of the deposited layer, i=1, 2,3,..;
According to the age of sampling point in certain sediment layer And a depth calibration relative to the surface of the earth is/>The average deposition rate of the deposited layer was calculated and scaled to/>The calculation formula is as follows:
wherein, Represents the depth of the (j+1) th sampling point of the (i) th deposition layer,/>Representing the age of the (j+1) th sampling point of the (i) th sediment layer;
Fitting by adopting a logarithmic function according to the duty ratio of the thickness D i of different deposition layers in the total thickness of all deposition layers to generate a weight coefficient f i of different deposition layers, wherein the formula for generating the weight coefficient f i is as follows:
wherein alpha represents a weighting coefficient, alpha is more than 0, the weighting coefficient is used for adjusting the deep learning model, and e represents the base number of natural logarithm;
According to the average deposition rate of different deposition layers And a weight coefficient f i, calculating the average deposition rate/>, of the positions of the talcum oresThe calculation formula is as follows:
the age of the existing talc ore is marked as Th, and Th and the average deposition rate of the talc ore are based on the age of the talc ore Generating a talcum ore depth Lh' calculated by a deep learning model, wherein the calculation formula is as follows:
lh' represents the depth of the talc ore calculated and generated by the deep learning model according to the position information, geological structure data, lithology data, stratum data, geomorphic topographic data, age data of the area where the talc ore is located and data acquired by the sampling points.
Preferably, the location information of the talc ore includes plane coordinates and depth of the talc ore, the geological structure data includes a structure map, a geological map, topographic data and stratum data of the area, the lithology data includes rock species distribution data of quartzite, chalcocite and rutile, the stratum data includes data of a sedimentary layer, the topographic data includes topographic line, topographic feature data showing relief of a contour line, and the age data includes age of the talc ore.
Preferably, the data of the rock species distribution of the quartzite, the chalcocite and the rutile comprise the content and the distribution sequence of the quartzite, the chalcocite and the rutile, and the data of the deposition layer comprise the number of layers and the thickness of the deposition layer and the ages corresponding to different depths.
Preferably, the sampling mode of the existing smooth rock ore is that the sampling is longitudinally performed at the central position of the plane coordinate of the existing smooth rock ore, the number of layers of the deposited layers and the thickness of each deposited layer are obtained, n sampling points are arranged in each deposited layer, the interval distances of the n sampling points in the same deposited layer are equal, and the age of each sampling point and the depth relative to the ground surface are measured.
Preferably, the determining logic for determining whether the deep learning model meets the requirements is:
Dividing data of all talcum ore depths generated by a deep learning model after multiple times of calculation into x data sets according to a time sequence, calibrating the data into W x, wherein x is a positive integer, numbering the data in the data sets W x, calibrating the data into Lh y ', y is a positive integer, p data are arranged in each data set W x, W x＝{Lh1′,Lh2′,Lh3′,......,Lhp' is a positive integer, the data in the data sets W x are approximately subjected to Gaussian distribution, a 95% confidence level is obtained, the actual depth of the existing talcum ore is calibrated into Lh, the actual depth of the existing talcum ore is taken as an ideal mean value in the data sets W x, the ideal standard deviation of the data sets W x is calibrated into S x, and a calculation formula is as follows:
An ideal confidence interval for the data set W x is generated from the ideal standard deviation S x, as follows:
the t value can be found in the t value distribution table according to the confidence level and p;
The calculated average value calculated by the data set W x is calibrated as The calculation formula is as follows:
calibrating the calculated standard deviation calculated by the data set W x as S x', wherein the calculation formula is as follows:
A calculated confidence interval for the dataset W x is generated from the calculated standard deviation S x' as follows:
Comparing the ideal confidence interval of the dataset W x with the calculated confidence interval, and if and only if the calculated confidence interval falls within the ideal confidence interval, calculating an average The actual depth Lh of the ore closer to the talcum powder is calculated more accurately, and the calculation is considered to be qualified at the moment, and the judgment formula is as follows:
When q data sets W x are continuously calculated to be qualified, the deep learning model is proved to meet the requirement, the verified deep learning model can be output, otherwise, the weighting coefficients and parameters of the model are adjusted through a back propagation and optimization algorithm according to errors.
