An electrical property feature identification method and system applied to sand bodies

Abstract

The application relates to the technical field of substance or object detection, in particular to an electrical property feature identification method and system applied to a sand body, which comprises the following steps: acquiring a standard electrical property feature data set and a standard geological data set of a sand body in a target area; performing initial sand body type estimation; performing feature determination on the standard electrical property feature data set; uploading a deep learning model to the determined features to obtain a type A identification model; performing feature determination on the geological data of the target area; uploading a deep learning model to the determined features to obtain a type B identification model; fusing the type A identification model and the type B identification model to obtain an optimal identification model; performing electrical property feature identification based on the optimal identification model to obtain a sand body type optional range. The application effectively guarantees the accuracy of sand body electrical property feature identification and is suitable for sand body electrical property feature identification under high-precision and complex geological conditions.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of substance or object detection, in particular to an electrical characteristic identification method applied to a sand body, and to an identification system. BACKGROUND

[0002] In the exploration and development process of oil and gas, accurate identification of sand body characteristics plays a key role in evaluating reservoir quality. Traditional sand body characteristic identification methods, such as core analysis and geological modeling, have certain effects, but are costly and inefficient, and are difficult to effectively cover large areas. Geophysical logging technology can obtain electrical characteristics and other information of underground geological bodies. However, the existing methods for identifying sand body characteristics using electrical characteristics still have a lot of room for improvement in terms of accuracy and adaptability to complex geological conditions.

[0003] A Chinese invention patent application with publication date of May 21, 2019, publication number CN109782360A, and title of a low-resistance coverage area deep sand body detection method, discloses that in a uranium mineralization prospective area, according to the main tectonic direction, a survey line is arranged, the number of known drill holes is ≥1, at least one of the survey lines passes through one of the known drill holes, and high-precision GPS and forest compasses are used for survey point positioning; a continuous tensor observation device is used to carry out wide-frequency magnetotelluric method measurement in a station type polar arrangement; impedance calculation is performed on the collected data to obtain apparent resistivity and phase information, two-dimensional data inversion is performed, and an inversion resistivity section is drawn according to the inversion result; according to the known drill hole exposure result, the inversion resistivity section is finely layered, and the electrical characteristics of the ore-bearing sand body are extracted; according to the electrical characteristics of the sand body, artificial intelligence technologies such as neural network method are used to predict the distribution range of the deep sand body.

[0004] The foregoing technical solution solves the problems of low resolution in low-resistance coverage area depth measurement and high difficulty in sand body identification, and achieves excellent results in sand body prediction, but cannot improve the accuracy and efficiency of sand body characteristic identification. SUMMARY

[0005] The inventors have found that conventional methods often cannot effectively distinguish different types of sand bodies with similar electrical characteristics, and cannot accurately reflect the heterogeneity inside the sand body, which leads to deviations in reservoir evaluation and affects the development and implementation of oil and gas production plans.

[0006] The present application aims to provide an electrical characteristic identification method and system applied to a sand body, which obtains an optimal identification model by fusing a type A identification model and a type B identification model, to solve the technical problem that the prior art cannot provide a sand body electrical characteristic identification method with high accuracy.

[0007] According to an aspect of the present application, a method for identifying electrical characteristics of a sand body is provided, which is executed by a processor and includes: obtaining a standard electrical characteristic data set of the sand body in a target area, and obtaining a standard geological data set of the target area; performing a preliminary sand body type estimation based on the geological data set of the target area; performing feature determination on the standard electrical characteristic data set based on a supervised learning algorithm; uploading the determined features to a deep learning model to obtain a type A identification model; performing feature determination on the geological data of the target area based on a supervised learning algorithm; uploading the determined features to a deep learning model to obtain a type B identification model; fusing the type A identification model and the type B identification model to obtain an optimal identification model; performing electrical characteristic identification based on the optimal identification model, and generating a visual chart containing the sand body type; comparing the visual chart with the preliminary estimated sand body type to obtain a selectable range of the sand body type.

[0008] In some embodiments, the process of obtaining a standard electrical characteristic data set of the sand body in the target area and obtaining a standard geological data set of the target area includes: obtaining electrical characteristic data of the sand body in the target area according to a geophysical logging device; performing standardization processing on the electrical characteristic data of the sand body in the target area to obtain normalized electrical characteristic data; obtaining a geological structure data set by analyzing the change of a geophysical field according to a geophysical method, and integrating the geological structure data set according to a geographic information system; and performing an interpolation algorithm on the geological structure data set and the normalized electrical characteristic data to fill in missing data, thereby obtaining the standard electrical characteristic data set and the standard geological data set.

[0009] In some embodiments, the process of performing a preliminary sand body type estimation based on the geological data set of the target area includes: performing data cleaning on core analysis data, logging data and regional geological background information in the geological data set of the target area to obtain a standard data set; combining a statistical method with the standard data set to establish a preliminary sand body type estimation model; and uploading the sand body types in the target area that have not been determined to the estimation model to obtain a preliminary estimation result of the sand body types.

[0010] In some embodiments, the process of performing feature determination on the standard electrical characteristic data set based on a supervised learning algorithm includes: constructing a resistivity variation coefficient based on the standard electrical characteristics; performing feature selection using a random forest algorithm, calculating the advantage score of the standard electrical characteristics and the resistivity variation coefficient for the classification of sand body characteristics; removing the standard electrical characteristics and the resistivity variation coefficient with an advantage score lower than a set threshold to obtain determined features, wherein the determined features at least include the standard electrical characteristics.

[0011] In some embodiments, the process of uploading the determined features to the deep learning model to obtain the type A recognition model is: inputting the standard electrical property features of the determined features into a convolutional neural network; obtaining the difference between the prediction result of the convolutional neural network and the true label based on a loss function; and adjusting the model parameters based on a gradient descent algorithm to obtain the type A recognition model.

[0012] In some embodiments, the process of performing feature determination based on a supervised learning algorithm on the target area geological data is: performing division on the target area geological data to obtain a training set and a test set; performing training based on a random forest; and determining the most contributing features based on the training result to obtain the determined features.

[0013] In some embodiments, the process of uploading the determined features to the deep learning model to obtain the type B recognition model is: inputting the geological data of the determined features into a convolutional neural network; obtaining the difference between the prediction result of the convolutional neural network and the true label based on a loss function; and adjusting the model parameters based on a gradient descent algorithm to obtain the type B recognition model.

[0014] In some embodiments, the process of fusing the type A recognition model and the type B recognition model to obtain the optimal recognition model is: constructing a meta-model, and fusing the standard electrical property feature dataset and the standard geological dataset to obtain a fused dataset; performing division on the fused dataset to obtain a training set and a test set; performing optimization on the type A recognition model and the type B recognition model based on the training set to obtain an optimization set; inputting the optimization set into the constructed meta-model to train the meta-model; sequentially performing prediction of the type A recognition model and the type B recognition model based on the test set to obtain a prediction dataset; inputting the prediction dataset into the meta-model to perform fusion to obtain the optimal recognition model.

[0015] In some embodiments, based on the optimal recognition model, the process of performing electrical property feature recognition and generating a visual chart containing sand body types, and performing comparison based on the visual chart and the initially estimated sand body types to obtain a selectable range of sand body types is: inputting the newly collected sand body electrical property feature data of the target area into the optimal recognition model to obtain a sand body property recognition result; generating a visual chart containing sand body types based on a visualization software; and performing comparison on the initially estimated result based on the visual chart to obtain a selectable sand body type.

[0016] According to another aspect of the present application, an electrical property feature identification system applied to a sand body is provided, the system comprising a processor, and further comprising: an acquisition module, configured to acquire a standard electrical property feature data set of the sand body in a target area, and acquire a standard geological data set of the target area; a preliminary estimation module, configured to perform preliminary sand body type estimation based on the geological data set of the target area; a verification module, configured to perform feature determination on the standard electrical property feature data set based on a supervised learning algorithm; upload the determined features to a deep learning model to obtain a type A identification model; perform feature determination on the geological data of the target area based on the supervised learning algorithm; upload the determined features to the deep learning model to obtain a type B identification model; a fusion module, configured to fuse the type A identification model and the type B identification model to obtain an optimal identification model; and an execution module, configured to perform electrical property feature identification based on the optimal identification model, and generate a visual chart containing the sand body type, and perform comparison between the visual chart and the preliminary estimated sand body type to obtain a selectable range of the sand body type.

[0017] Compared with the prior art, the present application has the following advantages and beneficial effects: the present application fuses the type A identification model and the type B identification model to obtain an optimal identification model, performs electrical property feature identification based on the optimal identification model, effectively ensures the accuracy of sand body electrical property feature identification, and is different from the limitations of the model identification of the prior art in a single aspect, which cannot be adapted to sand body electrical property feature identification with high precision and complex geological conditions; at the same time, the present application performs comparison between the visual chart and the preliminary estimated sand body type to obtain a selectable range of the sand body type, realizes accurate determination of the sand body type, and is different from the case that there are many confusing options when determining the sand body type in the prior art, and the sand body type determination range of the present application is more accurate. BRIEF DESCRIPTION OF DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or prior art description will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative effort.

[0019] Figure 1 is a flow chart of the electrical property feature identification method of the present application;

[0020] Figure 2 is a structure diagram of the identification system of the present application;

[0021] Figure 3 is a visual chart of the present application;

[0022] Figure 4 is another visual chart of the present application;

[0023] Figure 5 is a scenario of using the identification system of the present application. DETAILED DESCRIPTION

[0024] The technical solutions in the embodiments of the present application will be clearly and completely described with reference to the accompanying drawings of the embodiments of the present application. Figures 1-5 It should be apparent that the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments.

[0025] Embodiment One

[0026] Figure 1 The flowchart of the method for identifying the electrical characteristics of the sand body is provided for the present embodiment, and the method is executed by a processor. The processor can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor.

[0027] The method of the present application breaks through the limitations of the prior art, improves the accuracy and efficiency of identifying the electrical characteristics of the sand body, and provides a more reliable basis for oil and gas exploration and development. The method specifically includes:

[0028] The standard electrical characteristic data set of the sand body in the target area is obtained, and the standard geological data set of the target area is obtained. In some possible implementation manners, the electrical characteristic data of the sand body in the target area is obtained according to a geophysical logging device, wherein the geophysical logging device can be a resistivity logging instrument, a spontaneous potential logging instrument, a sonic logging instrument, etc., and the present embodiment preferably uses a resistivity logging instrument. The electrical characteristic data of the sand body in the target area can be resistivity, spontaneous potential, sonic time difference, etc., and the present embodiment preferably uses resistivity. The electrical characteristic data of the sand body in the target area is subjected to standardization processing to obtain normalized electrical characteristic data, and the electrical characteristic data of the sand body in the target area collected is processed to eliminate the differences in the measurement range and units of different logging devices, so that the data has comparability, and then the electrical characteristic data is normalized to the 【0.1】 interval by using a standardization formula. The standardization formula is

[0029]

[0030] wherein, X is original data, i.e. the electrical property characteristic data of the target area sand body collected; μ is the data mean; σ is the data standard deviation.

[0031] Then, the change of the geophysical field is analyzed according to the geophysical method to obtain a geological structure data set, and the geological structure data set is integrated according to the geographic information system, wherein the geophysical method can be gravity exploration, magnetic exploration, electrical exploration, seismic exploration, etc. The underground geological structure and the properties of the geological body are inferred by analyzing the change of the geophysical field, so as to obtain information such as geological structure, stratum distribution and potential mineral resources. According to the geological structure data set, the geological structure data is integrated and managed based on the geographic information system (GIS) technology, and the geological structure data set is established.

[0032] Finally, an interpolation algorithm is performed on the geological structure data set and the normalized electrical property characteristic data to fill in the missing data, and a standard electrical property characteristic data set and a standard geological data set are obtained, that is, whether the abnormally high value in the resistivity data is a measurement error is judged in combination with the adjacent data points and the geological background. If so, the weighted average value of the adjacent data is used to replace it.

[0033] Based on the target area geological data set, a preliminary sand body type estimation is performed. According to the target area geological data set, a preliminary sand body type is obtained, which can realize the limitation of the later sand body type and effectively distinguish the wider limited range of the sand body type in the prior art.

[0034] In some possible embodiments, the core analysis data, logging data and regional geological background information in the target area geological data are subjected to data cleaning to obtain a standard data set. The core analysis data can be lithology, porosity, permeability, etc., the logging data can be resistivity, acoustic time difference, natural gamma, etc., and the regional geological background information can be stratum sedimentary environment, tectonic evolution. In this embodiment, porosity, resistivity and stratum sedimentary environment are preferred as basic data for data cleaning. Then, the standard data set is combined with a statistical method to establish a preliminary sand body type estimation model. Specifically, based on the standard data set, cluster analysis is selected to establish a sand body type estimation model. A part of data with known sand body type is used as a training set to train and optimize the sand body type estimation model, and the parameters of the model are adjusted to improve the estimation accuracy. Another part of data is used as a test set to verify and evaluate the trained model, and the accuracy, recall rate and other evaluation indexes of the model are calculated to ensure the reliability of the model. Finally, the sand body types in the target area are uploaded to the estimation model to obtain a preliminary estimation result of the sand body type.

[0035] The standard electrical property feature dataset is based on a supervised learning algorithm, feature determination is performed, a deep learning model is uploaded to the determined features, and a type A identification model is obtained. In some possible embodiments, based on the standard electrical property features, a resistivity variation coefficient is constructed, that is, the resistivity variation coefficient is defined as:

[0036]

[0037] wherein σ R is the standard deviation of the resistivity data; is the average resistivity. It should be noted that this parameter can reflect the degree of change of the resistivity and can effectively represent the heterogeneity inside the sand body.

[0038] Then, the random forest algorithm is used to perform feature selection, the advantage score of the standard electrical property features and the resistivity variation coefficient for classifying the characteristics of the sand body is calculated, the standard electrical property features and the resistivity variation coefficient with an advantage score lower than a set threshold are removed, and the determined features are obtained, wherein the determined features at least include the standard electrical property features. For example, after random forest calculation, if the advantage score of a certain acoustic feature for classifying the permeability of the sand body is lower than the set threshold, it is removed. Then, the standard electrical property features of the determined features are input into the convolutional neural network, deep features in the data are extracted through multiple convolutional layers and pooling layers, and finally classification prediction is performed through the full connection layer. Further, based on the loss function, the difference between the prediction result of the neural network and the real label is obtained, wherein the loss function of the present embodiment is the cross-entropy loss function. Finally, the model parameters are adjusted based on the gradient descent algorithm to make the loss function minimum, and the type A identification model is obtained. Optionally, in the training process, the early stopping method is used to prevent model overfitting.

[0039] The target area geological data is based on a supervised learning algorithm, feature determination is performed, and a deep learning model is uploaded to the determined features to obtain a type B identification model. In some possible embodiments, the target area geological data is divided to obtain a training set and a test set; the random forest is used for training; based on the training result, the most contributing features are determined to obtain the determined features, wherein the most contributing features are the features with the largest weight proportion of neurons in the neural network. The geological data of the determined features is input into the convolutional neural network; based on the loss function, the difference between the prediction result of the neural network and the real label is obtained; and the model parameters are adjusted based on the gradient descent algorithm to obtain the type B identification model.

[0040] The type A recognition model and the type B recognition model are fused to obtain an optimal recognition model. In some possible implementation manners, a meta-model is constructed, and a standard electrical characteristic data set and a standard geological data set are fused to obtain a fused data set; the fused data set is divided to obtain a training set and a test set; the type A recognition model and the type B recognition model are optimized based on the training set to obtain an optimization set; the meta-model is trained based on the optimization set; the type A recognition model and the type B recognition model are sequentially predicted based on the test set to obtain a prediction data set; and the prediction data set is input into the meta-model to perform fusion to obtain the optimal recognition model.

[0041] Based on the optimal recognition model, electrical characteristic recognition is performed, and a visual chart containing a sand body type is generated; and based on the visual chart and the initially estimated sand body type, comparison is performed to obtain a selectable range of the sand body type. In some possible implementation manners, the newly collected electrical characteristic data of the target area is input into the optimal recognition model to obtain a sand body characteristic recognition result, and based on a visualization software, a visual chart containing a sand body type is generated, as shown in Figure 3 、 4 The visualization software can be a whale visualization software or a cabin visualization software, and the like. Based on the visual chart, comparison is performed on the initially estimated result to obtain a selectable sand body type, where the sand body type at least includes a river sand body, a lake sand body and a marine sand body, and the selectable sand body of this embodiment is preferably a beach sand body.

