Intelligent inversion parameter model establishing method and device
Through the intelligent inversion parameter model establishment method, seismic data and logging data are used for interpolation and nonlinear regression, the traditional method has solved the shortcomings in the accuracy and quality of complex tectonic regions, and achieved more accurate geological tectonic prediction and exploration risk reduction.
Patent Information
- Application Number
- CN202311673232.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2023-12-07
- Publication Date
- 2025-06-10
AI Technical Summary
The traditional inversion parameter model establishment method has insufficient accuracy and quality in complex tectonic areas, and depends on the number of drilling and the degree of exploration, making it difficult to effectively explain and predict geological structures.
The intelligent inversion parameter model establishment method is adopted. By converting seismic data into connected domain data, interpolation of logging data based on connected domain data, establishing a training data set for nonlinear regression, removing uncertain predicted parameter values, and obtaining an inversion parameter model.
The accuracy and quality of the inversion parameter model are improved, and the geological structure of complex target areas can be predicted more accurately without being constrained by well logging, reducing exploration risks.
Smart Images
Figure CN120122148A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of seismic data processing, and in particular to an intelligent inversion parameter model establishment method and device. Background Art
[0002] Intelligent interpretation technology is a key technology in seismic data processing. It analyzes, interprets and models seismic data through automated and intelligent methods to address the complexity and uncertainty in seismic data.
[0003] Intelligent inversion parameter modeling is one of the key technologies of intelligent interpretation technology. With the advent of the era of "geological big data", the potential of artificial intelligence in the field of oil and gas exploration is increasingly recognized. Compared with traditional methods that rely on expert experience and prior models, intelligent models represented by machine learning have advantages such as data mining, complex problem interpretation, and automated prediction. It is also one of the key technologies to reduce exploration risks.
[0004] The traditional inversion parameter modeling method is mainly realized by interpolation of layer and logging curve. Its accuracy is affected by the number of wells, well spacing, and exploration degree, and there is a certain irrationality. At present, there are relatively few inversion parameter modeling methods based on intelligent algorithms. In order to further improve the model accuracy of complex structural areas and improve the quality of inversion results in target areas, it is necessary to establish an intelligent inversion parameter modeling method. Summary of the invention
[0005] In view of the shortcomings of the above conventional methods, the present invention proposes an intelligent inversion parameter model establishment method, comprising:
[0006] Step 1: Convert seismic data into connected domain data;
[0007] Step 2: interpolating the well logging data based on the connected domain data;
[0008] Step 3: Establish a training data set with the interpolated logging data and the actual logging data, perform nonlinear regression of the parameter values, and obtain the predicted parameter values;
[0009] Step 4: Remove uncertain prediction parameter values and obtain the inversion parameter model.
[0010] Preferably, in the step 1, a 4-neighborhood or 8-neighborhood definition is adopted, and the seismic data is converted into connected domain data through a two-pass scanning method.
[0011] Preferably, in step 2, the well logging data is locally extrapolated along the local continuous interface of the connected domain data using a dynamic time programming method, and the well logging data is one of P-wave velocity, S-wave velocity, porosity and density.
[0012] Preferably, in the step 3, non-linear regression of parameter values is performed through a Feature Pyramid Network.
[0013] Preferably, in the step 4, Monte Carlo dropout is used to remove uncertain predicted parameter values, where the uncertain predicted parameter values refer to parameter values with a prediction accuracy lower than a predetermined threshold.
[0014] On the other hand, the present invention proposes an intelligent inversion parameter model establishment device, including:
[0015] A data conversion module for converting seismic data into connected domain data;
[0016] An interpolation module for interpolating well logging data based on the connected domain data;
[0017] A non-linear regression module for establishing a training data set with the interpolated well logging data and actual well logging data, performing non-linear regression of parameter values, and obtaining predicted parameter values;
[0018] A modeling module for removing uncertain predicted parameter values and obtaining an inversion parameter model.
[0019] Preferably, a 4-neighborhood or 8-neighborhood definition is adopted, and the seismic data is converted into connected domain data through a two-pass scanning method.
[0020] Preferably, the dynamic time warping method is used to locally extrapolate the well logging data along the local continuous boundary of the connected domain data, where the well logging data is one of longitudinal wave velocity, shear wave velocity, porosity, and density.
[0021] Preferably, non-linear regression of parameter values is performed through a Feature Pyramid Network.
