Submarine cable magnetic anomaly detection method, device and equipment based on deep learning
By employing a deep learning-based method for detecting magnetic anomalies in submarine cables, and utilizing magnetic field forward modeling and neural network training, the problem of inaccurate submarine cable detection in traditional methods is solved, achieving efficient and accurate automatic detection of submarine cable locations.
Patent Information
- Application Number
- CN202511160521.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-08-19
- Publication Date
- 2025-11-25
AI Technical Summary
Existing traditional magnetic data processing methods cannot accurately estimate the burial depth of submarine cables, and magnetic anomaly inversion methods suffer from problems such as numerous false solutions and significant influence from parameter selection, resulting in inaccurate detection of magnetic anomalies in submarine cables.
By employing a deep learning-based approach, magnetic field forward modeling is performed using magnetic anomaly theory to generate sample datasets and train a deep learning neural network model, thereby enabling the automatic detection of magnetic anomalies in submarine cables.
It improves the accuracy of magnetic anomaly detection for submarine cables, simplifies the processing procedure, and enables automated, real-time detection of submarine cable locations.
Smart Images
Figure CN121008328A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of artificial intelligence geophysics, in particular to a seabed cable magnetic anomaly detection method, device and equipment based on deep learning. BACKGROUND
[0002] Magnetic detection is a detection means for determining the position and depth of a target by measuring the magnetic field change caused by objects with different magnetization strengths in the geomagnetic field. The magnetic field change caused by the seabed cable in the geomagnetic field will be detected by the magnetometer, thereby determining the position and related physical properties of the ferromagnetic object.
[0003] At present, the positioning technology for seabed cable magnetic anomalies still relies on traditional manual magnetic anomaly processing means, mainly realized by two types of methods of magnetic data processing interpretation and magnetic data inversion. The traditional magnetic data processing mainly includes smoothing, derivative, component conversion, polar reduction, analytical continuation and empirical method and tangent method, etc. The traditional magnetic data processing method cannot accurately estimate the burial depth information of the submarine cable, and needs to rely on the magnetic anomaly data inversion method to obtain the burial depth information of the submarine cable. The magnetic anomaly inversion method mainly includes Werner deconvolution method, potential field imaging, Euler deconvolution method, etc. The Werner deconvolution method is only suitable for specific field source models, i.e. downward infinite deep plate and thick contact zone, and can only provide an approximate solution for the upper position of the complex model, which has limitations in practical application. The effect of potential field imaging is easily affected by the actual magnetization direction, resulting in inaccurate inversion results. Compared with the above potential field inversion method, the Euler deconvolution method can automatically or semi-automatically determine the position of the field source under the condition of less prior information, without being affected by the magnetization direction. However, the Euler inversion method still has certain limitations, one of which is that the inversion is easy to produce a large number of false solutions, and the other is that the selection of the structure index has a great influence on the convergence and accuracy of the inversion. SUMMARY
[0004] The purpose of the present application is to provide a seabed cable magnetic anomaly detection method, device and equipment based on deep learning, so as to improve the accuracy of seabed cable magnetic anomaly detection.
[0005] To achieve the above purpose, the present application provides the following solutions.
[0006] In a first aspect, the present application provides a seabed cable magnetic anomaly detection method based on deep learning, comprising:
[0007] Performing magnetic field forward simulation by magnetic anomaly theory to obtain the magnetic anomaly of the seabed cable with different geometric parameters and physical parameters;
[0008] Generating a background magnetic field of the seabed environment;
[0009] constructing a sample data set based on the magnetic anomaly of the submarine cable with different geometric parameters and physical parameters and the background magnetic field, wherein the sample data set comprises a plurality of samples without the magnetic anomaly of the submarine cable and a plurality of samples with the magnetic anomaly of the submarine cable, and the label of the sample comprises a classification result, a horizontal position and a vertical position of the submarine cable;
[0010] training a deep learning neural network model based on the sample data set to obtain a trained deep learning neural network model;
[0011] detecting the magnetic anomaly of the submarine cable based on the trained deep learning neural network model.
