A method for formation identification based on a multi-layer perceptron

Abstract

The application provides a stratum identification method based on a multilayer perceptron, relates to the technical field of oil drilling engineering, and comprises the following steps: for original data, data preprocessing based on a pipeline mechanism is performed to obtain preprocessed data; and the preprocessed data is input into a deep learning neural network model based on a multilayer perceptron to obtain a stratum identification result. The application pre-processes data through a pipeline mechanism, and establishes a deep learning model of a multilayer perceptron, so that the physical properties of underground rocks and soil are analyzed to realize rapid and accurate identification of strata.

Description

Technical Field

[0001] This invention relates to the field of oil drilling engineering technology, and more specifically, to a formation identification method based on a multilayer perceptron. Background Technology

[0002] In geophysics and geology, stratigraphic identification classifies and identifies subsurface strata by measuring and analyzing the physical properties of underground rocks and soils. These physical properties include resistivity, electrical conductivity, and dielectric constant, which can be measured using geophysical exploration equipment such as electrical resistivity, seismic methods, and well logging.

[0003] However, existing stratigraphic identification methods typically rely on manual interpretation and experience-based judgment, which not only requires extensive professional knowledge and experience but also makes it difficult to achieve rapid and accurate stratigraphic identification. Therefore, how to achieve automated and intelligent stratigraphic identification and improve its efficiency and accuracy is one of the urgent problems to be solved in the fields of geophysics and geology.

[0004] To address the problems of existing technologies, this invention provides a stratum identification method based on a multilayer perceptron. Summary of the Invention

[0005] To address the problems of existing technologies, this invention aims to preprocess data through a pipeline mechanism and establish a deep learning model of a multilayer perceptron, thereby achieving rapid and accurate identification of strata through the analysis of the physical properties of underground rocks and soil.

[0006] This invention provides a method for stratigraphic identification based on a multilayer perceptron, the method comprising:

[0007] For the raw data, perform data preprocessing based on a pipeline mechanism to obtain preprocessed data;

[0008] The preprocessed data is input into a deep learning neural network model based on a multilayer perceptron to obtain the formation identification results.

[0009] According to an embodiment of the present invention, the preprocessed data is obtained through the following steps: based on a pipeline mechanism, three operations—filling in missing data values, removing useless data, and converting stratigraphic category labels—are integrated into one pipeline and executed sequentially to obtain the preprocessed data.

[0010] According to one embodiment of the present invention, the missing data values ​​are filled by the following steps: filling the missing values ​​that may occur in the original data with the average value of the remaining data to obtain filled data.

[0011] According to one embodiment of the present invention, the removal of useless data is performed by the following steps: for the filling data, unnecessary data is removed to obtain the removed data, wherein the unnecessary data includes irrelevant feature columns or erroneous data rows.

[0012] According to an embodiment of the present invention, the stratigraphic category label conversion is performed through the following steps: for the removed data, the stratigraphic category column of the last column is converted into a uniquely hot code to obtain the preprocessed data.

[0013] According to an embodiment of the present invention, the deep learning neural network model includes: an input layer, a plurality of hidden layers and an output layer, wherein the plurality of hidden layers and the output layer are fully connected layers.

[0014] According to one embodiment of the present invention, the deep learning neural network model comprises:

[0015]

[0016] Where: H represents the hidden layer, H∈R n×h R represents a real number; n represents the number of input samples; h represents the number of hidden units; σ represents the activation function; X represents the input layer, X∈R n×d ;d represents the number of input features for each input sample; W (1) W represents the hidden layer weights. (1) ∈R d×h b (1) Indicates the hidden layer bias, b (1) ∈R 1×h W (2) W represents the output layer weights. (2) ∈R h×q b (2) Indicates the output layer bias, b (2) ∈R 1×q O represents the output layer, O∈R n×q ; q represents the number of output features of the output layer.

[0017] According to one embodiment of the present invention, the activation function is a ReLU function.

[0018] According to another aspect of the invention, a storage medium is also provided, which includes instructions for performing the methods described in any of the preceding claims.

[0019] According to another aspect of the present invention, a formation identification apparatus based on a multilayer perceptron is also provided, which performs the method as described in any of the preceding claims, the apparatus comprising:

[0020] The preprocessing module performs pipeline-based data preprocessing on the raw data to obtain preprocessed data.