[0022] Preferably, Monte Carlo dropout is used to remove uncertain predicted parameter values, where the uncertain predicted parameter values refer to parameter values with a prediction accuracy lower than a predetermined threshold.
[0023] On the other hand, the present invention provides a computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, the intelligent inversion parameter model establishment method is implemented.
[0024] On yet another aspect, the present invention provides an electronic device, including:
[0025] A memory storing executable instructions;
[0026] A processor that runs the executable instructions in the memory to implement the intelligent inversion parameter model establishment method.
[0027] The beneficial effects of the intelligent inversion parameter model establishment method of the present invention are as follows: Seismic data-driven is adopted to realize data connectivity domain processing. On this basis, a training data set is established based on well log curve interpolation for parameter modeling without considering the seismic wave wavelength, improving the ability of the inversion parameter model to indicate complex target area geological structures, and providing an important basis for reservoir prediction without well log constraints. This is of great significance for seismic structure interpretation, reservoir description and well location arrangement, as well as for reservoir prediction and subsequent processing. In addition, the present invention uses a feature pyramid network for non-linear regression of parameter values to obtain predicted parameter values, and removes uncertain predicted parameter values through the Monte Carlo dropout method, enabling a more accurate parameter model to be obtained.
[0028] The methods and apparatuses of the present invention have other characteristics and advantages that will be apparent in or will be described in detail in the accompanying drawings and the following detailed description incorporated herein. These accompanying drawings and detailed description are used together to explain the specific principles of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS
[0029] By describing the exemplary embodiments of the present invention in more detail in conjunction with the accompanying drawings, the above and other objects, features, and advantages of the present invention will become more apparent. Among them, in the exemplary embodiments of the present invention, the same reference numerals generally represent the same components.
[0030] Figure 1 FIG. shows a flowchart of an intelligent inversion parameter model establishment method according to an embodiment of the present invention.
[0031] Figure 2 FIG. shows seismic data according to an exemplary embodiment of the present invention.
[0032] Figure 3 FIG. shows a schematic diagram of well velocity interpolation according to an exemplary embodiment of the present invention.
[0033] Figure 4 FIG. shows a modeling result according to an exemplary embodiment of the present invention. DETAILED DESCRIPTION
[0034] The preferred embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although the preferred embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to make the present invention more thorough and complete, and to fully convey the scope of the present invention to those skilled in the art.
[0035] The present invention provides an intelligent inversion parameter model establishment method, including the following steps:
[0036] Step 1: Convert seismic data into connected component data;
[0037] Step 2: Interpolate well logging data based on the connected component data;
[0038] Step 3: Establish a training data set with the interpolated well logging data and the actual well logging data, perform non - linear regression of parameter values, and obtain predicted parameter values;
[0039] Step 4: Remove uncertain predicted parameter values to obtain an inversion parameter model.
[0040] The method for establishing an intelligent inversion parameter model of the present invention is based on interpolating well logging data to establish a training data set and perform parameter modeling, and can establish an effective parameter model.
[0041] Example 1
[0042] Figure 1 The flowchart of the method for establishing an intelligent inversion parameter model according to an embodiment of the present invention is shown. As Figure 1 shown, the method includes Steps 1 - 4.
[0043] Step 1: Convert seismic data into connected component data.
[0044] In Step 1, using the 4 - neighborhood and 8 - neighborhood definitions, the seismic data is converted into connected component data through a two - pass scanning method.
[0045] In seismic data, the definition methods of two adjacent pixels are 4 - neighborhood and 8 - neighborhood respectively. The 4 - neighborhood means taking the current pixel as the center position, and the four pixels above, below, left, and right of it are the 4 - neighborhood of the current pixel. That is to say, the 4 - neighborhood refers to the pixels in the four directions of up, down, left, and right of the current pixel position. The block distance between two adjacent pixels in the 4 - neighborhood is 1. For the 4 - neighborhood, the adjacent pixel points of the pixel point P0(x, y) are P1(x, y - 1), P2(x, y + 1), P3(x + 1, y), and P4(x - 1, y).