Preferably, the data analysis logic for the unexplored mine is:
Setting m sampling points at the same depth of talcum ore, measuring element content of each sampling point, wherein the silicon content is respectively Sulfur content is/>, respectively Iron content is/>, respectivelyWherein/> The content values of silicon, sulfur and iron at the r point are respectively represented, and the content values of silicon, sulfur and iron are subjected to the extremely-bad treatment, and the calculation formula is as follows:
In the middle of Extremely bad data of silicon, sulfur and iron are shown respectively,/>Respectively represent the maximum value of silicon, sulfur and iron contents in sampling points,/>Respectively representing the minimum values of silicon, sulfur and iron contents in the sampling points;
Calculating a range average value according to range data of silicon, sulfur and iron, and calculating a judgment coefficient according to the range average value, wherein the calculation formula of the judgment coefficient is as follows:
In the middle of The extremely poor average values of silicon, sulfur and iron are respectively shown;
The type of talc ore and possibly associated ore are judged according to the judgment coefficient theta 1、θ2、θ3, and the judgment method is as follows:
the smooth ore is rich in silicon, and the possible associated ore is quartz rock;
the smooth ore is sulfur-rich, and the possible associated ore is chalcocite;
The talc ore is rich in iron and the associated ore may be rutile.
The invention also provides a geochemical data analysis device based on deep learning, which is used for realizing the geochemical data analysis method, and comprises the following steps:
The sample construction module is used for acquiring the data of the existing talcum ores, including the position information, geological structure data, lithology data, stratum data, landform topographic data and age data of the area, forming a data set of each area, marking the geological map data, numbering the existing talcum ores, forming sample tags, and enabling the data set of each area to correspond to the sample tags of the area one by one;
The deep learning module is used for constructing a deep learning model based on a convolutional neural network, inputting the data of the existing smooth ore into the deep learning model, generating the average deposition rate of the position of the existing smooth ore, and calculating the depth of the smooth ore according to the age data;
the interval prediction module is used for generating an ideal confidence interval of the talc ore depth calculated by the deep learning model according to the actual depth of the existing talc ore, generating a calculated confidence interval according to the talc ore depth calculated by the deep learning model, and comparing the ideal confidence interval with the calculated confidence interval to evaluate the deep learning model;
And the type prediction module is used for carrying out field sampling on the unexplored mine according to the calculated talcum ore depth to obtain element content data of sampling points, analyzing the element content data, judging the type of the talcum ore at the sampling points and predicting associated ores possibly existing at the sampling points.
Compared with the prior art, the invention has the beneficial effects that:
according to the method, the depth learning model is built through the position information, the geological structure data, the lithology data, the stratum data, the landform topographic data and the age data of the existing area where the talcum ore is located, the average deposition rate of the geology of the talcum ore point can be calculated, so that the depth of the geology where the talcum ore is located is deduced according to the age of a small amount of collected talcum samples, and meanwhile, the depth of the talcum ore calculated by the depth learning model can be more approximate to the actual depth of the talcum ore through updating and iterating the parameters of the depth learning model, so that the prediction result of the depth learning model is more accurate.
Drawings
FIG. 1 is a schematic flow chart of the method of the present invention.
Detailed Description
The present invention will be further described in detail with reference to specific embodiments in order to make the objects, technical solutions and advantages of the present invention more apparent.
It is to be noted that unless otherwise defined, technical or scientific terms used herein should be taken in a general sense as understood by one of ordinary skill in the art to which the present invention belongs. The terms "first," "second," and the like, as used herein, do not denote any order, quantity, or importance, but rather are used to distinguish one element from another. The word "comprising" or "comprises", and the like, means that elements or items preceding the word are included in the element or item listed after the word and equivalents thereof, but does not exclude other elements or items. The terms "connected" or "connected," and the like, are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "up", "down", "left", "right" and the like are used only to indicate a relative positional relationship, and when the absolute position of the object to be described is changed, the relative positional relationship may be changed accordingly.
Examples:
referring to fig. 1, the present invention provides a technical solution:
a geochemical data analysis method based on deep learning specifically comprises the following steps:
The method comprises the steps of obtaining position information, geological structure data, lithology data, stratum data, geomorphic topographic data and age data of an area where the existing talcum ore is located, forming a data set of each area, marking the geological map data, numbering the existing talcum ore, forming a sample label, and enabling the data set of each area to correspond to the sample label of the area one by one.