[0012] Optionally, the magnetic field forward modeling is performed by a magnetic anomaly theory to obtain the magnetic anomaly of the submarine cable with different geometric parameters and physical parameters, and the magnetic anomaly is obtained by specifically comprising:
[0013] setting a position coordinate of the submarine cable according to a horizontal position coordinate range and a vertical position coordinate range, and setting geometric parameters and physical parameters of the submarine cable;
[0014] obtaining the magnetic anomaly of the submarine cable with different geometric parameters and physical parameters by simulating by using a magnetic field forward formula based on the position coordinate, the geometric parameters and the physical parameters of the submarine cable.
[0015] Optionally, the horizontal position coordinate range is 16m-48m, and the vertical position coordinate range is 2m-12m.
[0016] Optionally, the magnetic field forward formula is:
[0017]
[0018] wherein, ΔT x,D is the magnetic anomaly generated by the submarine cable, μ0 is the magnetic permeability in vacuum, M s is the effective magnetization, S is the cross-sectional area of the submarine cable, D is the central burial depth of the submarine cable, i s is the effective magnetization angle, I is the magnetic inclination, A is the magnetic azimuth, and x is the x-axis coordinate of the calculation point.
[0019] Optionally, the background magnetic field of the submarine environment is generated, and the background magnetic field is generated by specifically comprising:
[0020] generating three control points, wherein the horizontal coordinate of the control point is the horizontal distance, and the vertical coordinate of the control point is the induced magnetic field intensity of the submarine environment;
[0021] generating the background magnetic field of the submarine environment by spline interpolation on the three control points.
[0022] Optionally, based on the magnetic anomaly of the submarine cable with different geometric parameters and physical parameters and the background magnetic field, a sample data set is constructed, specifically including:
[0023] The magnetic anomaly of the submarine cable, the background magnetic field and 0-3% Gaussian random noise are added to obtain a sample containing the magnetic anomaly of the submarine cable;
[0024] The background magnetic field and 0-3% Gaussian random noise are added to obtain a sample not containing the magnetic anomaly of the submarine cable.
[0025] Optionally, the deep learning neural network model comprises a feature extraction module and a classification and positioning module;
[0026] The feature extraction module comprises four first convolution modules connected in sequence and a residual module connected with the last first convolution module, each of the convolution modules comprises a one-dimensional convolution layer, a ReLU activation function layer, an average pooling layer, a batch normalization layer and a random inactivation layer connected in sequence;
[0027] The classification and positioning module comprises a horizontal position regression submodule, a vertical position regression submodule and a classification submodule; the horizontal position regression submodule, the vertical position regression submodule and the classification submodule are connected with the residual module.
[0028] Optionally, the horizontal position regression submodule and the vertical position regression submodule each comprise two fully connected layers and a Linear activation function layer; the classification submodule comprises one fully connected layer and a Sigmoid activation function layer.
[0029] In a second aspect, the present application provides a deep learning-based submarine cable magnetic anomaly detection device, which applies the deep learning-based submarine cable magnetic anomaly detection method described above, and comprises:
[0030] A magnetic anomaly generation module is configured to perform magnetic field forward simulation through magnetic anomaly theory to obtain the magnetic anomaly of the submarine cable with different geometric parameters and physical parameters;
[0031] A background magnetic field generation module is configured to generate the background magnetic field of the submarine environment;
[0032] A sample data set construction module is configured to construct a sample data set based on the magnetic anomaly of the submarine cable with different geometric parameters and physical parameters and the background magnetic field; the sample data set comprises a plurality of samples not containing the magnetic anomaly of the submarine cable and a plurality of samples containing the magnetic anomaly of the submarine cable; the label of the sample comprises a classification result, a horizontal position and a vertical position of the submarine cable;
[0033] a model training module configured to train a deep learning neural network model based on the sample data set, and obtain a trained deep learning neural network model;
[0034] a magnetic anomaly detection module configured to detect a magnetic anomaly of the submarine cable based on the trained deep learning neural network model.