[0021] The stratigraphic identification module inputs the preprocessed data into a deep learning neural network model based on a multilayer perceptron to obtain stratigraphic identification results.

[0022] This invention provides a formation identification method based on a multilayer perceptron, which has the following advantages compared with existing technologies:

[0023] 1) This invention preprocesses the raw data through a pipeline mechanism, filling in missing data values, removing unnecessary data, and converting them into one-hot encodings, all of which are implemented one by one on a pipeline, thereby improving the usefulness of the data and making it more efficient and easier to use.

[0024] 2) Deep learning neural network models based on multilayer perceptrons possess powerful representational capabilities: Multilayer perceptrons can learn complex features of data and combine low-level features into high-level feature representations through a layer-by-layer pass-through method. This representational capability enables multilayer perceptrons to handle more complex tasks.

[0025] 3) Deep learning neural network models based on multilayer perceptrons possess nonlinear mapping capabilities: Multilayer perceptrons have the ability to map input data to a high-dimensional space, thereby better fitting the data. This is very useful for handling nonlinear problems.

[0026] 4) Deep learning neural network models based on multilayer perceptrons possess parallel computing capabilities: The structure of multilayer perceptrons makes them highly suitable for parallel computing. During training, the parameters of all layers can be updated simultaneously, thereby improving training speed.

[0027] 5) Deep learning neural network models based on multilayer perceptrons are easy to understand and debug: The structure of multilayer perceptrons is relatively simple, making them easy to understand and debug. This has led to their widespread application in many fields.

[0028] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the description, claims, and drawings. Attached Figure Description

[0029] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with the embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0030] Figure 1 A flowchart illustrating the steps of a formation identification method based on a multilayer perceptron according to an embodiment of the present invention is shown.

[0031] Figure 2 A technical roadmap according to an embodiment of the present invention is shown;

[0032] Figure 3 A schematic diagram of a data preprocessing procedure according to an embodiment of the present invention is shown;

[0033] Figure 4 A schematic diagram of a single-layer perceptron model according to an embodiment of the present invention is shown;

[0034] Figure 5 An implementation roadmap according to an embodiment of the present invention is shown.

[0035] In the accompanying drawings, the same parts use the same reference numerals. Also, the drawings are not drawn to scale. Detailed Implementation

[0036] To make the objectives, technical solutions, and advantages of the present invention clearer, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0037] In geophysics and geology, stratigraphic identification classifies and identifies subsurface strata by measuring and analyzing the physical properties of underground rocks and soils. These physical properties include resistivity, electrical conductivity, and dielectric constant, which can be measured using geophysical exploration equipment such as electrical resistivity, seismic methods, and well logging.

[0038] However, existing stratigraphic identification methods typically rely on manual interpretation and experience-based judgment, which not only requires extensive professional knowledge and experience but also makes it difficult to achieve rapid and accurate stratigraphic identification. Therefore, how to achieve automated and intelligent stratigraphic identification and improve its efficiency and accuracy is one of the urgent problems to be solved in the fields of geophysics and geology.

[0039] To address the problems of existing technologies, this invention aims to preprocess data through a pipeline mechanism and establish a deep learning model of a multilayer perceptron, thereby achieving rapid and accurate identification of strata through the analysis of the physical properties of underground rocks and soil.

[0040] Figure 1 A flowchart illustrating the steps of a multilayer perceptron-based stratum identification method according to an embodiment of the present invention is shown.

[0041] like Figure 1 As shown, in step S1, data preprocessing based on a pipeline mechanism is performed on the original data to obtain preprocessed data.

[0042] In one embodiment, preprocessed data is obtained through the following steps: A pipeline mechanism integrates three operations—data missing value imputation, useless data removal, and stratigraphic category label conversion—into a single pipeline, which are then executed sequentially to obtain the preprocessed data. Specifically, the pipeline-based data preprocessing method integrates the three operations of data missing value imputation, unnecessary column removal, and stratigraphic category label conversion into one-hot encoding into a single pipeline. This is similar to abstracting the above three steps into a multi-step pipeline workflow. For those using machine learning, pipelined machine learning is more efficient and easier to use than modeling individual steps independently.