[0042] Embodiment Two

[0043] Based on the same inventive concept as the electrical characteristic recognition method applied to the sand body in the foregoing embodiment one, as Figure 2As shown, the embodiment also provides an electrical property feature identification system applied to sand bodies, the system comprising a processor. The electrical property feature identification method applied to sand bodies in the foregoing embodiment can be divided into one or more modules, one or more modules are stored in the memory and executed by the processor to complete the present application. One or more modules can be a series of computer program instruction segments capable of completing a specific function, which is used to describe the execution process of the computer program. For example, the computer program can be divided into an acquisition module, an estimation module, a verification module, a fusion module, and an execution module, and the specific functions of each module are as follows: the acquisition module is used to acquire the standard electrical property feature data set of the sand bodies in the target area and acquire the standard geological data set of the target area; the estimation module is used to perform initial sand body type estimation based on the geological data set of the target area; the verification module is used to perform feature determination based on a supervised learning algorithm on the standard electrical property feature data set, upload the determined features to a deep learning model, and obtain a type A identification model; perform feature determination based on a supervised learning algorithm on the geological data of the target area, upload the determined features to a deep learning model, and obtain a type B identification model; the fusion module is used to fuse the type A identification model and the type B identification model to obtain an optimal identification model; and the execution module is used to perform electrical property feature identification based on the optimal identification model and generate a visual chart containing the sand body type, perform comparison based on the visual chart and the initial estimated sand body type, and obtain a selectable range of sand body types.

[0044] The electrical property feature identification method applied to sand bodies in the foregoing embodiment one is also applicable to the electrical property feature identification system applied to sand bodies of the present embodiment. Those skilled in the art can clearly understand the electrical property feature identification system applied to sand bodies in the present embodiment through the foregoing detailed description of the electrical property feature identification method applied to sand bodies. Therefore, for the sake of brevity of the specification, it will not be described in detail here.

[0045] The basic principles and main features of the present application and the advantages of the present application have been shown and described. It is obvious for those skilled in the art that the present application is not limited to the details of the above exemplary embodiments, and the present application can be realized in other specific forms without departing from the spirit or essential characteristics of the present application. Therefore, from any point of view, the embodiments should be regarded as exemplary and non-limiting, and the scope of the present application is defined by the appended claims rather than the above description, and therefore all changes falling within the meaning and scope of the equivalent elements of the claims are intended to be included in the present application. Any reference signs in the claims should not be regarded as limiting the claims involved.

[0046] Furthermore, it should be understood that although the specification is described in terms of embodiments, not every embodiment includes every feature or implementation described herein. The specification can include implicit combinations of explicitly mentioned features and / or implicit combinations of implicitly mentioned features. Such combinations are also expressly included within the scope of the specification and an embodiment.

Claims

1. A method for identifying the electrical characteristics of sand bodies, the method being executed by a processor, characterized in that, The application relates to a sand body type recognition method and device. The method comprises the following steps: acquiring a standard electrical characteristic data set of a sand body in a target area and acquiring a standard geological data set of the target area; performing initial sand body type estimation based on the geological data set of the target area; performing feature determination on the standard electrical characteristic data set based on a supervised learning algorithm; uploading the determined features to a deep learning model to obtain a type A recognition model; performing feature determination on the geological data of the target area based on a supervised learning algorithm; uploading the determined features to a deep learning model to obtain a type B recognition model; fusing the type A recognition model and the type B recognition model to obtain an optimal recognition model; performing electrical characteristic recognition based on the optimal recognition model and generating a visual chart containing the sand body type, and performing comparison between the visual chart and the initial estimated sand body type to obtain a selectable range of the sand body type. The process of uploading the determined features to the deep learning model to obtain the type A recognition model comprises the following steps: inputting the standard electrical characteristic of the determined features into a convolutional neural network; obtaining the difference between the prediction result of the convolutional neural network and the real label based on a loss function; and adjusting the model parameters based on a gradient descent algorithm to obtain the type A recognition model. The process of uploading the determined features to the deep learning model to obtain the type B recognition model comprises the following steps: inputting the geological data of the determined features into a convolutional neural network; obtaining the difference between the prediction result of the convolutional neural network and the real label based on a loss function; and adjusting the model parameters based on a gradient descent algorithm to obtain the type B recognition model.

2. The method of claim 1, wherein, The process of fusing the type A recognition model and the type B recognition model to obtain the optimal recognition model comprises the following steps: constructing a meta-model, fusing the standard electrical characteristic data set and the standard geological data set to obtain a fused data set, performing division on the fused data set to obtain a training set and a test set, optimizing the type A recognition model and the type B recognition model based on the training set to obtain an optimization set, training the meta-model based on the optimization set, and sequentially performing type A recognition model and type B recognition model prediction based on the test set to obtain a prediction data set, and inputting the prediction data set into the meta-model to perform fusion to obtain the optimal recognition model. The process of acquiring the standard electrical characteristic data set of the sand body in the target area and acquiring the standard geological data set of the target area comprises the following steps: acquiring the electrical characteristic data of the sand body in the target area according to a geophysical logging device; performing standardization processing on the electrical characteristic data of the sand body in the target area to obtain normalized electrical characteristic data; obtaining a geological structure data set according to the change of a geophysical field by a geophysical method, and integrating the geological structure data set according to a geographic information system; 3. The method of claim 2, wherein, performing an interpolation algorithm on the geological structure data set and the normalized electrical characteristic data to fill in missing data, and obtaining the standard electrical characteristic data set and the standard geological data set. The process of performing initial sand body type estimation based on the geological data set of the target area comprises the following steps: performing data cleaning on core analysis data, logging data and regional geological background data in the geological data set of the target area to obtain a standard data set; establishing an initial sand body type estimation model by combining a statistical method with the standard data set; uploading the sand body type in the target area which has not been determined to the estimation model to obtain an initial estimation result of the sand body type.

4. The method of claim 3, wherein, The process of determining features based on the standard electrical property characteristic data set based on a supervised learning algorithm is: Based on the standard electrical property characteristics, the coefficient of variation of resistivity is constructed; The random forest algorithm is used to perform feature selection, and the advantage score of the standard electrical property characteristics and the coefficient of variation of resistivity to the classification of sand body characteristics is calculated; Remove the standard electrical property characteristics and the coefficient of variation of resistivity whose advantage score is lower than the set threshold to obtain the determined features, wherein the determined features at least include the standard electrical property characteristics.

5. The method of claim 4, wherein, The process of determining features based on the target area geological data based on a supervised learning algorithm is: Divide the target area geological data to obtain a training set and a test set; Based on the random forest, perform training; Based on the training result, determine the most contributing features to obtain the determined features.

6. The method of claim 5, wherein, The process of performing electrical property feature recognition based on the optimal recognition model and generating a visualization chart containing sand body types, comparing the visualization chart with the initial estimated sand body types to obtain the selectable range of sand body types is: Input the newly collected sand body electrical property feature data of the target area into the optimal recognition model to obtain the sand body characteristic recognition result; Based on the visualization software, generate a visualization chart containing sand body types; Based on the visualization chart, compare the initial estimation result to obtain the selectable sand body types.

7. An electrical property identification system for use with a sand body, the system comprising a processor, wherein: Also includes: An acquisition module for acquiring a standard electrical property characteristic data set of a sand body in a target area and acquiring a target area standard geological data set; An estimation module for performing initial sand body type estimation based on the target area geological data set; A verification module for determining features based on the standard electrical property characteristic data set based on a supervised learning algorithm; uploading a deep learning model to the determined features to obtain a type A recognition model; determining features based on the target area geological data based on a supervised learning algorithm; uploading a deep learning model to the determined features to obtain a type B recognition model; the process of uploading a deep learning model to the determined features to obtain a type A recognition model is: input the standard electrical property characteristics of the determined features into a convolutional neural network; based on the loss function, obtain the difference between the convolutional neural network prediction result and the true label; based on the gradient descent algorithm, adjust the model parameters to obtain a type A recognition model; the process of uploading a deep learning model to the determined features to obtain a type B recognition model is: input the geological data of the determined features into a convolutional neural network; based on the loss function, obtain the difference between the convolutional neural network prediction result and the true label; based on the gradient descent algorithm, adjust the model parameters to obtain a type B recognition model; A fusion module for fusing the type A recognition model and the type B recognition model to obtain an optimal recognition model; the process of fusing the type A recognition model and the type B recognition model to obtain an optimal recognition model is: constructing a meta-model and fusing the standard electrical property characteristic data set and the standard geological data set to obtain a fusion data set; Divide the fusion data set to obtain a training set and a test set; based on the training set, optimize the type A recognition model and the type B recognition model to obtain an optimization set; Based on the optimization set, input the constructed meta-model into the training meta-model. Based on the test set in turn, the type A identification model and the type B identification model are predicted to obtain a prediction data set; the prediction data set is input into the meta model to perform fusion to obtain an optimal identification model; An execution module is configured to execute electrical characteristic identification based on the optimal identification model, and generate a visual chart containing a sand body type; and based on the visual chart and the initial estimated sand body type, comparison is performed to obtain a selectable range of the sand body type.

Images