[0035] In a third aspect, the present application provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor executes the computer program to implement the deep learning-based magnetic anomaly detection method of the submarine cable.
[0036] According to the embodiments provided in the present application, the following technical effects are achieved.
[0037] The present application provides a deep learning-based magnetic anomaly detection method, device and equipment of submarine cable. The present application firstly performs magnetic field forward simulation through magnetic anomaly theory to obtain the magnetic anomaly of submarine cable with different geometric parameters and physical parameters; generates the background magnetic field of submarine environment; constructs a sample data set based on the magnetic anomaly of submarine cable with different geometric parameters and physical parameters and the background magnetic field; then trains a deep learning neural network model based on the sample data set to obtain a trained deep learning neural network model; and then detects the magnetic anomaly of submarine cable based on the trained deep learning neural network model. The present application applies the neural network of deep learning architecture to the magnetic anomaly detection of submarine cable, automatically learns the feature representation of the magnetic anomaly of submarine cable from complex magnetic field data based on the deep learning method, realizes the automatic detection of magnetic signal containing submarine cable, and outputs the horizontal position and vertical distance of submarine cable, thereby improving the accuracy of magnetic anomaly detection of submarine cable. BRIEF DESCRIPTION OF DRAWINGS
[0038] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed in the embodiments 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 be obtained by those skilled in the art without creative labor.
[0039] Figure 1 A flowchart of a deep learning-based magnetic anomaly detection method, device and equipment of submarine cable provided by an embodiment of the present application.
[0040] Figure 2 A magnetic field forward model diagram of submarine cable provided by an embodiment of the present application;
[0041] Figure 3 A deep learning sample generation flowchart provided by an embodiment of the present application;
[0042] Figure 4 A structural schematic diagram of a deep learning neural network model provided by an embodiment of the present application is shown in the following figure.
[0043] Figure 5 A comparison diagram of positioning error distribution histograms of an Euler method and a deep learning method (ResCNN) based on a synthetic data set provided by an embodiment of the present application is shown in the following figure.
[0044] Figure 6 A comparison diagram of magnetic anomaly identification and positioning results of an Euler method and a deep learning method (ResCNN) based on field collected data provided by an embodiment of the present application is shown in the following figure.
[0045] Figure 7 A comparison diagram of the flow of an Euler method and a deep learning method (ResCNN) processing submarine cable magnetic data provided by an embodiment of the present application is shown in the following figure.
[0046] Figure 8 A structural schematic diagram of a deep learning based submarine cable magnetic anomaly detection device provided by an embodiment of the present application is shown in the following figure.
[0047] Figure 9 A structural schematic diagram of a computer device provided by an embodiment of the present application is shown in the following figure. DETAILED DESCRIPTION
[0048] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the scope of protection of the present application.
[0049] The above purposes, features and advantages of the present application will be more apparent and easy to understand. The present application will be described in further detail below with reference to the drawings and specific embodiments.
[0050] In an exemplary embodiment, a deep learning based submarine cable magnetic anomaly detection method is provided, as shown in the following figure. Figure 1 The method includes the following steps 101-105.
[0051] Step 101, magnetic field forward modeling is performed by magnetic anomaly theory to obtain the magnetic anomaly of the submarine cable with different geometric parameters and physical parameters;
[0052] Step 102, the background magnetic field of the submarine environment is generated;
[0053] Step 103, based on the magnetic anomaly of the seabed cable with different geometric parameters and physical parameters and the background magnetic field, a sample data set is constructed; the sample data set includes a plurality of samples not containing the magnetic anomaly of the seabed cable and a plurality of samples containing the magnetic anomaly of the seabed cable; the label of the sample includes the classification result, the horizontal position and the vertical position of the seabed cable;
[0054] Step 104, training a deep learning neural network model based on the sample data set to obtain a trained deep learning neural network model;
[0055] Step 105, based on the trained deep learning neural network model, the magnetic anomaly of the seabed cable is detected.