The area where the talcum ore adopted in the embodiment is located is the talcum ore which is mined and excavated in China, the position information of the point comprises the plane coordinates and depth of the talcum ore, the geological structure data comprise the structure diagram, the geological diagram, the topographic data and the stratum data of the area, the lithology data comprise the rock species distribution data of quartz rock, chalcocite and rutile, the stratum data comprise the data of a sedimentary layer, the topographic data comprise topographic feature data of topographic lines and contour lines showing relief, and the age data comprise the age of the talcum ore.
Further, the rock species distribution data of the quartz rock, the chalcocite and the rutile comprise the content and the distribution sequence of the quartz rock, the chalcocite and the rutile, the quartz rock, the chalcocite and the rutile are common associated ores of the talc ores, the type of the talc ores can be predicted through the content and the distribution sequence of the quartz rock, the chalcocite and the rutile, and the data of the deposit layer comprise the layer number and the thickness of the deposit layer and the ages corresponding to different depths.
The sampling mode of the existing smooth rock ore is that the central position of the plane coordinate of the existing smooth rock ore is longitudinally sampled, the number of layers of the deposited layers and the thickness of each deposited layer are obtained, n sampling points are arranged in each deposited layer, the interval distances of the n sampling points in the same deposited layer are equal, and the age of each sampling point and the depth relative to the ground surface are measured.
The method comprises the steps of constructing a deep learning model based on a convolutional neural network, inputting position information, geological structure data, lithology data, stratum data, geomorphic topographic data and age data of an area where the existing talcum ore is located into the deep learning model, generating an average deposition rate of the position where the existing talcum ore is located, and calculating the depth of the talcum ore according to the age data.
Calibrating data acquired by sampling points, numbering the data from the earth surface to the sedimentary layers, calibrating the thickness of each sedimentary layer to be Di, and calibrating the age of each sampling point to be DiThe depth of each sampling point relative to the surface is calibrated as/>I denotes the number of layers of the deposited layer, i=1, 2,3,...
According to the age of sampling point in certain sediment layerAnd a depth calibration relative to the surface of the earth is/>The average deposition rate of the deposited layer was calculated and scaled to/>The calculation formula is as follows:
wherein, Represents the depth of the (j+1) th sampling point of the (i) th deposition layer,/>Representing the age of the (j+1) th sampling point of the (i) th sediment layer;
Fitting by adopting a logarithmic function according to the duty ratio of the thickness D i of different deposition layers in the total thickness of all deposition layers to generate a weight coefficient f i of different deposition layers, wherein the formula for generating the weight coefficient f i is as follows:
Where α represents a weighting coefficient, α > 0, for adjusting the depth model, and e represents a base of natural logarithm.
According to the average deposition rate of different deposition layersAnd a weight coefficient f i, calculating the average deposition rate/>, of the positions of the talcum oresThe calculation formula is as follows:
the age of the existing talc ore is marked as Th, and Th and the average deposition rate of the talc ore are based on the age of the talc ore Generating a talcum ore depth Lh' calculated by a deep learning model, wherein the calculation formula is as follows:
lh' represents the depth of the talc ore calculated and generated by the deep learning model according to the position information, geological structure data, lithology data, stratum data, geomorphic topographic data, age data of the area where the talc ore is located and data acquired by the sampling points.
According to the actual depth of the existing talcum ore, an ideal confidence interval of the talcum ore depth calculated by the deep learning model is generated, then according to the talcum ore depth calculated by the deep learning model, a calculated confidence interval is generated, if and only if the calculated confidence interval falls in the ideal confidence interval, the condition is met, the deep learning model is proved to meet the requirement, the verified deep learning model can be output, and otherwise, the weighting coefficient and the parameter of the model are adjusted through the counter-propagation and optimization algorithm according to the error.
Dividing data of all talcum ore depths generated by a deep learning model after multiple times of calculation into x data sets according to a time sequence, calibrating the data into W x, wherein x is a positive integer, numbering the data in the data sets W x, calibrating the data into Lh y ', y is a positive integer, p data are arranged in each data set W x, W x＝{Lh1′,Lh2′,Lh3′,......,Lhp' is a positive integer, the data in the data sets W x are approximately subjected to Gaussian distribution, a 95% confidence level is obtained, the actual depth of the existing talcum ore is calibrated into Lh, the actual depth of the existing talcum ore is taken as an ideal mean value in the data sets W x, the ideal standard deviation of the data sets W x is calibrated into S x, and a calculation formula is as follows:
the ideal confidence interval for the dataset W x is regenerated as follows:
the t value can be found from the t value distribution table based on the confidence level and p.