[0046] The 8-neighborhood refers to taking the current pixel as the central position, and the eight pixels above, below, left, right, top-left, top-right, bottom-left, and bottom-right are the 8-neighborhood of the current pixel. That is to say, the 8-neighborhood refers to the pixels in the eight directions of above, below, left, right, top-left, top-right, bottom-left, and bottom-right of the current pixel position. The chessboard distance between two adjacent pixels in the 8-neighborhood is 1. For the 8-neighborhood, the adjacent pixel points of the pixel point P0(x, y) are P1(x - 1, y - 1), P2(x - 1, y), P3(x - 1, y + 1), P4(x, y - 1), P5(x, y + 1), P6(x + 1, y - 1), P7(x + 1, y), and P8(x + 1, y + 1). According to different definitions of pixel neighborhoods, the obtained connected regions are also different.
[0047] The two-pass scanning method means traversing the image twice in a scanning manner. In the first traversal, the image is traversed in the order from top to bottom and from left to right, and a digital label is assigned to each non-zero element (the digital label of zero pixels is defaulted to the number 0). From the traversal order, the pixel points directly above and directly to the left of the currently visited pixel point have been assigned digital labels; when the currently visited pixel point is a non-zero pixel, there are the following four cases:
[0048] 1) The pixel points in the upper neighborhood and the left neighborhood of the currently visited pixel point are both zero, then a new digital label is assigned to the currently visited pixel point, and at the same time, the label value is recorded;
[0049] 2) The pixel point in the upper neighborhood of the currently visited pixel point is zero, and the pixel point in the left neighborhood is not zero, then the digital label of the currently visited pixel point is the same as the digital label of the left pixel point;
[0050] 3) The pixel point in the upper neighborhood of the currently visited pixel point is not zero, and the pixel point in the left neighborhood is zero, then the digital label of the currently visited pixel point is the same as the digital label of the upper neighborhood pixel point;
[0051] 4) The pixel points in the upper neighborhood and the left neighborhood of the currently visited pixel point are both not zero, then the digital label of the currently visited pixel point is the minimum value of the digital labels of the pixel points in the left neighborhood and the upper neighborhood.
[0052] After the first traversal, the same connected region may be assigned one or more digital labels. The purpose of the second traversal is to merge different digital labels belonging to the same connected region, and finally make the digital labels of all pixel points in the same connected region consistent. Therefore, before the second traversal starts, a union-find (disjoint-set) process needs to be performed on the array storing the digital labels, so that different labels in the same connected region all point to the same label. Then, the different digital labels belonging to the same connected region are merged.
[0053] Based on seismic data for image analysis and analyzing the longitudinal and transverse connectivity relationships, the conversion from seismic data to connected domain data can be achieved.
[0054] Step 2: Interpolate well logging data based on the connected domain data.
[0055] Specifically, the Dynamic Time Warping (DTW) method is used to perform local extrapolation on the well logging data along the local continuous boundary of the connected domain data to obtain the interpolated well logging data.
[0056] The Dynamic Time Warping method is a technique for time series pattern matching that can handle time series of different lengths and rates. In DTW, the pattern matching of two time series is completed by calculating the distance between them. This distance is calculated by comparing the corresponding elements in the two sequences point by point and accumulating the errors. To handle time series of different lengths and rates, DTW adopts a dynamic time warping method, that is, allowing local stretching or compression of the time series during the matching process.
[0057] Specifically, the goal of DTW is to find a time warping function that minimizes the cumulative error between two time series. This function describes the time correspondence between the input sequence and the reference sequence and satisfies certain conditions, such as being monotonically increasing and non - negative. Through dynamic programming techniques, DTW can effectively solve this problem and find the shortest distance between the two sequences.
[0058] In this embodiment, DTW is used to align two signals s 1 (z) and s 2 (z), which can be used to estimate the relative deviation between the two signals (such as time difference in the time - domain signal). Specifically, it includes the following 3 steps:
[0059] The first step is to establish an error matrix e(Δz,z), and the elements in this matrix represent the Euclidean (L2) distance between the signals s1(z) and s2(z + Δz).
[0060] e(Δz,z) = ||s 1 (z)-s 2 (z + Δz)|| 2 2
[0061] The second step is to search and establish an accumulation matrix d(Δz,z). The elements in this matrix are the optimal accumulation distances searched from the previous sample point s 1 (z - 1). This search process can be expressed as:
[0062] d(Δz,z) = e(Δz,z), if z = 1
[0063] If z = 2, 3, …, ns 1
[0064] This formula indicates that the elements in this accumulation matrix can be considered as the optimal distance from sample point s 1 (z) to sample point s 2 (z + Δz).
[0065] In the third step, the relative deviation u(z) between the two signals is obtained by backtracking the accumulation matrix d, and the final accumulation matrix is obtained based on the minimum value of the relative deviation u(z). The correspondence between the connected domain data and the logging curve labels is confirmed to achieve local extrapolation.
[0066] The logging data can be the longitudinal wave velocity, shear wave velocity, porosity, density, etc.
[0067] Step 3: Establish a training data set with the interpolated logging data and the actual logging data, perform non-linear regression on the parameter values, and obtain the predicted parameter values.
[0068] Using the actual logging data as the data label, a training data set can thus be established with the interpolated logging data and the actual logging data, perform non-linear regression on the parameter values, and obtain the predicted parameter values.
[0069] In this embodiment, non-linear regression of the parameter values is performed through a Feature Pyramid Network (FPN).
[0070] The FPN network mainly solves the multi-scale problem in object detection. By simply changing the network connection, the performance of small object detection is greatly improved without significantly increasing the computational complexity of the original model. Through upsampling the high-level features and top-down connection of the low-level features, and predictions are made at each layer. That is, the problem of low semantic information level of the low-level features is solved through the top-down process and horizontal connection. While maintaining the advantage of the low-level features for detecting small targets, the detection accuracy is improved. The specific process is as follows:
[0071] (1) The convolutional network obtains feature layers 1, 2, 3
[0072] (2) Copy feature layer 3 as the highest feature layer 4 for top-down
[0073] (3) Upsample feature layer 4 to obtain feature layer 4', ensuring its size is the same as that of feature layer 2. At the same time, use 1*1 conv on feature layer 2 to obtain feature layer 4”, ensuring its number of channels is the same as that of feature layer 4'. Finally, add feature layer 4' and feature layer 4” to obtain feature layer 5 (in order to eliminate the aliasing effect of upsampling, actually a 3*3 convolutional kernel will be used to convolve each fusion result again)
[0074] (4) Repeat the above step to obtain low-level features in sequence.
[0075] Step 4: Remove the uncertain predicted parameter values to obtain a parameter model
[0076] Specifically, the Monte Carlo dropout method is used to remove the uncertain predicted parameter values. The uncertain predicted parameter values refer to the parameter values with a prediction accuracy lower than a predetermined threshold. The Monte Carlo dropout method is a regularization method based on the neural network structure. It randomly discards some neurons, thereby reducing the coupling between neurons and preventing overfitting. During the training process, the Monte Carlo dropout method will randomly set the outputs of some neurons to 0 in each training iteration, so that these neurons do not participate in the calculation in that iteration. The effect of this is that the model of each iteration is different, so overfitting can be avoided. By removing the uncertain predicted parameter values through the Monte Carlo dropout method, the stability of the method can be improved.
[0077] Example 2
[0078] Embodiment 2 provides an intelligent inversion parameter model establishment method, including the following steps:
[0079] Step 1: Convert seismic data into connected domain data;
[0080] Step 2: Interpolate well logging data based on the connected domain data;
[0081] Step 3: Establish a training dataset with the interpolated well logging data and the actual well logging data, perform non-linear regression of parameter values, and obtain predicted parameter values;
[0082] Step 4: Remove the uncertain predicted parameter values to obtain a parameter model.
[0083] In this embodiment, in Step 1, the 4-neighborhood or 8-neighborhood definition is adopted, and the seismic data is converted into connected domain data through a two-pass scanning method.
[0084] In this embodiment, in Step 2, the dynamic time warping method is used to perform local extrapolation on the well logging data along the local continuous boundary of the connected domain data.
[0085] In this embodiment, in Step 3, non-linear regression of parameter values is performed through a feature pyramid network.
[0086] In this embodiment, in Step 4, the Monte Carlo dropout method is used to remove the uncertain predicted parameter values. The uncertain predicted parameter values refer to the parameter values with a prediction accuracy lower than a predetermined threshold.
[0087] In this embodiment, the well logging data is well velocity, the parameter value is velocity value, and the inversion parameter model is a velocity model.
[0088] In this embodiment, an intelligent inversion parameter model establishment method is used to model the migrated seismic image, and the recognition result is as Figure 2 shown. The dynamic time warping method is used to locally extrapolate the well velocity along the local continuous boundary of the connected domain data, and the result is as Figure 3 shown. Figure 4 The modeling result of this embodiment is shown. Compared with the reading of the relative comparison software, the modeling result has a high degree of coincidence with the actual logging data. This embodiment upgrades the migrated model where full waveform inversion (FWI) does not converge to an FWI model that can converge, and the effect is simple and practical.