[0056] Deep learning is essentially constructing a deeper multi-level neural network structure to extract and learn more useful features. It has been proven that the neural network based on the deep learning architecture can successfully convert the original input into a higher and more abstract representation, and performs better than traditional methods in establishing a learning model of the seabed cable magnetic anomaly to the seabed cable position parameter.
[0057] The above steps 101-105 of the present application apply the neural network of the deep learning architecture to the seabed cable magnetic anomaly detection, automatically learn the feature representation of the seabed cable magnetic anomaly from the complex magnetic field data based on the deep learning method, realize the automatic detection of the seabed cable magnetic signal, and output the horizontal position and vertical distance of the seabed cable, thereby improving the accuracy of the seabed cable magnetic anomaly detection.
[0058] In another exemplary embodiment, the above step 101, according to the geometric parameters and physical parameters of the common seabed cable, the magnetic anomaly of the seabed cable with different geometric parameters and physical parameters is obtained by magnetic anomaly theoretical forward modeling, Figure 2 The magnetic field forward model of the optical cable in the seabed cable (i.e. Figure 2 The xoy plane is the observation plane, the y axis is along the seabed cable, the x axis is perpendicular to the trend, the z axis is vertically downward, and the coordinate origin is selected at the center of the magnetic body. The coordinates of the calculation points outside the magnetic body are represented by (x, y, z). The coordinates of the body points inside the magnetic body are (ξ, η, ζ), the cross-sectional area of the seabed cable is S, the center depth of the seabed cable is D, the effective magnetization is M s , the magnetic azimuth is A, the magnetic inclination is I, and the effective magnetization inclination is i s . Therefore, the magnetic field generated in the direction perpendicular to the seabed cable is:
[0059]
[0060] Where, ΔT x,DThe magnetic anomaly is generated by the submarine cable, and μ0 is the permeability in vacuum, which is 4π×H / m.
[0061] The above formula represents the magnetic anomaly generated when the submarine cable runs perpendicular to the observation line (e.g., Figure 2 The calculation formula is shown below.
[0062] In the embodiments of this application, the calculation point can be equivalent to any point in the induced magnetic field of the submarine cable.
[0063] In this embodiment of the application, the process of generating the random submarine cable model is as follows:
[0064] Step 201, randomly set the coordinates of the submarine cable location, such as... Figure 3 As shown in (d), the horizontal position coordinates range from 16 to 48 meters, and the vertical distance ranges from 2 to 12 meters.
[0065] Step 202: Set the property parameters of the submarine cable. The effective cross-sectional radius of the cylinder composed of armored steel wire is 1 cm, and the relative magnetic permeability of the composite steel is 300.
[0066] Step 203: Randomly set the generated geomagnetic field parameters, perform forward modeling, and generate the magnetic anomaly of the submarine cable, such as... Figure 3 As shown in (a) above. The effective magnetization tilt angle is -90° to 90°. The angle between the observation profile and the direction of horizontal magnetization is 0° to 90°.
[0067] In another exemplary embodiment, the background magnetic field generation process is as follows:
[0068] Step 301: Generate the coordinates of three control points. Two control points have x-coordinates of 1 and 64 respectively, and the x-coordinate of the third control point is randomly generated within the range of 2-63m. The y-coordinate of each point is randomly generated within the range of 47300-47350nT. Figure 3 (e) in the middle.
[0069] Step 302: Generate the background magnetic field using spline interpolation of control points, i.e., generate a generated random background field using control points, such as... Figure 3 As shown in (b) above. Finally, two types of sample data are synthesized: one containing samples with submarine cable magnetic anomalies and the other without. Specifically, the submarine cable magnetic anomalies are added to the background magnetic field, and 0-3% Gaussian random noise is added to obtain the synthesized submarine cable magnetic anomaly samples, as shown in (b). Figure 3 As shown in (c) in the figure; the background magnetic field phase is directly added to 0-3% Gaussian random noise to obtain a synthetic sample that does not contain magnetic anomalies of submarine cables, as shown in the figure.Figure 3 The sample labels are the classification results, the horizontal position and the vertical distance of the submarine cable.