The calculated average value of the data set W x is then calibrated asThe calculation formula is as follows:
The calculated standard deviation calculated by the dataset W x is calibrated as S x', and the calculation formula is:
The calculated confidence interval for the dataset W x is regenerated, and the ideal confidence interval is as follows:
The ideal confidence interval of the dataset W x is compared with the calculated confidence interval, and if and only if the calculated confidence interval falls within the ideal confidence interval, an average value is calculated The actual depth Lh of the ore closer to the talcum powder is calculated more accurately, and the calculation is considered to be qualified at the moment, and the judgment formula is as follows:
When q data sets W x are continuously calculated to be qualified, the deep learning model is proved to meet the requirement, the verified deep learning model can be output, otherwise, the weighting coefficients and parameters of the model are adjusted through a back propagation and optimization algorithm according to errors.
According to the calculated talcum ore depth, the non-mined ore field is sampled to obtain element content data of sampling points, the element content data is analyzed, the type of the talcum ore at the position is judged, accompanying ores possibly existing at the position are predicted, and the judgment logic is as follows:
Setting m sampling points at the same depth of talcum ore, measuring element content of each sampling point, wherein the silicon content is respectively Sulfur content is/>, respectively Iron content is/>, respectivelyWherein/> The content values of silicon, sulfur and iron at the r point are respectively represented, and the content values of silicon, sulfur and iron are subjected to the extremely-bad treatment, and the calculation formula is as follows:
In the middle of Extremely bad data of silicon, sulfur and iron are shown respectively,/>Respectively represent the maximum value of silicon, sulfur and iron contents in sampling points,/>Respectively represent the minimum values of silicon, sulfur and iron contents in the sampling points.
And calculating a range average value according to the range data of the silicon, the sulfur and the iron, and calculating a judgment coefficient according to the range average value, wherein the calculation formula of the judgment coefficient is as follows:
In the middle of The very poor averages of silicon, sulfur, and iron are shown, respectively.
Under normal conditions, the distribution of silicon, sulfur and iron in a plurality of sampling points is relatively uniform, and the difference value of the content values among samples is relatively small, namely the extremely poor data of the silicon, the sulfur and the ironThe fluctuations are also small and are all approximately equal to 1, the calculated range average/>All are equal to about 1, at the moment, the judging coefficients theta 1、θ2、θ3 are all at the zero position, when the silicon element enrichment occurs in the sampling area, the difference value of the silicon element in the sampling point is increased, and the extremely bad data/>, of the silicon elementFluctuation in [0,1], extremely bad average/>The data of sulfur and iron elements are reduced to be less than 1, at the moment, theta 1＜0,θ2＝0,θ3 is more than 0, the associated ore which possibly exists is quartz rock, when sulfur element enrichment occurs in a sampling area, theta 1＞0,θ2＜0,θ3 =0, the associated ore which possibly exists is chalcocite, when iron element enrichment occurs in the sampling area, theta 1＝0,θ2＞0,θ3 is less than 0, the associated ore which possibly exists is iron-rich, and the associated ore which possibly exists is rutile.
The invention also provides a geochemical data analysis device based on deep learning, which is used for realizing the geochemical data analysis method, and comprises the following steps:
The sample construction module is used for acquiring the data of the existing talcum ores, including the position information, geological structure data, lithology data, stratum data, landform topographic data and age data of the area, forming a data set of each area, marking the geological map data, numbering the existing talcum ores, forming sample tags, and enabling the data set of each area to correspond to the sample tags of the area one by one;
The deep learning module is used for constructing a deep learning model based on a convolutional neural network, inputting the data of the existing smooth ore into the deep learning model, generating the average deposition rate of the position of the existing smooth ore, and calculating the depth of the smooth ore according to the age data;
the interval prediction module is used for generating an ideal confidence interval of the talc ore depth calculated by the deep learning model according to the actual depth of the existing talc ore, generating a calculated confidence interval according to the talc ore depth calculated by the deep learning model, and comparing the ideal confidence interval with the calculated confidence interval to evaluate the deep learning model;
And the type prediction module is used for carrying out field sampling on the unexplored mine according to the calculated talcum ore depth to obtain element content data of sampling points, analyzing the element content data, judging the type of the talcum ore at the sampling points and predicting associated ores possibly existing at the sampling points.