[0089] For other detailed descriptions of this exemplary embodiment, reference can be made to the corresponding descriptions in the foregoing embodiments, which will not be elaborated herein.
[0090] Example 3
[0091] This embodiment provides an intelligent inversion parameter model establishment device, including:
[0092] A data conversion module for converting seismic data into connected domain data;
[0093] An interpolation module for interpolating logging data based on the connected domain data;
[0094] A non-linear regression module for establishing a training data set with the interpolated logging data and the actual logging data, performing non-linear regression of parameter values, and obtaining predicted parameter values;
[0095] A modeling module for removing uncertain predicted parameter values and obtaining a parameter model.
[0096] In this embodiment, a 4-neighborhood or 8-neighborhood definition is adopted, and the seismic data is converted into connected domain data by a two-pass scanning method.
[0097] In this embodiment, the dynamic time warping method is used to locally extrapolate the logging data along the local continuous boundary of the connected domain data.
[0098] In this embodiment, non-linear regression of parameter values is performed through a feature pyramid network.
[0099] In this embodiment, uncertain predicted parameter values are removed by the Monte Carlo dropout method, and the uncertain predicted parameter values refer to parameter values with a prediction accuracy lower than a predetermined threshold.
[0100] For other detailed descriptions of this exemplary embodiment, reference can be made to the corresponding descriptions in the foregoing embodiments, which will not be elaborated herein.
[0101] Example 4
[0102] This embodiment provides a computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, the foregoing method for establishing an intelligent inversion parameter model is implemented.
[0103] A computer-readable storage medium may be a tangible device that can hold and store instructions for use by an instruction execution device. A computer-readable storage medium may be, for example, but not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of the computer-readable storage medium include: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a static random access memory (SRAM), a portable compact disc read-only memory (CD-ROM), a digital versatile disc (DVD), a memory stick, a floppy disk, a mechanical encoding device, such as a punched card or raised structures in a groove having instructions stored thereon, and any suitable combination of the foregoing. The computer-readable storage medium used herein is not construed as an instantaneous signal itself, such as a radio wave or other freely propagating electromagnetic wave, an electromagnetic wave propagated through a waveguide or other transmission medium (e.g., an optical pulse through an optical fiber cable), or an electrical signal transmitted through a wire.
[0104] For other detailed descriptions of this exemplary embodiment, reference may be made to the corresponding descriptions in the foregoing embodiments, which will not be elaborated herein.
[0105] Example 5
[0106] This embodiment provides an electronic device, including:
[0107] A memory storing executable instructions;
[0108] A processor that runs the executable instructions in the memory to implement the foregoing method for establishing an intelligent inversion parameter model.
[0109] The computer-readable program instructions described herein may be downloaded from a computer-readable storage medium to various computing / processing devices, or downloaded to an external computer or external storage device through a network, such as the Internet, a local area network, a wide area network, and / or a wireless network. The network may include a copper transmission cable, an optical fiber transmission, a wireless transmission, a router, a firewall, a switch, a gateway computer, and / or an edge server. A network adapter or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards the computer-readable program instructions for storage in the computer-readable storage medium in each computing / processing device.
[0110] The computer program instructions for performing the operations of the present disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine - related instructions, microcode, firmware instructions, state - setting data, or source code or object code written in any combination of one or more programming languages, including object - oriented programming languages such as Smalltalk, C++, etc., and conventional procedural programming languages such as the "C" language or similar programming languages. The computer - readable program instructions may be executed entirely on the user's computer, partially on the user's computer, executed as a stand - alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or, alternatively, may be connected to an external computer (e.g., via the Internet service provider through the Internet). In some embodiments, by using the state information of the computer - readable program instructions to customize an electronic circuit, such as a programmable logic circuit, a field - programmable gate array (FPGA), or a programmable logic array (PLA), the electronic circuit can execute the computer - readable program instructions to implement various aspects of the present disclosure.
[0111] Aspects of the present disclosure are described herein with reference to the flowcharts and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the present disclosure. It should be understood that each block of the flowcharts and / or block diagrams, and combinations of blocks in the flowcharts and / or block diagrams, can be implemented by computer - readable program instructions.
