Rock core identification method and system based on regional geological data

Abstract

The invention discloses a rock core identification method and system based on regional geological data, and the method comprises the steps: S1, collecting a rock core sample image, and constructing a rock core image data set; s2, dividing the core image data set into a training set and a test set, and training and testing the constructed classification model based on artificial intelligence; and S3, extracting formation lithology information in the regional geological map, and adjusting and optimizing the trained classification model by using the extracted formation lithology information. When the rock core recognition model is constructed, the regional geological data is introduced to optimize the model, the initial weight assignment of the model is optimized by introducing the regional geological data, the training efficiency, the accuracy and the generalization ability of the model are improved, and the model can be applied to different geological conditions and different regions.

Description

Technical Field

[0001] The present invention relates to the technical field of core identification, and particularly to a core identification method and system based on regional geological data. Background Art

[0002] During the engineering exploration process, due to the large variety of cores, significant differences in the same type of rock, and large differences in core categories in different regions, it is difficult for engineering technicians to identify cores on site.

[0003] In order to improve the accuracy of on-site core identification by engineering technicians, artificial intelligence technology can be used to process core images to assist in core identification. The accuracy rate of the core identification system based on neural network established by some researchers can reach over 90%. However, during the research process, due to the limited sample set, the generalization ability of the model is often insufficient. The model applicable to Area A is difficult to be applicable to Area B, and the regional limitation of the model is obvious. At the same time, due to the limited types of cores used to train the model, the model does not have the ability to identify new types of cores. The existence of the above problems limits the large-scale engineering promotion and application of core image recognition.

[0004] During the engineering exploration process, exploration engineers often rely on regional geological maps to preliminarily understand the geological overview of the project, thereby establishing a preliminary understanding of the geological conditions of the site. The regional geological map outlines the stratigraphic distribution within the project scope and is an important geological data during the engineering exploration process.

[0005] Therefore, it is considered that for the core identification model based on artificial intelligence technology, if regional geological data can be introduced to optimize the model performance, the accuracy and generalization performance of the model can be significantly improved, and the model training efficiency can be increased. Furthermore, the core intelligent identification system can step out of the laboratory and have the ability of engineering application. Summary of the Invention

[0006] The present application provides a core identification method and system based on regional geological data to solve the problems such as insufficient generalization ability and obvious regional limitation existing in the existing artificial intelligence models.

[0007] According to the first aspect, in one embodiment, a core identification method based on regional geological data is provided. The method includes:

[0008] Step S1, collecting core sample images and constructing a core image data set;

[0009] Step S2, dividing the core image data set into a training set and a test set, and training and testing the constructed classification model based on artificial intelligence;

[0010] Step S3: Extract the stratigraphic lithology information from the regional geological map, and optimize the trained classification model by using the extracted stratigraphic lithology information.

[0011] Further, the step S1 specifically includes:

[0012] Step S11: Collect various types of core sample images from a wide geographical area;

[0013] Step S12: Process the collected core sample images, including orthorectification, cutting, adding labels, and unifying the size, where the added label information includes the geographical coordinate position and the name of the geotechnical material.

[0014] Further, the step S2 specifically includes:

[0015] Step S21: Divide the core image dataset into a training set and a test set according to a preset ratio;

[0016] Step S22: Construct an artificial intelligence-based classification model, including an input layer, a fully connected layer, and an output layer, with each layer including multiple neurons;

[0017] Step S23: Use the training set and the test set to train and test the artificial intelligence-based classification model respectively.

[0018] Further, the step S22 specifically includes:

[0019] Step S221: The number of neurons in the input layer depends on the size of the input core photo. If the matrix size after the core photo is converted is x*y and the color image has three RGB channels, the total number of neurons in the input layer is 3*x*y;

[0020] Step S222: The number of layers and neurons in the fully connected layer depends on the complexity of the input data. Initially set the number of layers and neurons in the fully connected layer according to the complexity of the input data, and optimize the number of neurons and layers in the fully connected layer of the model during the training process;

[0021] Step S223: The number of neurons in the output layer is equal to the number of geotechnical categories in the training set, and each output layer neuron outputs a geotechnical name.