When a small amount of scattered talcum ore is found out in a certain place, a researcher samples and analyzes geological structure data, lithology data and landform topographic data of the place, if the geological structure data, the lithology data and the landform topographic data are in accordance with the aggregation conditions of the talcum ore, a large amount of aggregated talcum ore veins can be judged to exist in the place, a mine field can be established, a small amount of the collected talcum ore is detected to obtain age data, the age data are input into a deep learning model, and therefore the depth of the talcum ore is calculated, and subsequent mining work is facilitated.
The above embodiments may be implemented in whole or in part by software, hardware, firmware, or any other combination. When implemented in software, the above-described embodiments may be implemented in whole or in part in the form of a computer program product. Those of ordinary skill in the art will appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the solution.
The units described as separate units may or may not be physically separate, and units shown as units may or may not be physical units, may be located in one place, or may be distributed over a plurality of network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
The foregoing is merely illustrative of the present application, and the present application is not limited thereto, and any person skilled in the art will readily recognize that variations or substitutions are within the scope of the present application.
Claims (7)
1. The geochemical data analysis method based on deep learning is characterized by comprising the following steps:
acquiring data of existing talcum ores, including position information, geological structure data, lithology data, stratum data, landform topographic data and age data of the areas, forming a data set of each area, marking on geological map data, numbering the existing talcum ores, forming sample tags, and enabling the data set of each area to correspond to the sample tags of the area one by one;
a deep learning model based on a convolutional neural network is constructed, data of the existing talc ore is input into the deep learning model, the average deposition rate of the position of the existing talc ore is generated, and the depth of the talc ore is calculated according to the age data;
According to the actual depth of the existing talcum ore, generating an ideal confidence interval of the talcum ore depth calculated by a deep learning model, then according to the talcum ore depth calculated by the deep learning model, generating a calculated confidence interval, and comparing the ideal confidence interval with the calculated confidence interval to evaluate the deep learning model;
According to the calculated talcum ore depth, carrying out field sampling on an unexplored ore field to obtain element content data of sampling points, analyzing the element content data, judging the type of the talcum ore at the sampling points, and predicting possible associated ores at the sampling points;
logic for calculating the depth of the talc ore based on the age data is:
Calibrating the data acquired by the sampling points, numbering the data from the ground surface to the sedimentary layers, calibrating the thickness of each sedimentary layer to be D i, and calibrating the age of each sampling point to be The depth of each sampling point relative to the ground surface is calibrated asI represents the number of layers of the deposited layer, i=1, 2,3, … …, N, i is a positive integer, j represents the number of sampling points, j=1, 2,3, … …, N, j is a positive integer;
According to the age of sampling point in certain sediment layer And a depth calibration relative to the surface of the earth is/>The average deposition rate of the deposited layer was calculated and scaled to/>The calculation formula is as follows:
wherein, Represents the depth of the (j+1) th sampling point of the (i) th deposition layer,/>Representing the age of the (j+1) th sampling point of the (i) th sediment layer;
Fitting by adopting a logarithmic function according to the duty ratio of the thickness D i of different deposition layers in the total thickness of all deposition layers to generate a weight coefficient f i of different deposition layers, wherein the formula for generating the weight coefficient f i is as follows:
wherein alpha represents a weighting coefficient, alpha is more than 0, the weighting coefficient is used for adjusting the deep learning model, and e represents the base number of natural logarithm;
According to the average deposition rate of different deposition layers And a weight coefficient f i, calculating the average deposition rate/>, of the positions of the talcum oresThe calculation formula is as follows:
the age of the existing talc ore is marked as Th, and Th and the average deposition rate of the talc ore are based on the age of the talc ore Generating a talcum ore depth Lh' calculated by a deep learning model, wherein the calculation formula is as follows:
lh' represents the depth of the talc ore calculated and generated by the deep learning model according to the position information, geological structure data, lithology data, stratum data, geomorphic topographic data, age data of the area where the talc ore is located and data acquired by the sampling points.