[0112] These computer - readable program instructions can be provided to a processor of a general - purpose computer, a special - purpose computer, or other programmable data - processing apparatus to produce a machine such that the instructions, when executed by the processor of the computer or other programmable data - processing apparatus, create a means for implementing the functions / acts specified in one or more blocks of the flowchart and / or block diagram. These computer - readable program instructions can also be stored in a computer - readable storage medium, which causes a computer, a programmable data - processing apparatus, and / or other devices to operate in a particular manner, so that the computer - readable medium storing the instructions includes a manufacture, which includes instructions for implementing various aspects of the functions / acts specified in one or more blocks of the flowchart and / or block diagram.
[0113] Computer-readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other devices, causing a series of operational steps to be performed on the computer, other programmable data processing apparatus, or other devices to generate a computer-implemented process such that the instructions executed on the computer, other programmable data processing apparatus, or other devices implement the functions / acts specified in one or more blocks of the flowchart and / or block diagram.
[0114] For other detailed descriptions of this exemplary embodiment, reference may be made to the corresponding descriptions in the foregoing embodiments, which will not be repeated here.
[0115] The embodiments of the present invention have been described above. The above description is exemplary and not exhaustive, and is not limited to the disclosed embodiments. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The choice of terms used herein is intended to best explain the principles of the embodiments, practical applications, or technical improvements in the technology in the market, or to enable other ordinary skill in the art to understand the embodiments disclosed herein.
Claims
1. An intelligent inversion parameter model establishment method, characterized in that, it includes: Step 1: Convert seismic data into connected domain data; Step 2: Interpolate well logging data based on the connected domain data; Step 3: Establish a training dataset with the interpolated well logging data and actual well logging data, perform non-linear regression of parameter values, and obtain predicted parameter values; Step 4: Remove uncertain predicted parameter values to obtain an inversion parameter model.
2. The intelligent inversion parameter model establishment method according to claim 1, characterized in that, in Step 1, using 4-neighborhood or 8-neighborhood definition, convert the seismic data into connected domain data by a two-pass scanning method.
3. The intelligent inversion parameter model establishment method according to claim 1, characterized in that, in Step 2, use the dynamic time warping method to perform local extrapolation on the well logging data along the local continuous boundary of the connected domain data, and the well logging data is one of longitudinal wave velocity, transverse wave velocity, porosity, and density.
4. The intelligent inversion parameter model establishment method according to claim 1, characterized in that, in Step 3, perform non-linear regression of parameter values through a feature pyramid network.
5. The intelligent inversion parameter model establishment method according to claim 1, characterized in that, in Step 4, remove uncertain predicted parameter values through the Monte Carlo dropout method, and the uncertain predicted parameter values refer to parameter values with a prediction accuracy lower than a predetermined threshold.
6. An intelligent inversion parameter model establishment device, characterized in that, it includes: A data conversion module for converting seismic data into connected domain data; An interpolation module for interpolating well logging data based on the connected domain data; A non-linear regression module for establishing a training dataset with the interpolated well logging data and actual well logging data, performing non-linear regression of parameter values, and obtaining predicted parameter values; A modeling module for removing uncertain predicted parameter values to obtain an inversion parameter model.
7. The intelligent inversion parameter model establishment data device according to claim 6, characterized in that, using 4-neighborhood or 8-neighborhood definition, convert the seismic data into connected domain data by a two-pass scanning method.
8. The intelligent inversion parameter model establishment device according to claim 6, characterized in that, use the dynamic time warping method to perform local extrapolation on the well logging data along the local continuous boundary of the connected domain data, and the well logging data is one of longitudinal wave velocity, transverse wave velocity, porosity, and density.
9. The intelligent inversion parameter model establishment device according to claim 6, characterized in that, perform non-linear regression of parameter values through a feature pyramid network.
10. The intelligent inversion parameter model establishment device according to claim 6, characterized in that, remove uncertain predicted parameter values through the Monte Carlo dropout method, and the uncertain predicted parameter values refer to parameter values with a prediction accuracy lower than a predetermined threshold.
11. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, it implements the method for establishing an intelligent inversion parameter model according to any one of claims 1-5.
12. An electronic device, characterized in that, the electronic device includes: a memory storing executable instructions; a processor that runs the executable instructions in the memory to implement the method for establishing an intelligent inversion parameter model according to any one of claims 1-5.