[0022] Further, the step S3 specifically includes:

[0023] Step S31: Extract the stratigraphic lithology information at the geographical coordinate position where the core sample image i used as the model input is located from the regional geological map;

[0024] Step S32: Optimize the initialization assignment of the weights w and the bias terms b between adjacent layers of neurons in the model by using the extracted stratigraphic lithology information.

[0025] Step S33: Use the defined loss function and update the weights w and the bias term b with the specified learning rate until the changes in the weights w and the bias term b are lower than the preset threshold or the preset number of iterations is reached.

[0026] Further, the step S31 specifically includes:

[0027] Step S311: Obtain the regional geological map and perform GIS processing;

[0028] Step S312: Select an arbitrary core sample image i from the core image dataset, and obtain the formation lithology information at the geographical coordinate position where it is located from the regional geological map according to the geographical coordinate position information of the selected core sample image i;

[0029] Step S313: Perform simplification processing on the obtained formation lithology information to obtain the simplified formation and rock names.

[0030] Further, in the step S32, it specifically includes:

[0031] Assign different numerical ranges to the weights w of the corresponding neurons in the output layer according to the output results of the neurons in the output layer of the classification model to reflect their importance in the model; randomly initialize the bias term b.

[0032] Further, in the step S33, the loss function used is the cross-entropy loss function.

[0033] According to the second aspect, in an embodiment, a core identification system based on regional geological data is provided. The system includes:

[0034] A dataset construction module, configured to collect core sample images and construct a core image dataset;

[0035] A model training module, configured to divide the core image dataset into a training set and a test set, and train and test the constructed artificial intelligence-based classification model;

[0036] A model tuning module, configured to extract the formation lithology information from the regional geological map, and use the extracted formation lithology information to tune the trained classification model.

[0037] According to the third aspect, in an embodiment, an electronic device is provided. The device includes: a processor and a memory;

[0038] The memory is used to store one or more program instructions;

[0039] The processor is used to run one or more program instructions to execute the steps of a core identification method based on regional geological data as described in any one of the above.

[0040] According to a fourth aspect, in one embodiment, a computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of a core identification method based on regional geological data as described in any one of the above are implemented.

[0041] This application provides a core identification method and system based on regional geological data, having the following beneficial effects:

[0042] (1) The present invention improves the accuracy of the core identification model, can significantly improve the accuracy of field core naming, reduce exploration costs and speed up the exploration speed, thereby improving the efficiency of engineering construction; enhances the consistency and comparability of data, and reduces deviations caused by human factors;

[0043] (2) The present invention supports exploration activities under complex geological conditions and improves the generalization ability of the model in different environments; can improve the quality of decision-making and provide a more reliable scientific basis for geological exploration and resource development;

[0044] (3) The present invention can promote technological innovation, especially in the application of artificial intelligence and machine learning technologies in the field of geological exploration; improve operation safety and reduce the exposure of personnel in dangerous or harmful environments. These effects jointly promote the progress and efficiency improvement of the geological exploration industry. Description of the Drawings

[0045] Figure 1 It is a flowchart of a core identification method based on regional geological data provided by an embodiment of the present invention;

[0046] Figure 2 It is a core photo before processing in a core identification method based on regional geological data provided by an embodiment of the present invention;

[0047] Figure 3 It is a core photo after processing in a core identification method based on regional geological data provided by an embodiment of the present invention;

[0048] Figure 4 It is an example diagram of a neural network in a core identification method based on regional geological data provided by an embodiment of the present invention;

[0049] Figure 5 It is an example of stratigraphic layering in a core identification method based on regional geological data provided by an embodiment of the present invention;

[0050] Figure 6Schematic diagram of a core identification system based on regional geological data provided by an embodiment of the present invention. Detailed implementation manners