2. The method for analyzing geochemical data based on deep learning according to claim 1, wherein: the position information of the smooth stone ore comprises plane coordinates and depth of the smooth stone ore, the geological structure data comprise a structure diagram, a geological diagram, topographic data and stratum data of the area, the lithology data comprise rock species distribution data of quartz rock, chalcocite and rutile, the stratum data comprise data of a sedimentary layer, the topographic data comprise topographic feature data of topographic lines and contour lines for showing relief, and the age data comprise ages of the smooth stone ore.
3. The method for analyzing geochemical data based on deep learning according to claim 2, wherein: the rock species distribution data of quartzite, chalcocite and rutile comprise the content and distribution sequence of the quartzite, chalcocite and rutile, and the data of the deposition layer comprise the number of layers and the thickness of the deposition layer and the ages corresponding to different depths.
4. The method for analyzing geochemical data based on deep learning according to claim 1, wherein: the sampling mode of the existing smooth rock ore is that the central position of the plane coordinate of the existing smooth rock ore is longitudinally sampled, the number of layers of the deposited layers and the thickness of each deposited layer are obtained, n sampling points are arranged in each deposited layer, the interval distances of the n sampling points in the same deposited layer are equal, and the age of each sampling point and the depth relative to the ground surface are measured.
5. The method for analyzing geochemical data based on deep learning according to claim 1, wherein: the decision logic for determining whether the deep learning model meets the requirements is:
Dividing data of all talcum ore depths generated by a deep learning model after multiple times of calculation into x data sets according to a time sequence, calibrating the data into W x, wherein x is a positive integer, numbering the data in the data sets W x, calibrating the data into Lh y ', y is a positive integer, p data are arranged in each data set W x, W x＝{Lh1′,Lh2′,Lh3′,……,Lhp' is a positive integer, the data in the data sets W x are approximately subjected to Gaussian distribution, a 95% confidence level is obtained, the actual depth of the existing talcum ore is calibrated into Lh, the actual depth of the existing talcum ore is taken as an ideal mean value in the data sets W x, the ideal standard deviation of the data sets W x is calibrated into S x, and a calculation formula is as follows:
An ideal confidence interval for the data set W x is generated from the ideal standard deviation S x, as follows:
the t value can be found in the t value distribution table according to the confidence level and p;
The calculated average value calculated by the data set W x is calibrated as The calculation formula is as follows:
calibrating the calculated standard deviation calculated by the data set W x as S x', wherein the calculation formula is as follows:
A calculated confidence interval for the dataset W x is generated from the calculated standard deviation S x' as follows:
Comparing the ideal confidence interval of the dataset W x with the calculated confidence interval, and if and only if the calculated confidence interval falls within the ideal confidence interval, calculating an average The actual depth Lh of the ore closer to the talcum powder is calculated more accurately, and the calculation is considered to be qualified at the moment, and the judgment formula is as follows:
When q data sets W x are continuously calculated to be qualified, the deep learning model is proved to meet the requirement, the verified deep learning model can be output, otherwise, the weighting coefficients and parameters of the model are adjusted through a back propagation and optimization algorithm according to errors.
6. The method for analyzing geochemical data based on deep learning according to claim 1, wherein: the data analysis logic for the unexplored mine is:
Setting m sampling points at the same depth of talcum ore, measuring element content of each sampling point, wherein the silicon content is respectively Sulfur content is/>, respectively Iron content is/>, respectivelyWherein/> The content values of silicon, sulfur and iron at the r point are respectively represented, and the content values of silicon, sulfur and iron are subjected to the extremely-bad treatment, and the calculation formula is as follows:
In the middle of Extremely bad data of silicon, sulfur and iron are shown respectively,/>Respectively represent the maximum value of silicon, sulfur and iron contents in sampling points,/>Respectively representing the minimum values of silicon, sulfur and iron contents in the sampling points;
Calculating a range average value according to range data of silicon, sulfur and iron, and calculating a judgment coefficient according to the range average value, wherein the calculation formula of the judgment coefficient is as follows:
In the middle of The extremely poor average values of silicon, sulfur and iron are respectively shown;
The type of talc ore and possibly associated ore are judged according to the judgment coefficient theta 1、θ2、θ3, and the judgment method is as follows:
the smooth ore is rich in silicon, and the possible associated ore is quartz rock;
the smooth ore is sulfur-rich, and the possible associated ore is chalcocite;
The talc ore is rich in iron and the associated ore may be rutile.