[0070] In another exemplary embodiment, the deep learning based submarine cable magnetic anomaly identification and positioning network is inspired by the multi-task learning network with shared hidden layers. The deep learning neural network model of the present application extracts high-order features through shared convolutional layers, and then performs a classification task to predict whether there is a submarine cable magnetic anomaly in the magnetic field data, and performs two regression tasks to fit the horizontal and vertical positions of the submarine cable. For the one-dimensional characteristics of the magnetic signal, the data is not converted to the two-dimensional spectral domain for image-like identification and classification. In order to improve the efficiency of model training and execution, a network structure for classification and fitting of one-dimensional magnetic anomaly signals is designed, as shown in Figure 4 Figure 4 In the figure, Flatten is data flattening, Horizontal position is horizontal position, Depth is depth, Submarine Cable is submarine cable, Detection is distance, and Probability is probability. Figure 4 is a structural diagram of the deep learning neural network model, and the input is one-dimensional magnetic field data. The number of sampling points of the input is 64. In the initial shared feature extraction stage, 4 one-dimensional convolutional layers are used to extract data features. After each convolutional layer, a ReLU activation function, an average pooling, a batch normalization and a random deactivation operation are applied. The size of the convolution kernel is 3, the pooling size of the average pooling layer is 2, and the deactivation rate of the random deactivation layer is 0.1. The pooling operation can reduce the redundant information in the features. The batch normalization operation can prevent overfitting and speed up the training process. Random deactivation realizes the regularization of the neural network model. In the latter half of the shared feature extraction stage, residual connection helps to extract higher-level features, and also ensures that the model does not degrade during the training process, which is mainly achieved by allowing the gradient to flow directly from the early layers to the later layers. After shared feature extraction, the output includes three branch tasks: one branch task is data classification, mainly used to determine whether the magnetic field data contains a submarine cable magnetic anomaly; the other two branch tasks are horizontal position regression and vertical distance regression, mainly used to determine the horizontal position and vertical distance of the submarine cable.
[0071] Figure 4 In the figure, the rectangles represent the outputs of the layers of the neural network. The output dimension of each layer is shown on its side. The arrows represent different operations. In the last layer of the data classification task, a Sigmoid function is applied to predict whether there is a submarine cable magnetic anomaly. In the last layer of the horizontal position regression and vertical distance regression tasks, a linear regression (Linear) function is applied to fit the actual position of the submarine cable.
[0072] In the embodiments of the present application, a loss function is used to evaluate the feedforward results of the network throughout the training process. Among them, the classification function and the regression loss function are binary cross-entropy and mean square error respectively, and the following is the total loss function:
[0073] Loss = Loss cls + Loss h_loc + Loss v_dist
[0074] Among them, Loss is the total loss function, which quantifies the closeness of the magnetic field data to the classification result and the seabed cable position mapping. Loss cls , Loss h_loc and Loss v_dist are the classification loss, horizontal position regression loss and vertical distance regression loss respectively.
[0075] In order to enable those skilled in the art to further understand the deep learning-based seabed cable magnetic anomaly recognition and positioning method of the embodiments of the present application, the simulation model and results are described in detail below.
[0076] Figure 5 The positioning error distribution histogram comparison of the Euler method (Euler) and the deep learning method (ResCNN) based on the synthetic data set. Figure 5 (a) in FIG. 1 is a horizontal position error distribution histogram comparison. Figure 5 (b) in FIG. 1 is a vertical distance error distribution comparison. Compared with the deep learning method, the error range of the Euler method in the horizontal position is larger, most of which is between-2 and 2 m. The vertical distance error range of the deep learning method is between-1 and 1 m, which is significantly smaller than that of the Euler method.
[0077] Further, in order to enable those skilled in the art to further understand the deep learning-based seabed cable magnetic anomaly recognition and positioning method of the embodiments of the present application in the actual data application situation, the field magnetic data and the recognition and positioning results are described in detail below.