[0051] The present invention will be further described in detail below in conjunction with the accompanying drawings through specific implementation manners. Similar elements in different implementation manners are labeled with related similar element numbers. In the following implementation manners, many detailed descriptions are provided to enable a better understanding of the present application. However, those skilled in the art can easily recognize that some of the features can be omitted in different situations, or can be replaced by other elements, materials, and methods. In some cases, some operations related to the present application are not shown or described in the specification to avoid overwhelming the core part of the present application with excessive descriptions. For those skilled in the art, it is not necessary to describe these related operations in detail, and they can fully understand the related operations based on the descriptions in the specification and general technical knowledge in the art.

[0052] In addition, the features, operations, or characteristics described in the specification can be combined in any appropriate manner to form various implementation manners. At the same time, the steps or actions in the method description can also be reordered or adjusted in an obvious manner by those skilled in the art. Therefore, the various sequences in the specification and drawings are only for clearly describing a certain embodiment and do not mean a necessary sequence, unless it is stated that a certain sequence must be followed.

[0053] A core identification method based on regional geological data provided by the first embodiment of the present invention will be described in detail below in conjunction with Figure 1 and Figure 2 for detailed illustration.

[0054] As Figure 1 shown, in step S1, a core sample image is collected and a core image data set is constructed.

[0055] The above steps specifically include:

[0056] Step S11, collect various types of core sample images from a wide geographical area.

[0057] Step S12, process the collected core sample images, including orthorectification, cutting, adding labels, and unifying the size, where the added label information includes geographical coordinate positions and geotechnical names.

[0058] Specifically, the requirements for the data set are as follows: The geographical area covered should be wide and the types of cores should be complete. There should be labels, and the geographical location, geotechnical name, etc. should be marked on the labels.

[0059] 1) By collecting and gathering core photos, and performing orthorectification, cutting, and tagging on the core photos, a core photo set is obtained. All core photos are cut into a unified length size, and the length can generally be taken as 10 cm. The core photo data set should include common geotechnical categories. The core images before processing are as Figure 2 shown, and after processing are as Figure 3 shown, Figure 2 is 5 * 1 = 5 m, Figure 3 only 10 cm of it is taken as an example.

[0060] 2) Make tags for the cut core photos. The tags of the core photos include the name of the geotechnical material, geographical location (generally given by longitude and latitude), etc. For example, the tag of Figure 3 is "Geotechnical material name: silty clay, geographical location (111.02512132°, 36.52132554°)".

[0061] 3) Adjust the core photos to a unified size through matrix transformation. To facilitate the input of the model, all core photos are adjusted to a unified size, and the number of rows and columns of all core photos is the same.

[0062] As Figure 1 shown, in step S2, the core image data set is divided into a training set and a test set, and the constructed artificial intelligence-based classification model is trained and tested.

[0063] The above steps specifically include:

[0064] Step S21, divide the core image data set into a training set and a test set according to a preset ratio;

[0065] Step S22, construct an artificial intelligence-based classification model, including an input layer, a fully connected layer, and an output layer, and each layer includes multiple neurons;

[0066] Step S23, use the training set and the test set to train and test the artificial intelligence-based classification model respectively.

[0067] Specifically, the selected artificial intelligence algorithm needs to have a classification function, such as support vector machine, random forest, artificial neural network, convolutional neural network, etc. Taking a common artificial intelligence network as an example, the construction of the classification model is described.

[0068] 1) Divide the core data set into a training set and a test set. The training set generally accounts for about 70%, and the test set accounts for 30%. The training set and the test set are generally randomly divided.

[0069] 2) Construct an artificial neural network classification model. The artificial neural network classification model includes an input layer, a fully connected layer, and an output layer. Both the input layer and the output layer have only 1 layer, and the fully connected layer can have multiple layers.