7. A geochemical data analysis device based on deep learning, characterized in that: the data analysis device is used for realizing the geochemical data analysis method according to any one of claims 1 to 6, and comprises the following steps:
The sample construction module is used for acquiring the data of the existing talcum ores, including the position information, geological structure data, lithology data, stratum data, landform topographic data and age data of the area, forming a data set of each area, marking the geological map data, numbering the existing talcum ores, forming sample tags, and enabling the data set of each area to correspond to the sample tags of the area one by one;
The deep learning module is used for constructing a deep learning model based on a convolutional neural network, inputting the data of the existing smooth ore into the deep learning model, generating the average deposition rate of the position of the existing smooth ore, and calculating the depth of the smooth ore according to the age data;
the interval prediction module is used for generating an ideal confidence interval of the talc ore depth calculated by the deep learning model according to the actual depth of the existing talc ore, generating a calculated confidence interval according to the talc ore depth calculated by the deep learning model, and comparing the ideal confidence interval with the calculated confidence interval to evaluate the deep learning model;
And the type prediction module is used for carrying out field sampling on the unexplored mine according to the calculated talcum ore depth to obtain element content data of sampling points, analyzing the element content data, judging the type of the talcum ore at the sampling points and predicting associated ores possibly existing at the sampling points.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311219804.6A CN117238405B (en) | 2023-09-20 | 2023-09-20 | Geochemical data analysis method and device based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311219804.6A CN117238405B (en) | 2023-09-20 | 2023-09-20 | Geochemical data analysis method and device based on deep learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117238405A CN117238405A (en) | 2023-12-15 |
| CN117238405B true CN117238405B (en) | 2024-05-31 |
Family
ID=89096321
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311219804.6A Active CN117238405B (en) | 2023-09-20 | 2023-09-20 | Geochemical data analysis method and device based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117238405B (en) |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| RU2012110069A (en) * | 2012-03-15 | 2013-09-20 | Федеральное государственное бюджетное учреждение науки Институт минералогии Уральского отделения Российской академии наук | METHOD FOR DETERMINING THE EROSION SECTION LEVEL OF THE EROSION SECTION OF THE ORE-CONTAINING GEOCHEMICAL STRUCTURE BY A SPECIFIC ALGORITHM FOR CORRELATION RELATIONS AND FORMULAS OF RANGEED ELEMENTS OF ELEMENTS |
| CN110060173A (en) * | 2019-04-27 | 2019-07-26 | 烟台市牟金矿业有限公司 | A kind of deep gold deposit method of prospecting |
| CN110442666A (en) * | 2019-08-02 | 2019-11-12 | 中国地质调查局发展研究中心 | A kind of mineral resource prediction method and system based on neural network model |
| CN113344050A (en) * | 2021-05-28 | 2021-09-03 | 中国石油天然气股份有限公司 | Lithology intelligent identification method and system based on deep learning |
| RU2767159C1 (en) * | 2020-10-02 | 2022-03-16 | Акционерное общество "Сибирский научно-исследовательский институт геологии, геофизики и минерального сырья" | Method for searching for gold-ore and gold-bearing deposits by ore and geochemical associations |
| CN114203268A (en) * | 2021-12-07 | 2022-03-18 | 昆明理工大学 | Machine learning-based geochemical data classification method and device |
| CN115081685A (en) * | 2022-05-30 | 2022-09-20 | 成都理工大学 | Three-dimensional visual positioning prediction method for metal deposit deep resource |
| CN116665067A (en) * | 2023-08-01 | 2023-08-29 | 吉林大学 | Ore finding target area optimization system and method based on graph neural network |
-
2023
- 2023-09-20 CN CN202311219804.6A patent/CN117238405B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| RU2012110069A (en) * | 2012-03-15 | 2013-09-20 | Федеральное государственное бюджетное учреждение науки Институт минералогии Уральского отделения Российской академии наук | METHOD FOR DETERMINING THE EROSION SECTION LEVEL OF THE EROSION SECTION OF THE ORE-CONTAINING GEOCHEMICAL STRUCTURE BY A SPECIFIC ALGORITHM FOR CORRELATION RELATIONS AND FORMULAS OF RANGEED ELEMENTS OF ELEMENTS |