[0078] Figure 6 The magnetic anomaly recognition and positioning result comparison chart of the Euler method (Euler) and the deep learning method (ResCNN) based on the field collected data. Figure 6 (a) in FIG. 2 is the original observed magnetic field and the background magnetic field. Figure 6 (b) in FIG. 2 is the magnetic anomaly of the submarine cable obtained by removing the background magnetic field. Figure 6 (c) in FIG. 2 is a comparison of the deep learning method and the Euler method result. The colored points represent the different positions of the submarine cable position obtained by the deep learning method. The black circles represent the positions of the submarine cable obtained by the Euler method.
[0079] Further, to further understand the advantages of the deep learning-based submarine cable magnetic anomaly recognition and positioning method of the embodiments of the present application over the Euler method, the data processing flow is described in detail.
[0080] Figure 7 A flowchart for comparing the processes of the Euler method and the deep learning method (ResCNN) for processing submarine cable magnetic data. The deep learning method is different from the traditional Euler method. The traditional Euler method processing flow first eliminates the background magnetic field from the original data to obtain the submarine cable magnetic anomaly. In the magnetic field separation process, it is difficult to completely separate the submarine cable magnetic anomaly from the magnetic field data regardless of the method chosen. The background magnetic field has a certain degree of influence on the submarine cable magnetic anomaly obtained by field separation, thus inevitably affecting the inversion result. Secondly, the submarine cable magnetic anomaly must be smoothed to reduce the influence on the inversion result. In the noise smoothing or filtering process, the submarine cable magnetic anomaly will inevitably be affected, thus affecting the inversion result. The implementation of the Euler method requires the adjustment of many related parameters, including the structure index, window size, and horizontal gradient filter coefficient. These parameters play a crucial role in obtaining accurate inversion results and need to be carefully adjusted and optimized. Finally, the traditional magnetic field separation, denoising, and Euler method require human-computer interaction for parameter adjustment, i.e., selecting appropriate parameters according to the characteristics of the magnetic field data. Parameter tuning requires a lot of time, which reduces the efficiency of data processing. Using the Euler method to obtain the position of the submarine cable has multiple solutions, and a clustering method must be used to eliminate false solutions and retain reliable solutions.
[0081] The deep learning method provides a direct mapping from the preprocessed magnetic data to the position of the submarine cable, without the need for field separation and denoising. Compared with the traditional Euler method processing flow, the processing flow is simplified and the efficiency is improved. The traditional Euler method processing process relies on human-computer interaction to optimize the inversion result and cannot realize real-time processing of data. The deep learning method directly processes the observed data after preprocessing without human intervention. Real-time processing capability test was conducted on a desktop computer equipped with an Intel(R) Core(TM) i9 processor, 64 Gb memory, and a GeForce GTX 1660 graphics card. The processing time for 67 lines with a total of 15436 sampling points was 35.7 seconds. This method can analyze data at a speed of 432 sampling points per second, which is significantly faster than the 10Hz sampling frequency of general magnetic field data, so the deep learning method has the ability of real-time processing.
[0082] Based on the same inventive concept, the embodiments of the present application also provide a deep learning based submarine cable magnetic anomaly detection device for implementing the deep learning based submarine cable magnetic anomaly detection method described above. The implementation scheme for solving problems provided by the device is similar to the implementation scheme described in the above method, so the specific limitations in one or more deep learning based submarine cable magnetic anomaly detection device embodiments provided below can refer to the limitations of the deep learning based submarine cable magnetic anomaly detection method described above, and will not be repeated here.