[0070] The number of neurons in the input layer depends on the size of the input core photo. Generally speaking, the core photo dataset is a color image, so it has three RGB channels. If the size of the matrix after the core photo is transformed is x*y, then the total number of neurons in the input layer is 3*x*y. Each neuron in the input layer receives one of the RGB values of a pixel point in the core photo.

[0071] The number of layers and neurons in the fully connected layer depends on the complexity of the input data. The number of layers and neurons in the fully connected layer can be initially set according to the complexity of the input data, and the number of neurons and layers in the fully connected layer of the model will be optimized during the training process.

[0072] The number of neurons in the output layer is equal to the number of geotechnical categories in the training set. Each neuron in the output layer outputs a geotechnical name.

[0073] Figure 4 A simple neural network structure is given. In this figure, the input layer contains three neurons and can only receive the RGB image information of 1 pixel. The fully connected layer has 1 layer and contains 4 neurons, and the output layer contains 2 neurons, which can output two core categories. It should be noted that during the core recognition process, the number of neurons in the input layer depends on the size of the core photo and may reach thousands or even tens of thousands of neurons.

[0074] Between any two neurons in any two adjacent layers, the following related operations are performed:

[0075]

[0076] In the formula, x i represents the input data of the previous neuron, w ki represents the weight between the connections of these two neurons, b k represents the bias term, which allows the neuron to still produce an output when there is no input or the sum of the weights is zero. f(x) represents the activation function, and the softmax function can be selected as the activation function.

[0077] Taking the output layer and the previous layer as an example to illustrate the application method of the softmax function:

[0078]

[0079] In the formula, z i represents any class label in the input sample dataset, and n represents the number of neurons in the output layer. Calculate the Softmax(z) i value corresponding to each core class z i in the output, and select the core class corresponding to its maximum value as the output result, thereby realizing the classification function.

[0080] For exampleFigure 1 As shown, in step S3, the formation lithology information in the regional geological map is extracted, and the trained classification model is optimized using the extracted formation lithology information.

[0081] The above steps specifically include:

[0082] Step S31: Extract the formation lithology information at the geographical coordinate position where the core sample image i, which is used as the model input at will, is located from the regional geological map.

[0083] Among them, step S31 specifically includes:

[0084] Step S311: Obtain the regional geological map and perform GIS processing;

[0085] Step S312: Select a core sample image i from the core image dataset at will, and obtain the formation lithology information at the geographical coordinate position where it is located from the regional geological map according to the geographical coordinate position information of the selected core sample image i;

[0086] Step S313: Perform simplification processing on the obtained formation lithology information to obtain the simplified formation and rock names.

[0087] Specifically:

[0088] 1) Collect the regional geological map. Free and publicly available regional geological data can be collected, or regional geological data can be obtained through means such as borrowing. For example, obtain publicly available 1:200,000 regional geological maps and regional geological maps with smaller scales through platforms such as Geological Cloud.

[0089] 2) Use GIS technology to perform coordinate registration and vectorization on all regional geological data. According to the regional geological map mapping standard, unify the collected paper geological maps and electronic geological maps through GIS means to establish a unified regional geological database. Thus, it is possible to extract the formation information at this coordinate by inputting the coordinates.

[0090] 3) Due to the limited accuracy of regional geological data, the accuracy of the divided rock stratigraphic units reaches different levels among "group, formation, member, bed", and each rock stratigraphic unit divided contains more than one geotechnical body. Therefore, the rock mass names at the same coordinate position found on the regional geological map are often not unique.

[0091] The regional geological map is made with relatively coarse accuracy and lacks detail. In the actual work process, the work degree is often more refined. Therefore, some new geotechnical bodies may be found during the actual work process. At the same time, due to the complexity of geological conditions, the formations shown on the regional geological map may not exist on-site.

[0092] Figure 5As part of the legend of a geological map of a public area, it can be seen from Figure 5 that there are 3 lithostratigraphic units of the Qingbaikou System on the geological map of this area, namely the Meiyaogou Formation, the Yaolinghe Group, and the Yunxi Group. The rock and soil body category of the Meiyaogou Formation is unique, which is dolomite intercalated with algal dolomite; the rock and soil bodies included in the Yaolinghe Group are not unique, including epidote albite schist, chlorite quartz albite schist, and meta - keratophyre respectively; the rock and soil bodies included in the Yunxi Group are also not unique, which are quartz keratophyre, hornblende plagiogneiss intercalated with graphite quartz schist respectively.

[0093] 4) Compared with the naming of rock and soil bodies in engineering construction, the naming of rock and soil bodies on regional geological maps is often more refined. Therefore, simplify the names of the rock and soil bodies obtained in the above steps. Remove the description of mineral composition and the description of secondary components mixed in to obtain the simplified names.

[0094] Example: Dolomite intercalated with algal dolomite, after removing the intercalated components, is simplified to dolomite.

[0095] Epidote albite schist, after removing the description of mineral composition, is simplified to schist.

[0096] Chlorite quartz albite schist, after removing the description of mineral composition, is simplified to schist.

[0097] Meta - keratophyre, without the description of mineral composition and intercalated components, is not simplified.

[0098] Quartz keratophyre, after removing the description of mineral composition, is simplified to keratophyre.

[0099] Hornblende plagiogneiss intercalated with graphite quartz schist, after removing the intercalated components and mineral description, is simplified to plagiogneiss.

[0100] To improve the accuracy, generalization rate, and training efficiency of the core recognition system, and make the model applicable to different geographical regions and different geological conditions, in this embodiment, regional geological data is introduced to optimize the model during the construction of the core recognition system.

[0101] Use the stratigraphic information obtained from the regional geological data to optimize the artificial neural network model:

[0102] Step S32, use the extracted stratigraphic lithology information to optimize the initialization assignment of the weight w and bias term b between adjacent layers of neurons in the model.

[0103] In the said step S32, it specifically includes: assigning different numerical ranges to the weight w of the corresponding neurons in the output layer according to the output results of each neuron in the output layer of the classification model to reflect its importance in the model; randomly initialize the bias term b. The specific initialization process is as follows: If the name of the rock and soil in the label information of the core sample image i is d i, the set E=(e1, e2, ……, e m ) of the names of various strata and rock soils at the geographical coordinate position where the core sample image i is located, e m is the name of the rock and soil corresponding to stratum m, calculate d i ∩E = F. If the category output by a certain neuron S j in the output layer of the classification model is F, then for the neuron S j in the output layer, the weights w between all neurons connected to the previous layer are initialized and assigned values in a relatively large numerical range (0.5, 1), and the weights w between neurons other than neuron S j in the output layer and the neurons connected to the previous layer are initialized and assigned values in a relatively small numerical range (0, 0.5).

[0104] Step S33, adopt the defined loss function, and use the specified learning rate to update the weights w and the bias term b until the changes in the weights w and the bias term b are lower than the preset threshold or reach the preset number of iterations.

[0105] Specifically:

[0106] For the construction of the model loss function, the cross-entropy loss function can be selected to represent the loss of the model, and the formula is as follows:

[0107]

[0108] Among them, N is the total number of samples, C is the total number of categories, and y ij is the true label (one-hot encoding) indicating whether the i-th sample belongs to the j-th category. is the probability that the model predicts that the i-th sample belongs to the j-th category.

[0109] Assume a simple core classification problem. There are 3 input core photo samples, belonging to three different core categories, namely granite (one hot encoding is [1, 0, 0]), limestone (one hot encoding is [0, 1, 0]), and sandstone (onehot encoding is [0, 0, 1]). There are the following model output probabilities and true labels:

[0110] Sample Model Output Probability True Label (one - hot encoding) 1 [0.7,0.2,0.1] [1,0,0] 2 [0.1,0.8,0.1] [0,1,0] 3 [0.2,0.3,0.5] [0,0,1]

[0111] Loss calculation for sample 1 (granite)

[0112] L2 = -{1 + log(0.7) + 0 * log(0.2) + 0 * log(0.1)} = 0.35667494

[0113] Loss calculation for sample 2 (limestone)

[0114] L1 = - {0 * log(0.1) + 1 * log(0.8) + 0 * log(0.1)} = 0.22314355

[0115] Loss calculation for Sample 3 (sandstone):

[0116] L3 = - { * log(0.2) + 0 * log(0.3) + 1 * log(0.5)} = 0.63914718

[0117] Total loss:

[0118] L = 1 / 3 * (0.35667494 + 0.22314355 + 0.69314718) = 0.42433856

[0119] Update iteration:

[0120] Update w and b using the learning rate α, the value of α is set by the user. The updated w and b are respectively:

[0121]

[0122] By repeatedly calculating the partial derivatives and updating w and b until the changes in w and b are very small or reach the preset number of iterations.

[0123] Verify the model performance through the validation set: Apply the model trained with the training set to the validation set, and further adjust the model according to the performance feedback on the validation set. The model can be adjusted by means such as increasing or decreasing the number of fully connected layers, changing the activation function, and adjusting the learning rate, so that the model has good performance on both the training set and the test set.

[0124] Engineering application of the model: During the model application process, the geographical location information of all areas of the project is automatically obtained through positioning devices (such as smartphones, GPS, etc.). According to the location information, the formation information of the geology of the area at this point is automatically extracted from the background. The core photo is input into the model, and the model is optimized using the formation information on the regional geological map. The model then outputs the predicted geotechnical name of the core photo.

[0125] Corresponding to the above - disclosed core identification method based on regional geological data, an embodiment of the present invention also discloses a core identification system based on regional geological data, as Figure 6 shown, which specifically includes:

[0126] Dataset construction module, used to collect core sample images and construct a core image dataset;

[0127] Model training module, used to divide the core image dataset into a training set and a test set, and train and test the constructed classification model based on artificial intelligence;

[0128] A model tuning module, configured to extract formation lithology information from a regional geological map, and use the extracted formation lithology information to tune the trained classification model.

[0129] It should be noted that for the detailed description of a core identification system based on regional geological data provided in the embodiments of the present invention, reference can be made to the relevant description of a core identification method based on regional geological data provided in the embodiments of the present application, which will not be elaborated here.

[0130] In addition, an electronic device is further provided in the embodiments of the present invention. The device includes: a processor and a memory; the memory is used to store one or more program instructions; the processor is used to run one or more program instructions to execute the steps of a core identification method based on regional geological data as described in any one of the above.

[0131] It should be noted that for the detailed description of an electronic device provided in the embodiments of the present invention, reference can be made to the relevant description of a core identification method based on regional geological data provided in the embodiments of the present application, which will not be elaborated here.

[0132] In addition, a computer-readable storage medium is further provided in the embodiments of the present invention. A computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, the steps of a core identification method based on regional geological data as described in any one of the above are implemented.

[0133] It should be noted that for the detailed description of a computer-readable storage medium provided in the embodiments of the present invention, reference can be made to the relevant description of a core identification method based on regional geological data provided in the embodiments of the present application, which will not be elaborated here.

[0134] Those skilled in the art can understand that all or part of the functions of the various methods in the above embodiments can be implemented in a hardware manner or in a computer program manner. When all or part of the functions in the above embodiments are implemented in a computer program manner, the program can be stored in a computer-readable storage medium. The storage medium can include: read-only memory, random access memory, magnetic disk, optical disk, hard disk, etc. The above functions can be realized by a computer executing the program. For example, the program is stored in the memory of the device, and when the processor executes the program in the memory, the above all or part of the functions can be realized. In addition, when all or part of the functions in the above embodiments are implemented in a computer program manner, the program can also be stored in a storage medium such as a server, another computer, magnetic disk, optical disk, flash drive or mobile hard disk, and saved to the memory of the local device by downloading or copying, or the system of the local device is updated. When the processor executes the program in the memory, all or part of the functions in the above embodiments can be realized.