[0043] like Figure 1 As shown, in step S2, the preprocessed data is input into a deep learning neural network model based on a multilayer perceptron to obtain the formation identification result.

[0044] In one embodiment, a deep learning neural network model includes an input layer, multiple hidden layers, and an output layer, wherein the multiple hidden layers and the output layer are fully connected layers. Specifically, in a deep learning neural network model based on a multilayer perceptron: after data preprocessing, a deep learning model using a multilayer perceptron is used to train the model; since ordinary linear models and simple preprocessing cannot meet the requirements for model training, this invention overcomes the limitations of linear models by adding one or more hidden layers to the model, enabling it to handle more general types of functional relationships.

[0045] Figure 2 A technology roadmap according to an embodiment of the present invention is shown.

[0046] like Figure 2 As shown, a pipeline-based data preprocessing method is used for preprocessing; then, a deep learning neural network model based on a multilayer perceptron is used to model and train the processed data in order to obtain the stratigraphic identification results.

[0047] like Figure 2 As shown, the data preprocessing method based on the pipeline mechanism integrates the three cases in data preprocessing into a pipeline. That is, the operations of filling missing data values, removing unnecessary data, and changing the formation type to unique thermal encoding are placed on a pipeline and executed sequentially.

[0048] like Figure 2As shown, a deep learning neural network model based on a multilayer perceptron is presented. The core idea of ​​a multilayer perceptron is to overcome the limitations of linear models by adding one or more hidden layers, enabling it to handle more general types of functional relationships. Specifically, many fully connected layers are stacked together, with each layer outputting to the layers above it, until the final output is generated. The first L-1 layers can be viewed as representations, and the last layer as a linear predictor. This architecture is called a multilayer perceptron, abbreviated as MLP.

[0049] Figure 3 A schematic diagram of a data preprocessing procedure according to an embodiment of the present invention is shown.

[0050] like Figure 3 As shown, missing data imputation is performed through the following steps: fill in any missing values ​​that may occur in the original data with the average of the remaining data to obtain the imputed data.

[0051] like Figure 3 As shown, the following steps are used to remove useless data: For the populated data, unnecessary data is removed to obtain the removed data. The unnecessary data contains irrelevant feature columns or erroneous data rows.

[0052] like Figure 3 As shown, the stratigraphic category label conversion is performed through the following steps: For the removed data, the stratigraphic category column in the last column is converted into one-hot encoding to obtain preprocessed data.

[0053] This invention preprocesses the raw data through a pipeline mechanism, filling in missing data values, removing unnecessary data, and converting to one-hot encoding, all in a pipeline to achieve this step by step. This improves the usefulness of the data and makes it more efficient and easier to use.

[0054] Figure 4 A schematic diagram of a single-layer perceptron model according to an embodiment of the present invention is shown.

[0055] like Figure 4 As shown, suppose a multilayer perceptron (deep learning neural network model) has 3 inputs (x1, x2, x3) and 2 outputs (o1, o2), and its hidden layer contains 4 hidden units (h1, h2, h3, h4).

[0056] The input layer does not involve any computation, so generating output using this network only requires implementing the computation of the hidden and output layers, such as... Figure 4 The multilayer perceptron shown has two layers; it should also be noted that both layers are fully connected, and each input affects every neuron in the hidden layer, while each neuron in the hidden layer affects every neuron in the output layer.

[0057] First, through matrix X∈Rn×d Let represent a mini-batch of n samples, where each sample has d input features.

[0058] For a single-hidden-layer multilayer perceptron with h hidden units, let H∈R n×h The output of the hidden layer is called the hidden representation; in mathematics or code, H is also called the hidden layer variable or hidden variable.

[0059] Since both the hidden layer and the output layer are fully connected, the weight W of the hidden layer is... (1) ∈R d×h and hidden layer bias b( 1 )∈R 1×h and output layer weights W( 2 )∈R h×q and output layer bias b( 2 )∈R 1×q .

[0060] Formally, the output O∈R of a single hidden layer multilayer perceptron is calculated as follows: n×q :

[0061]

[0062] However, at this point, compared to a linear model, this single-hidden-layer model offers no advantage. The reason is that the hidden units are given by the affine functions of the input, while the output is simply the affine function of the hidden units. Since the affine function of an affine function is itself an affine function, it remains a linear model.

[0063] This equivalence can be proven: for any weight value, simply merging the hidden layers will produce a layer with parameters W = W. (1) W (2) and b = b (1) W (1) +b (2) Equivalent single-layer model:

[0064] O=(XW (1) +b (1) W (2) +b (2) =XW (1) W (2) +b (1) W (1) +b (2) =XW+b

[0065] At this point, in order to realize the potential of the multi-layer architecture, an additional key element is needed: applying a nonlinear activation function σ to each hidden unit after the affine transformation. The output of the activation function is called the activity value.

[0066] Generally speaking, with an activation function, it is no longer possible to degenerate a multilayer perceptron into a linear model. Therefore, the deep learning neural network model proposed in this invention includes:

[0067]

[0068] Where: H represents the hidden layer, H∈R n×h R represents a real number; n represents the number of input samples; h represents the number of hidden units; σ represents the activation function; X represents the input layer, X∈R n×d ;d represents the number of input features for each input sample; W (1) W represents the hidden layer weights. 1 )∈R d×h b (1) Indicates the hidden layer bias, b( 1 )∈R 1×h W (2) W represents the output layer weights. (2) ∈R h×q b (2) Indicates the output layer bias, b (2) ∈R 1×q O represents the output layer, O∈R n×q ; q represents the number of output features of the output layer.

[0069] Activation functions determine whether a neuron should be activated by calculating a weighted sum plus a bias. They convert input signals into differentiable outputs, and most activation functions are nonlinear.

[0070] In one embodiment, the activation function is the ReLU function. Specifically, the rectified linear unit, i.e., the ReLU function, is used because it is simple to implement and performs well in various prediction tasks.

[0071] Given an element x, the ReLU function is defined as the maximum value of that element and 0:

[0072] ReLU(x) = max{x, 0}

[0073] In other words, the ReLU function retains only positive elements and discards all negative elements by setting the corresponding activity value to 0.

[0074] This invention uses a deep learning neural network model based on a multilayer perceptron, which has the following four advantages:

[0075] I. Powerful Representation Capabilities: Multilayer perceptrons can learn complex features of data and combine lower-level features into higher-level feature representations through a layer-by-layer transfer mechanism. This representation capability enables multilayer perceptrons to handle more complex tasks.

[0076] II. Nonlinear Mapping Capability: Multilayer perceptrons possess nonlinear mapping capabilities, enabling them to map input data to a higher-dimensional space, thereby achieving a better fit to the data. This is extremely useful for handling nonlinear problems.

[0077] III. Parallel Computing Capability: The structure of the multilayer perceptron makes it highly suitable for parallel computing. During training, the parameters of all layers can be updated simultaneously, thereby improving training speed.

[0078] Fourth, it is easy to understand and debug: The structure of the multilayer perceptron is relatively simple, making it easy to understand and debug. This has led to its widespread application in many fields.

[0079] Figure 5 An implementation roadmap according to an embodiment of the present invention is shown.

[0080] like Figure 5 As shown, a formation drilling dataset was selected to collect data.

[0081] like Figure 5 As shown, a pipeline mechanism is used to preprocess the original dataset: First, missing values ​​are imputed by averaging the remaining data. Next, unnecessary data is removed, including rows or columns such as irrelevant feature columns or erroneous rows. Finally, the last column of the original data, the stratigraphic category column, is converted into one-hot encoding. The pipeline mechanism executes these three steps sequentially.

[0082] like Figure 5 As shown, after data processing, the data can be input into the multilayer perceptron model for training. First, determine the input dimension, which is the dimension of the geological data features. Then, determine the output dimension, which is the number of categories to be classified for the geological formation. Next, determine the number and dimension of the hidden layers; these parameters can be set independently, but generally, the dimension of the first hidden layer should be larger than the input dimension. Finally, determine the activation function; here, the ReLU function is chosen. Finally, initialize parameters such as batch size and initial weights. This will yield the trained model.

[0083] like Figure 5 As shown, the trained deep learning neural network model based on multilayer perceptron is used for stratum identification to obtain stratum identification results.

[0084] In summary, when classifying strata based on data related to the physical properties of underground rocks and soil, this invention first requires data preprocessing. This process encounters three scenarios: missing data values, unnecessary data, and the need to modify strata categories using one-hot encoding. This invention employs a pipeline mechanism to abstract these three steps into a multi-step workflow. For machine learning users, pipelined machine learning is more efficient and easier to use than individual step-by-step modeling. The pipeline mechanism's application in machine learning stems from the reuse of parameter sets on new datasets, achieving a streaming encapsulation and management of all steps. It's more of a programming technique innovation than an algorithmic innovation. After data preprocessing, this invention uses this data for modeling, specifically a multilayer perceptron model. A multilayer perceptron is a deep learning neural network model composed of multiple layers of neurons. Each layer is connected to the layer above it, receiving input from it; simultaneously, each layer is also connected to the layer below it, influencing the neurons in that layer. When training a large model, there is a risk of overfitting. Therefore, weight decay is also needed to regularize the model, as well as methods for parameter initialization.

[0085] The multilayer perceptron-based stratigraphic identification method provided by this invention can also be used in conjunction with a computer-readable storage medium. The storage medium stores a computer program, which is executed to run the multilayer perceptron-based stratigraphic identification method. The computer program is capable of executing computer instructions, which include computer program code. The computer program code can be in the form of source code, object code, executable file, or some intermediate form.

[0086] Computer-readable storage media can include: any entity or device capable of carrying computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc.

[0087] It should be noted that the contents of computer-readable storage media may be appropriately added to or subtracted from the contents according to the requirements of legislation and patent practice in a jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable storage media may not include electrical carrier signals and telecommunication signals.

[0088] According to another aspect of the present invention, a formation identification device based on a multilayer perceptron is also provided, which performs a formation identification method based on a multilayer perceptron. The device includes: a preprocessing module and a formation identification module.

[0089] In one embodiment, the preprocessing module performs pipeline-based data preprocessing on the original data to obtain preprocessed data; the stratigraphic identification module inputs the preprocessed data into a deep learning neural network model based on a multilayer perceptron to obtain stratigraphic identification results.

[0090] In summary, this invention provides a stratum identification method based on a multilayer perceptron, which has the following advantages compared with existing technologies:

[0091] 1) This invention preprocesses the raw data through a pipeline mechanism, filling in missing data values, removing unnecessary data, and converting them into one-hot encodings, all of which are implemented one by one on a pipeline, thereby improving the usefulness of the data and making it more efficient and easier to use.

[0092] 2) Deep learning neural network models based on multilayer perceptrons possess powerful representational capabilities: Multilayer perceptrons can learn complex features of data and combine low-level features into high-level feature representations through a layer-by-layer pass-through method. This representational capability enables multilayer perceptrons to handle more complex tasks.

[0093] 3) Deep learning neural network models based on multilayer perceptrons possess nonlinear mapping capabilities: Multilayer perceptrons have the ability to map input data to a high-dimensional space, thereby better fitting the data. This is very useful for handling nonlinear problems.

[0094] 4) Deep learning neural network models based on multilayer perceptrons possess parallel computing capabilities: The structure of multilayer perceptrons makes them highly suitable for parallel computing. During training, the parameters of all layers can be updated simultaneously, thereby improving training speed.

[0095] 5) Deep learning neural network models based on multilayer perceptrons are easy to understand and debug: The structure of multilayer perceptrons is relatively simple, making them easy to understand and debug. This has led to their widespread application in many fields.

[0096] It should be understood that the embodiments disclosed herein are not limited to the specific structures, processing steps, or materials disclosed herein, but should be extended to equivalent substitutions of these features as understood by those skilled in the art. It should also be understood that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting.

[0097] In the description of this invention, unless otherwise stated, "a plurality of" means two or more; the terms "upper," "lower," "left," "right," "inner," "outer," "front end," "rear end," "head," "tail," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, the terms "first," "second," "third," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0098] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "connected" and "linked" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0099] Certain terms are used throughout this application to refer to specific system components. As those skilled in the art will recognize, the same components may often be referred to by different names, and therefore this application is not intended to distinguish those components that differ only in name and not in function. In this application, the terms “comprise,” “include,” and “have” are used in an open-ended manner and should therefore be interpreted as meaning “including, but not limited to…”. Furthermore, the terms “substantially,” “materially,” or “approximately” as used herein refer to industry-accepted tolerances for the corresponding terms. The term “coupling,” as may be used herein, includes direct coupling and indirect coupling via additional components, elements, circuits, or modules, wherein, for indirect coupling, the intermediate component, element, circuit, or module does not alter the information of the signal but may adjust its current level, voltage level, and / or power level. Inferred coupling (e.g., one element is inferredly coupled to another element) includes direct and indirect coupling between two elements in the same manner as “coupling.”

[0100] The phrase "an embodiment" or "an embodiment" used in this specification means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment of the invention. Therefore, the phrase "an embodiment" or "an embodiment" appearing in various places throughout the specification does not necessarily refer to the same embodiment.

[0101] The embodiments of the present invention are given for illustrative and descriptive purposes only, and are not intended to be exhaustive or to limit the invention to the forms disclosed. Many modifications and variations will be apparent to those skilled in the art. The embodiments were chosen and described in order to better illustrate the principles and practical application of the invention, and to enable those skilled in the art to understand the invention and to design various embodiments with various modifications suitable for a particular purpose.

[0102] While the embodiments disclosed in this invention are as described above, the content is merely for the purpose of facilitating understanding of the invention and is not intended to limit the invention. Any person skilled in the art to which this invention pertains may make any modifications and variations in form and detail of the implementation without departing from the spirit and scope disclosed herein; however, the scope of patent protection for this invention shall still be determined by the scope defined in the appended claims.

Claims

1. A method for stratigraphic identification based on a multilayer perceptron, characterized in that, The method includes: For the raw data, perform data preprocessing based on a pipeline mechanism to obtain preprocessed data; The preprocessed data is input into a deep learning neural network model based on a multilayer perceptron to obtain the formation identification results.

2. The formation identification method based on a multilayer perceptron as described in claim 1, characterized in that, The preprocessed data is obtained through the following steps: based on a pipeline mechanism, three operations—filling in missing data values, removing useless data, and converting stratigraphic category labels—are integrated into one pipeline and executed sequentially to obtain the preprocessed data.

3. The formation identification method based on a multilayer perceptron as described in claim 2, characterized in that, The missing data is filled by the following steps: filling the missing values ​​that may occur in the original data with the average value of the remaining data to obtain the filled data.

4. The formation identification method based on a multilayer perceptron as described in claim 3, characterized in that, The removal of useless data is performed through the following steps: For the fill data, unnecessary data is removed to obtain the removed data, wherein the unnecessary data includes irrelevant feature columns or erroneous data rows.

5. The formation identification method based on a multilayer perceptron as described in claim 4, characterized in that, The stratigraphic category label conversion is performed through the following steps: For the removed data, the stratigraphic category column in the last column is converted into a one-hot code to obtain the preprocessed data.

6. A method for stratigraphic identification based on a multilayer perceptron as described in any one of claims 1-5, characterized in that, The deep learning neural network model includes an input layer, multiple hidden layers, and an output layer, wherein the multiple hidden layers and the output layer are fully connected layers.

7. A method for stratigraphic identification based on a multilayer perceptron as described in any one of claims 1-6, characterized in that, The deep learning neural network model includes: Where: H represents the hidden layer, H∈R n×h R represents a real number; n represents the number of input samples; h represents the number of hidden units; σ represents the activation function; X represents the input layer, X∈R n×d ;d represents the number of input features for each input sample; W (1) W represents the hidden layer weights. (1) ∈R d×h b (1) Indicates the hidden layer bias, b (1) ∈R 1×h W (2) W represents the output layer weights. (2) ∈R h×q b (2) Indicates the output layer bias, b (2) ∈R 1×q O represents the output layer, O∈R n×q ; q represents the number of output features of the output layer.

8. The formation identification method based on a multilayer perceptron as described in claim 7, characterized in that, The activation function is the ReLU function.

9. A storage medium, characterized in that, It contains instructions for performing the method as described in any one of claims 1-8.

10. A formation identification device based on a multilayer perceptron, characterized in that, The apparatus for performing the method as described in any one of claims 1-8 comprises: The preprocessing module performs pipeline-based data preprocessing on the raw data to obtain preprocessed data. The stratigraphic identification module inputs the preprocessed data into a deep learning neural network model based on a multilayer perceptron to obtain stratigraphic identification results.

Images