[0083] In one exemplary embodiment, a deep learning based submarine cable magnetic anomaly detection device is provided, as shown in Figure 8 comprises:
[0084] a magnetic anomaly generation module configured to obtain magnetic anomalies of submarine cables with different geometric parameters and physical parameters by magnetic field forward modeling based on magnetic anomaly theory;
[0085] a background magnetic field generation module configured to generate a background magnetic field of a submarine environment;
[0086] a sample data set construction module configured to construct a sample data set based on the magnetic anomalies of the submarine cables with different geometric parameters and physical parameters and the background magnetic field; the sample data set includes a plurality of samples without magnetic anomalies of submarine cables and a plurality of samples with magnetic anomalies of submarine cables; and labels of the samples include classification results, horizontal positions and vertical positions of submarine cables;
[0087] a model training module configured to train a deep learning neural network model based on the sample data set to obtain a trained deep learning neural network model;
[0088] a magnetic anomaly detection module configured to perform submarine cable magnetic anomaly detection based on the trained deep learning neural network model.
[0089] In one exemplary embodiment, a computer device is provided, which can be a server or a terminal, and an internal structure diagram thereof can be as shown in Figure 9As shown in the figure. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. Among them, the processor, the memory and the input / output interface are connected through the system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capability. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and the external device. The communication interface of the computer device is used to communicate with the terminal outside through the network connection. The computer program is executed by the processor to implement a deep learning-based submarine cable magnetic anomaly detection method.
[0090] Those skilled in the art can understand that, Figure 9 The structure shown in the figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. The specific computer device can include more or fewer components than those shown in the figure, or combine certain components, or have a different component arrangement. In one exemplary embodiment, a computer device is provided, including a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps in each of the above method embodiments.
[0091] In one exemplary embodiment, a computer readable storage medium is provided, storing a computer program, which is executed by a processor to implement the steps in each of the above method embodiments.
[0092] In one exemplary embodiment, a computer program product is provided, including a computer program, which is executed by a processor to implement the steps in each of the above method embodiments.
[0093] It should be noted that the user information (including but not limited to user equipment information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present application are all information and data authorized by the user or authorized by all parties.
[0094] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer readable storage medium, and when the computer program is executed, the processes of the above-mentioned embodiments of the methods can be included. Any reference to memory, databases or other media used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical storage, high-density embedded non-volatile memory, resistive memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration but not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc.
[0095] The database involved in the embodiments provided in the present application can include at least one of a relational database and a non-relational database. The non-relational database can include a distributed database based on a blockchain, etc., without being limited thereto. The processor involved in the embodiments provided in the present application can be a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., without being limited thereto.
[0096] The technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, all possible combinations of the technical features in the above embodiments are not described, but as long as the combinations of the technical features do not exist, they should be considered as the scope of the present application.
[0097] The principles and implementation modes of the present application are described by applying specific examples herein. The above description of the embodiments is only used to help understand the method and its core idea of the present application; meanwhile, for those skilled in the art, according to the idea of the present application, the specific implementation mode and application range can be changed. In conclusion, the content of the present application should not be understood as a limitation.
Claims
1. A method for detecting seabed cable magnetic anomaly based on deep learning, characterized in that, The method comprises the following steps: obtaining the magnetic anomaly of the seabed cable with different geometric parameters and physical parameters through magnetic anomaly theory and magnetic field forward simulation; generating the background magnetic field of the seabed environment; constructing a sample data set based on the magnetic anomaly of the seabed cable with different geometric parameters and physical parameters and the background magnetic field; the sample data set comprises a plurality of samples without the magnetic anomaly of the seabed cable and a plurality of samples with the magnetic anomaly of the seabed cable; the label of the sample comprises a classification result, a horizontal position and a vertical position of the seabed cable; training a deep learning neural network model based on the sample data set to obtain a trained deep learning neural network model; detecting the magnetic anomaly of the seabed cable based on the trained deep learning neural network model. 2.The deep learning-based seabed cable magnetic anomaly detection method of claim 1, wherein, The method for obtaining the magnetic anomaly of the seabed cable with different geometric parameters and physical parameters through magnetic anomaly theory and magnetic field forward simulation comprises the following steps: setting the position coordinates of the seabed cable according to the horizontal position coordinate range and the vertical position coordinate range, and setting the geometric parameters and physical parameters of the seabed cable; simulating the magnetic anomaly of the seabed cable with different geometric parameters and physical parameters by using a magnetic field forward formula based on the position coordinates, geometric parameters and physical parameters of the seabed cable. 3.The deep learning-based seabed cable magnetic anomaly detection method of claim 2, wherein, The horizontal position coordinate range is 16m-48m, and the vertical position coordinate range is 2m-12m. 4.The deep learning-based seabed cable magnetic anomaly detection method of claim 2, wherein, The magnetic field forward formula is: where ΔT x,D is the magnetic anomaly generated by the submarine cable, μ0 is the magnetic permeability in vacuum, M s is the effective magnetization, S is the cross-sectional area of the submarine cable, D is the central burial depth of the submarine cable, i s is the effective magnetization inclination, I is the magnetic inclination, A is the magnetic azimuth, and x is the x-axis coordinate of the calculation point. 5.The deep learning-based seabed cable magnetic anomaly detection method of claim 2, wherein, The method for generating the background magnetic field of the seabed environment comprises the following steps: generating three control points; the horizontal coordinates of the control points are horizontal distances, and the vertical coordinates of the control points are the induced magnetic field intensity of the seabed environment; generating the background magnetic field of the seabed environment by spline interpolation on the three control points. 6.The deep learning-based seabed cable magnetic anomaly detection method of claim 1, wherein, The method for constructing a sample data set based on the magnetic anomaly of the seabed cable with different geometric parameters and physical parameters and the background magnetic field comprises the following steps: adding the magnetic anomaly of the seabed cable, the background magnetic field and 0-3% Gaussian random noise to obtain a sample containing the magnetic anomaly of the seabed cable; adding the background magnetic field and 0-3% Gaussian random noise to obtain a sample without the magnetic anomaly of the seabed cable. 7.The deep learning-based seabed cable magnetic anomaly detection method of claim 1, wherein, The deep learning neural network model comprises a feature extraction module and a classification and positioning module; the feature extraction module comprises four first convolution modules connected in sequence and a residual module connected with the last first convolution module; each convolution module comprises a one-dimensional convolution layer, a ReLU activation function layer, an average pooling layer, a batch normalization layer and a random inactivation layer connected in sequence; the classification and positioning module comprises a horizontal position regression submodule, a vertical position regression submodule and a classification submodule; the horizontal position regression submodule, the vertical position regression submodule and the classification submodule are connected with the residual module. 8.The deep learning-based seabed cable magnetic anomaly detection method according to claim 7, characterized in that, The horizontal position regression submodule and the vertical position regression submodule each comprise two full connection layers and a Linear activation function layer; the classification submodule comprises a full connection layer and a Sigmoid activation function layer. 9.A deep learning based seabed cable magnetic anomaly detection device, characterized in that, The deep learning-based seabed cable magnetic anomaly detection device is applied to the deep learning-based seabed cable magnetic anomaly detection method in any one of claims 1-8, and comprises The magnetic anomaly generation module is configured to perform forward simulation of a magnetic field by using a magnetic anomaly theory to obtain a magnetic anomaly of the seabed cable with different geometric parameters and physical parameters; The background magnetic field generation module is configured to generate a background magnetic field of a seabed environment; The sample data set construction module is configured to construct a sample data set based on the magnetic anomaly of the seabed cable with different geometric parameters and physical parameters and the background magnetic field; the sample data set includes a plurality of samples not containing the magnetic anomaly of the seabed cable and a plurality of samples containing the magnetic anomaly of the seabed cable; and a label of a sample includes a classification result, a horizontal position and a vertical position of the seabed cable; The model training module is configured to train a deep learning neural network model based on the sample data set to obtain a trained deep learning neural network model; The magnetic anomaly detection module is configured to perform seabed cable magnetic anomaly detection based on the trained deep learning neural network model.
10. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, and characterized in that the processor executes the computer program to implement the deep learning-based seabed cable magnetic anomaly detection method in any one of claims 1-8.