[0135] The above uses specific examples to elaborate on the present invention, which is only used to help understand the present invention and is not intended to limit the present invention. For those skilled in the art of the present invention, according to the idea of the present invention, several simple deductions, deformations or substitutions can also be made.

Claims

1. A core identification method based on regional geological data, characterized in that The method includes: Step S1, collecting core sample images and constructing a core image dataset; Step S2, dividing the core image dataset into a training set and a test set, and training and testing the constructed artificial intelligence-based classification model; Step S3, extracting formation lithology information from the regional geological map, and using the extracted formation lithology information to optimize the trained classification model.

2. The core identification method based on regional geological data according to claim 1, characterized in that The step S1 specifically includes: Step S11, collecting various types of core sample images from a wide geographical area; Step S12, processing the collected core sample images, including orthorectification, cutting, adding labels, and unifying the size, where the added label information includes geographical coordinate positions and geotechnical names.

3. The core identification method based on regional geological data according to claim 1, characterized in that, The step S2 specifically includes: Step S21, dividing the core image dataset into a training set and a test set according to a preset ratio; Step S22, constructing an artificial intelligence-based classification model, including an input layer, a fully connected layer, and an output layer, and each layer includes multiple neurons; Step S23, using the training set and the test set to train and test the artificial intelligence-based classification model respectively.

4. The core identification method based on regional geological data according to claim 3, wherein, The step S22 specifically includes: Step S221, determining the number of neurons in the input layer according to the size of the input core photo; Step S222, initially setting the number of layers and neurons in the fully connected layer according to the complexity of the input data, and optimizing the number of neurons and layers in the fully connected layer of the model during the training process; The number of neurons in the output layer is equal to the number of geotechnical categories in the training set, and each neuron in the output layer outputs a geotechnical name.

5. A core identification method based on regional geological data according to claim 1, characterized in that The step S3 specifically includes: Step S31, extracting the formation lithology information of the geographical coordinate position where the core sample image i used as the model input is located from the regional geological map; Step S32, using the extracted formation lithology information to optimize the initialization assignment of the weights w and bias terms b between adjacent layers of neurons in the model; Step S33, adopting the defined loss function, and using the specified learning rate to update the weights w and bias terms b until the changes in the weights w and bias terms b are lower than the preset threshold or reach the preset number of iterations.

6. The core identification method based on regional geological data according to claim 5, wherein, The step S31 specifically includes: Step S311, obtaining the regional geological map and performing GIS processing; Step S312, randomly selecting a core sample image i from the core image dataset, and obtaining the formation lithology information of the geographical coordinate position where it is located from the regional geological map according to the geographical coordinate position information of the selected core sample image i; Step S313, performing simplification processing on the obtained formation lithology information to obtain the simplified formation geotechnical names.

7. The core identification method based on regional geological data according to claim 5, characterized in that, In the step S32, it specifically includes: Assigning different numerical ranges to the weights w of the corresponding neurons in the output layer according to the output results of the neurons in the output layer of the classification model to reflect their importance in the model; randomly initializing the bias terms b.

8. A core identification method based on regional geological data according to claim 5, characterized in that, In the step S33, the adopted loss function is the cross-entropy loss function.

9. A core identification system based on regional geological data, characterized in that, The system includes: A dataset construction module, used for collecting core sample images and constructing a core image dataset; A model training module, configured to divide the core image dataset into a training set and a test set, and train and test a constructed artificial intelligence-based classification model; A model tuning module, configured to extract formation lithology information from a regional geological map, and use the extracted formation lithology information to tune the trained classification model.

10. An electronic device, characterized in that, The device includes: a processor and a memory; The memory is used to store one or more program instructions; The processor is configured to run one or more program instructions to execute the steps of a core identification method based on regional geological data according to any one of claims 1 to 8.

Images