| CN110060173A (en) * | 2019-04-27 | 2019-07-26 | 烟台市牟金矿业有限公司 | A kind of deep gold deposit method of prospecting |
| CN113488117A (en) * | 2019-04-27 | 2021-10-08 | 烟台市牟金矿业有限公司 | Deep gold deposit mining and prospecting method with deep learning ability |
| CN110442666A (en) * | 2019-08-02 | 2019-11-12 | 中国地质调查局发展研究中心 | A kind of mineral resource prediction method and system based on neural network model |
| RU2767159C1 (en) * | 2020-10-02 | 2022-03-16 | Акционерное общество "Сибирский научно-исследовательский институт геологии, геофизики и минерального сырья" | Method for searching for gold-ore and gold-bearing deposits by ore and geochemical associations |
| CN113344050A (en) * | 2021-05-28 | 2021-09-03 | 中国石油天然气股份有限公司 | Lithology intelligent identification method and system based on deep learning |
| CN114203268A (en) * | 2021-12-07 | 2022-03-18 | 昆明理工大学 | Machine learning-based geochemical data classification method and device |
| CN115081685A (en) * | 2022-05-30 | 2022-09-20 | 成都理工大学 | Three-dimensional visual positioning prediction method for metal deposit deep resource |
| CN116665067A (en) * | 2023-08-01 | 2023-08-29 | 吉林大学 | Ore finding target area optimization system and method based on graph neural network |
Non-Patent Citations (1)
| Title |
|---|
| 《再论元素丰度的成矿学意义》;赵博;于蕾;张德会;张荣臻;石成龙;田颖;《桂林理工大学学报》;20150215(01期);29-39 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117238405A (en) | 2023-12-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11599790B2 (en) | Deep learning based reservoir modeling | |
| EP3989132A1 (en) | Reservoir performance system | |
| CN115879648B (en) | Ternary deep mining prediction method and system based on machine learning | |
| Chambers et al. | Petroleum geostatistics for nongeostatisticians: Part 1 | |
| CN106920176B (en) | Mining area scale mineral resource estimation method and system | |
| CN116665067B (en) | Ore finding target area optimization system and method based on graph neural network | |
| CN113267822B (en) | Method for improving submarine topography inversion precision based on topography constraint factor weight optimization | |
| CN108957554B (en) | Seismic inversion method in geophysical exploration | |
| CN104899928A (en) | Three-directional geology modeling method based on sparse borehole points | |
| CN106446514A (en) | Fuzzy theory and neural network-based well-log facies recognition method | |
| Nourani et al. | Comparison of machine learning techniques for predicting porosity of chalk | |
| CN117609848B (en) | Mineral resource classification prediction method and system based on deep learning | |
| EP3929630B1 (en) | Computer implemented method for correcting a reservoir model of a reservoir geological formation based on seismic images | |
| CN113627657A (en) | Sandstone-type uranium mineralization interest area prediction method using machine learning model | |
| CN117238405B (en) | Geochemical data analysis method and device based on deep learning | |
| CN111967677A (en) | Prediction method and device for unconventional resource dessert distribution | |
| Sarkheil et al. | The fracture network modeling in naturally fractured reservoirs using artificial neural network based on image loges and core measurements | |
| Gulbrandsen et al. | Automatic mapping of the base of aquifer—A case study from Morrill, Nebraska | |
| Spichak | Modern methods for joint analysis and inversion of geophysical data | |
| Wong | A novel technique for modeling fracture intensity: A case study from the Pinedale anticline in Wyoming | |
| CN112818603A (en) | Method, terminal and storage medium for adaptively selecting optimal mineralization prediction elements | |
| Zhang et al. | Deep Subsurface Pseudo-Lithostratigraphic Modeling Based on Three-Dimensional Convolutional Neural Network (3D CNN) Using Inversed Geophysical Properties and Shallow Subsurface Geological Model | |
| Jo et al. | Sensitivity analysis of geological rule-based subsurface model parameters on fluid flow | |
| CN118194728B (en) | Magnetotelluric physical and data hybrid driving inversion method | |
| CN118378101B (en) | Intelligent cloud management analysis system for geological survey data |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |