GIS electromagnetic field rough simulation data enhancement method, system and storage medium based on differential fusion neural network

Abstract

The present invention discloses a GIS electromagnetic field coarse simulation data enhancement method, system and storage medium based on differential fusion neural network, comprising obtaining a GIS electromagnetic field coarse grid simulation data set; inputting the GIS electromagnetic field coarse grid simulation data set into a differential deep learning network model, the differential deep learning network model comprising an enhancement network and a structural similarity network connected in sequence, the enhancement network comprising a self-adjusting module and a differential convolution module; using the self-adjusting module and the differential convolution module to calculate the coarse grid structure data and the coarse grid field strength data respectively, to obtain fine grid enhancement grid structure characteristics and fine grid enhancement field strength characteristics; using the structural similarity network to match the fine grid enhancement grid structure characteristics and the fine grid enhancement field strength characteristics, and calculating the similarity characteristics; based on the similarity characteristics, calculating the GIS fine grid enhancement data; the present invention can take into account both the accuracy and efficiency of GIS electromagnetic simulation.

Description

Technical Field

[0001] The invention belongs to the technical field of electromagnetic field simulation, and in particular relates to a GIS electromagnetic field rough simulation data enhancement method, system and storage medium based on differential fusion neural network. Background Art

[0002] Gas Insulated Switchgear (GIS) plays a vital role in modern power systems. This equipment uses inert gas as an insulating medium to achieve safe and reliable transmission and distribution of high-voltage electricity in a relatively small space. Compared with traditional air-insulated switchgear, GIS has higher reliability and longer service life, while reducing the impact of the equipment on the environment. In addition, due to its compact design, GIS is particularly suitable for urban areas or space-constrained environments where land utilization costs are high. The high performance and reliability of GIS are essential to ensure the stable operation of the power system, improve the transmission efficiency of the power grid, and ensure the security of power supply.

[0003] With the rapid development of power systems and the increasing requirements for grid security performance, electromagnetic simulation of GIS has become one of the research focuses. Electromagnetic simulation methods can help engineers and researchers deeply understand the electromagnetic field distribution, electromagnetic interference and its impact on GIS performance.

[0004] Currently, electromagnetic simulation mainly uses finite element method, boundary element method and traditional analytical methods, etc. These methods can provide detailed information on the electromagnetic field inside GIS, help optimize GIS design and reduce failure rates.

[0005] In recent years, with the development of computing technology, more advanced simulation technologies such as multi-physics coupling simulation and machine learning-based simulation optimization methods have been gradually applied to GIS research. These studies have not only promoted the advancement of GIS technology, but also provided important support for improving the overall performance and safety of power systems. Nevertheless, electromagnetic simulation still faces challenges when dealing with extremely high voltages and complex structures, which requires the development of more accurate and efficient simulation tools and methods in future research.

[0006] In related technologies, researchers have established broadband equivalent circuit models of 1000kV UHV AC GIS equipment bushings, disconnectors, branch busbars, tube busbars, insulating flanges and other components, grounding pillars and grounding grids, and proposed a joint simulation model of fast transient overvoltage and transient casing voltage coupled with multi-conductor transmission lines and lumped broadband components. The effectiveness of the modeling method and simulation model was verified by comparing the measured results of each fast transient overvoltage and transient casing voltage measurement point in the UHV GIS test circuit with the simulation results. However, this method has a slow simulation speed and cannot meet the requirements of high precision and speed.

[0007] Some researchers have also established a transmission characteristic model of the casing of a semi-enclosed gas-insulated combination electrical appliance, and used multi-conductor transmission line theory and phase mode transformation methods to perform detailed calculations and analyses of the ultra-fast transient overvoltages and transient casing voltages generated by the operation of the disconnector. This approach has improved the simulation speed to a certain extent by simplifying the model, but model simplification and assumptions may affect the accuracy of the results. The simplification of the grounding grid model may not fully capture the complexity of the actual situation, and the generalization ability and scope of application of the research results may be limited, resulting in low accuracy of the simulation results.

[0008] In the patent application document with publication number CN117725769A, a method for fast three-dimensional forward modeling of complex models of airborne electromagnetics based on multi-scale finite elements is proposed. This scheme is used to realize fast three-dimensional forward modeling of airborne electromagnetics. Based on the multi-scale finite element method, the amount of calculation is reduced by dividing coarse and fine grids and optimizing calculations, and the efficiency of geophysical electromagnetic calculations is improved by reducing the order of the forward equation. However, since GIS electromagnetic field simulation focuses more on the accurate simulation of detailed interactions between equipment and electromagnetic fields, rather than large-scale geological structure detection, GIS electromagnetic field simulation needs to consider the influence of the specific structure and material properties of power equipment on the electromagnetic field, as well as how the electromagnetic field is distributed and changed inside these equipment. This requires the model to be able to perform accurate calculations at a finer scale and adapt to specific physical and engineering constraints. Moreover, the application of the multi-scale finite element method proposed in the patent document in airborne electromagnetic forward modeling focuses on processing electromagnetic wave reflection, refraction and other phenomena at the interface of the medium, as well as electromagnetic wave attenuation inside the medium, which is sensitive to large-scale changes and is not enough to finely process the electromagnetic field details inside the GIS equipment.

[0009] In the patent application document with publication number CN113221403A, a finite element mesh optimization method based on deep reinforcement learning is proposed. This method directly optimizes the quality of the finite element mesh through deep reinforcement learning to improve the accuracy of finite element mesh division. However, the application of deep reinforcement learning adopted by this method focuses on directly optimizing and adjusting the mesh structure through learning and decision-making processes, which requires a clear reward or penalty mechanism to evaluate the effect of mesh optimization. It is an optimization process centered on mesh quality. This method proposes the application of deep reinforcement learning in mesh optimization. Although it can improve the mesh quality of finite element analysis, it focuses on the improvement of mesh structure and does not directly optimize the quality of electromagnetic field simulation data and the physical accuracy of the simulation process. Moreover, deep reinforcement learning algorithms are usually trained for specific environments or tasks. They learn how to perform tasks through interaction with the environment. The decision-making strategy of the model is highly dependent on the environmental characteristics during its training. This means that if the environment changes (for example, from finite element mesh optimization to GIS electromagnetic field simulation), the previously trained model may not be able to effectively adapt to the new environment.

[0010] In the patent application document with publication number CN113722951A, it is proposed to use a neural network to establish a mapping relationship from A-dimensional grid feature variables to the root mean square error of the far-field radar scattering cross section, and use a differential evolution algorithm to solve the optimal grid feature variable value to obtain the optimization result of the three-dimensional finite element grid of the scatterer; however, this scheme uses the differential evolution algorithm to optimize the established neural network model, and the goal is parameter optimization, specifically optimizing the network weights to achieve the optimization of the prediction performance. That is to say, the differential evolution algorithm is independent of the training process of the neural network and is used as a subsequent parameter adjustment step. Summary of the invention

[0011] The technical problem to be solved by the present invention is how to balance the accuracy and efficiency of GIS electromagnetic simulation.

[0012] The present invention solves the above technical problems by the following technical means:

[0013] The present invention proposes a GIS electromagnetic field rough simulation data enhancement method based on differential fusion neural network, and the method comprises the following steps:

[0014] Acquire a GIS electromagnetic field coarse grid simulation data set, wherein the GIS electromagnetic field coarse grid simulation data set includes coarse grid structure data and coarse grid field strength data;

[0015] Inputting the GIS electromagnetic field coarse grid simulation data set into a differential deep learning network model, wherein the differential deep learning network model includes an enhanced network and a structural similarity network connected in sequence, and the enhanced network includes a self-adjusting module and a differential convolution module;

[0016] The self-adjusting module and the differential convolution module are used to calculate the coarse grid structure data and the coarse grid field strength data respectively to obtain fine grid enhanced grid structure characteristics and fine grid enhanced field strength characteristics;

[0017] Using the structural similarity network, the fine grid enhanced grid structure feature and the fine grid enhanced field strength feature are matched to calculate a similarity feature;

[0018] Based on the similarity features, GIS fine grid enhancement data is calculated.

[0019] Furthermore, the self-adjusting module includes a first convolutional layer and a second convolutional layer connected in sequence, and the output features of the first convolutional layer and the output features of the second convolutional layer are output to the activation function layer after addition operation;

[0020] The coarse grid structure data is used as the input of the first convolutional layer, the grid size supplementary information of the coarse grid structure data is used as the input of the second convolutional layer, the bias vector of the coarse grid structure data is used as the input of the first addition operation, and the coarse grid structure data and the grid size supplementary information of the coarse grid structure data are output to the first addition operation via residual connection.

[0021] Furthermore, the process formula for calculating the coarse grid structure data using the self-adjusting module is expressed as:

[0022]

[0023] In the formula, Enhance the mesh structure features for fine meshes, is the convolution operation, is the coarse grid structure data, Supplementary information for the grid size of the coarse grid structure data, is the deviation vector of the coarse grid structure data, is the residual link, is the activation function.

[0024] Furthermore, the differential convolution module includes a differential convolution layer, a size integration layer, a self-attention mechanism layer and a third convolution layer connected in sequence, and the third convolution layer is connected to an activation function;

[0025] The coarse grid field intensity data is used as the input of the differential convolution layer, and the differential convolution layer is used to calculate the field intensity features of different sizes of the coarse grid field intensity data by using a differential algorithm;

[0026] The size integration layer is used to sum the field intensity features of different sizes calculated by the differential convolution layer to obtain the reorganized field intensity features.

[0027] Furthermore, the process formula for calculating the coarse grid field strength data using the differential convolution module is expressed as:

[0028]

[0029]

[0030]

[0031] In the formula, is the field intensity feature generated by differential convolution, is the differential convolution operation, is the coarse grid field strength data, is the field strength characteristic after reorganization, represents the size convolution operation, Enhance the field strength characteristics for the fine grid, For the self-attention mechanism operation, is the convolution operation, is the activation function.

[0032] Furthermore, the convolution kernel of the differential convolution layer adopts any one of a five-point differential convolution kernel, a weighted differential convolution kernel, a multi-scale differential convolution kernel, a directional differential convolution kernel, a nine-point differential convolution kernel, and a mixed mode differential convolution.

[0033] Furthermore, the structural similarity network includes a first branch network, a second branch network, a second addition operation and a first multilayer perceptron, the outputs of the first branch network and the second branch network are both connected to the second addition operation, the output of the second addition operation is connected to the first multilayer perceptron, and the multilayer perceptron is followed by an activation function.

[0034] Furthermore, the first branch network includes a convolutional neural network layer CNN and a batch normalization operation connected in sequence, and the regularization operation is followed by an activation function;

[0035] The second branch network includes a second multilayer perceptron, and the second multilayer perceptron is followed by an activation function.

[0036] Furthermore, the using the structural similarity network to match the fine grid enhanced grid structure feature and the fine grid enhanced field strength feature to calculate the similarity feature includes:

[0037] The structural feature information of the fine grid enhanced grid structure feature is extracted by using the first branch network, and the formula is expressed as:

[0038]

[0039] In the formula, For the The fine grid enhances the characteristic information of the grid structure characteristics, Enhance the grid structure characteristics for the fine grid, Indicates the use of stacked convolution operations, is the batch normalization operation, is the activation function;

[0040] The second branch network is used to extract the field intensity feature information of the fine grid enhanced field intensity feature, and the formula is expressed as:

[0041]

[0042] In the formula, For the The characteristic information of the fine grid enhanced field strength characteristics, Enhance the field strength characteristics for the fine grid, Operations performed by the second multilayer perceptron;

[0043] The field strength and structure combination feature is calculated by using the modulation weights corresponding to the feature information of the fine grid enhanced grid structure feature and the feature information of the fine grid enhanced field strength feature. The formula is expressed as:

[0044]

[0045] In the formula, is the field strength and structure combination feature, and The modulation weights corresponding to the feature information of the fine grid enhanced grid structure feature and the feature information of the fine grid enhanced field strength feature are respectively represented;

[0046] Based on the field strength and structure combination characteristics, the similarity characteristics are calculated, and the formula is expressed as:

[0047]

[0048] In the formula, For the similarity feature, Operations performed by the first multilayer perceptron.

[0049] Furthermore, the calculating of GIS fine grid enhancement data based on the similarity feature includes:

[0050] Calculating regularized coarse grid structure data, regularized coarse grid field intensity data, regularized fine grid enhanced grid structure characteristics, and regularized fine grid enhanced field intensity characteristics based on the coarse grid structure data, the coarse grid field intensity data, the fine grid enhanced grid structure characteristics, and the fine grid enhanced field intensity characteristics, respectively;

[0051] Calculating grid structure consistency based on the regularized coarse grid structure data and the regularized fine grid enhanced grid structure features;

[0052] Calculating grid field strength consistency based on the regularized coarse grid field strength data and the regularized fine grid enhanced field strength characteristics;

[0053] Calculating grid field strength loss compensation based on the grid structure consistency and the grid field strength consistency;

[0054] Calculating fine grid enhancement structure data based on the fine grid enhancement grid structure characteristics, the grid field strength loss compensation and the similarity characteristics;

[0055] Based on the fine grid enhanced field strength characteristics, the grid field strength loss compensation and the similarity characteristics, fine grid enhanced field strength data is calculated.

[0056] Furthermore, before inputting the GIS electromagnetic field coarse grid simulation data set into the differential deep learning network model, the method further includes:

[0057] A GIS electromagnetic field coarse grid historical simulation data set is used, and the differential deep learning network model is trained using the historical simulation data set until the total loss function of the model is minimized, wherein the total loss function includes a constraint loss function based on Maxwell's equations and a field strength and structural consistency loss function, and the formula is expressed as:

[0058]

[0059] In the formula, is the total loss function, is the field strength and structure consistency loss function, is the Maxwell constraint loss function, and is the weight parameter.

[0060] Furthermore, the formula of the field strength and structure consistency loss function is expressed as:

[0061]

[0062]

[0063]

[0064] In the formula, represents the gradient obtained by applying the difference method, is the structural similarity loss, is the field strength consistency loss, and is the scale factor, Enhance the mesh structure features for fine meshes, is the coarse grid structure data, To enhance the field strength characteristics for fine grids, is the coarse grid field strength data, is the 2-norm of the vector.

[0065] Furthermore, the formula of the constraint loss function based on Maxwell equations is expressed as:

[0066]

[0067]

[0068]

[0069]

[0070]

[0071]

[0072] In the formula, is the magnetic field component, is the Gaussian law loss, is Gauss's magnetic law loss, is the loss due to Faraday's law of electromagnetic induction, is Ampere's law loss, The proportionality constant, , , , is the scale factor, represents the curl, is the vector 2 norm; is the magnetic permeability of vacuum, is the permittivity of vacuum, It is the fine grid enhanced field strength data.

[0073] In addition, the present invention also proposes a GIS electromagnetic field coarse grid data enhancement system based on differential GAN, the system comprising:

[0074] A data acquisition module is used to acquire a GIS electromagnetic field coarse grid simulation data set, wherein the GIS electromagnetic field coarse grid simulation data set includes coarse grid structure data and coarse grid field strength data;

[0075] A data enhancement module is used to process the GIS electromagnetic field coarse grid simulation data set using the deployed differential deep learning network model, wherein the differential deep learning network model includes an enhancement network and a structural similarity network connected in sequence, and the enhancement network includes a self-adjusting module and a differential convolution module; the coarse grid structure data and the coarse grid field strength data are calculated using the self-adjusting module and the differential convolution module respectively to obtain fine grid enhanced grid structure features and fine grid enhanced field strength features; the fine grid enhanced grid structure features and the fine grid enhanced field strength features are matched using the structural similarity network to calculate similarity features;

[0076] The data calculation module is used to calculate GIS fine grid enhanced data based on the similarity features.

[0077] In addition, the present invention also proposes a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the GIS electromagnetic field coarse simulation data enhancement method based on differential fusion neural network as described above is implemented.

[0078] The advantages of the present invention are:

[0079] (1) Since GIS electromagnetic field simulation focuses more on the accurate simulation of the detailed interaction between equipment and electromagnetic fields, the present invention designs and constructs a differential deep learning network model, which uses convolution combined with a differential algorithm to process coarse grid field strength data, and can accurately simulate the electromagnetic field distribution of GIS, especially when dealing with complex geometric structures and boundary conditions; and uses a deep learning network to optimize the simulation process, improve the simulation accuracy, and accelerate the simulation process, effectively shortening the design and evaluation cycle, thereby taking into account both the accuracy and efficiency of GIS electromagnetic simulation.

[0080] (2) The present invention designs a self-adjusting module for realizing refined grid enhancement. The self-adjusting module realizes refined enhancement of the coarse grid structure data of the electromagnetic field of the GIS equipment by combining convolution operation, residual connection and sigmoid activation function. This not only improves the detail expression of the grid structure data, but also the self-adjusting module can realize adaptive grid enhancement of network structure input data of different sizes by stacking different convolution layers, different convolution kernel sizes and step sizes. It can adaptively adjust the enhancement strategy according to the original grid data, thereby ensuring the accuracy and adaptability of the enhanced grid structure.

[0081] (3) The present invention designs a differential convolution module for accurate extraction of field intensity data. The differential convolution module uses a specially designed differential convolution operation, and the concept of difference is embedded in the training process of the neural network. In particular, as a part of the network, the differential algorithm is directly used in the generation process through a specially designed convolution kernel, so that the differential algorithm is used at the data level and directly affects the data generated by the model. Therefore, the differential algorithm is used as part of the data processing step to capture the spatial features in the electromagnetic field data, rather than as a tool for optimizing network parameters. The designed differential convolution module can accurately extract the electromagnetic field intensity features and effectively simulate the physical distribution of the electromagnetic field. In combination with the self-attention mechanism, it can highlight important field intensity features, improve the extraction accuracy of field intensity data and the attention concentration of the model; and by extracting and summing features with convolution kernels of different sizes, it can capture the multi-scale features of the electromagnetic field data, which is particularly important for understanding and simulating the complex changes of the electromagnetic field. This design improves the generalization ability of the model, enabling it to better adapt to electromagnetic field changes of different scales.

[0082] (4) The present invention designs a multi-dimensional loss function. By combining structural similarity loss, field strength consistency loss and physical constraint loss, multiple important characteristics of simulation data are comprehensively considered. The loss function combines the advantages of traditional physical models and deep learning. Through precise mathematical expressions and physical constraints, it helps to optimize the stability and convergence speed of the algorithm and improve training efficiency. The design of this multi-dimensional loss function is conducive to ensuring the accuracy of the simulation while also being able to adjust the weights between different losses to adapt to specific simulation needs and optimization goals.

[0083] (5) By setting the structural similarity loss and field strength consistency loss, it is ensured that the enhanced electromagnetic field data is highly consistent with the original data in terms of structure and field strength. This consistency not only promotes the convergence of the simulation data, but also ensures the physical authenticity of the simulation results and improves the reliability of the simulation.

[0084] (6) By introducing a scaling factor into the structural similarity loss function, a flexible loss function adjustment mechanism is provided. This mechanism can not only adjust the importance of different parts of the loss function according to specific tasks, but also improve the generalization ability and adaptability of the model when processing different types of electromagnetic field data.

[0085] (7) The physical constraint loss function ensures that the enhanced electromagnetic field data strictly follows the basic laws of electromagnetism through the key components of Maxwell's equations - Gauss's law, Gauss's law of magnetism, Faraday's law of electromagnetic induction, and Ampere's law. This constraint makes the electromagnetic field data output by the model not only numerically accurate but also strict in physical laws, which increases the scientificity and practicality of the model.

[0086] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0087] Figure 1 It is a flow chart of a GIS electromagnetic field rough simulation data enhancement method based on a differential fusion neural network proposed in one embodiment of the present invention;

[0088] Figure 2 It is a principle block diagram of GIS electromagnetic field coarse grid data enhancement based on differential network in one embodiment of the present invention;

[0089] Figure 3 It is a schematic diagram of the structure of an enhanced network in a differential deep learning network model according to an embodiment of the present invention;

[0090] Figure 4 Schematic diagram of the design principle of a differential convolution module in one embodiment of the present invention;

[0091] Figure 5 It is a schematic diagram of the structure of a structural similarity network in a differential deep learning network model according to an embodiment of the present invention;

[0092] Figure 6 It is a structural schematic diagram of a GIS electromagnetic field coarse grid data enhancement system based on a differential network proposed in one embodiment of the present invention. DETAILED DESCRIPTION

[0093] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in combination with the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0094] like Figure 1 to Figure 2 As shown, the first embodiment of the present invention proposes a GIS electromagnetic field rough simulation data enhancement method based on differential fusion neural network, and the method includes the following steps:

[0095] S10, obtaining a GIS electromagnetic field coarse grid simulation data set, wherein the GIS electromagnetic field coarse grid simulation data set includes coarse grid structure data and coarse grid field strength data;

[0096] It should be noted that this embodiment can specifically start from the finite element method, use modeling software such as Solidworks to establish a three-dimensional model of the GIS equipment, and then use finite element analysis software such as ANSYS Workbench to simulate and calculate the three-dimensional electric field strength of the three-dimensional model of the GIS equipment.

[0097] S20, inputting the GIS electromagnetic field coarse grid simulation data set into a differential deep learning network model, wherein the differential deep learning network model includes an enhanced network and a structural similarity network connected in sequence, and the enhanced network includes a self-adjusting module and a differential convolution module;

[0098] S30, using the self-adjusting module and the differential convolution module to respectively calculate the coarse grid structure data and the coarse grid field strength data to obtain fine grid enhanced grid structure features and fine grid enhanced field strength features;

[0099] S40, using the structural similarity network to match the fine grid enhanced grid structure feature and the fine grid enhanced field strength feature, and calculating a similarity feature;

[0100] S50. Calculate GIS fine grid enhancement data based on the similarity features.

[0101] This embodiment designs and constructs a differential deep learning network model, in which convolution combined with a differential algorithm is used to process coarse grid field strength data, which can accurately simulate the electromagnetic field distribution of GIS, especially when dealing with complex geometric structures and boundary conditions; and uses a deep learning network to optimize the simulation process, improve simulation accuracy, and accelerate the simulation process, effectively shortening the design and evaluation cycle, thereby taking into account both the accuracy and efficiency of GIS electromagnetic simulation.

[0102] As a further preferred technical solution, this embodiment constructs a GIS electromagnetic field coarse grid simulation data set by using finite element analysis software; the GIS equipment electromagnetic field coarse grid simulation data is segmented according to the data type to obtain a coarse grid structure data set and a coarse grid field strength data set, which are recorded as ,in, represents a coarse grid structured dataset, and , Indicates GIS equipment electromagnetic field coarse grid data, and , Indicates Coarse grid structure coordinates of electromagnetic field of GIS equipment; represents the coarse-grid field strength dataset, and , Indicates GIS equipment electromagnetic field coarse grid field strength data, and , Indicates The field strength of the electromagnetic field coarse grid data of GIS equipment at the grid points; , Indicates the total number of simulation data; ， , , 、 and Represents the coarse grid simulation size.

[0103] As a further preferred technical solution, Figure 3 As shown, the self-adjusting module includes a first convolutional layer Conv1 and a second convolutional layer Conv2 connected in sequence, and the outputs of the first convolutional layer Conv1 and the second convolutional layer Conv2 are connected through a first addition operation and then output to the activation function layer;

[0104] The coarse grid structure data is used as the input of the first convolution layer Conv1, the grid size supplementary information of the coarse grid structure data is used as the input of the second convolution layer Conv2, the deviation vector of the coarse grid structure data is used as the input of the first addition operation, and the coarse grid structure data and the grid size supplementary information of the coarse grid structure data are connected via residual Output to the first adding operation.

[0105] It should be noted that this embodiment realizes the refined enhancement of the coarse grid structure data by designing a self-adjusting module. The self-adjusting module realizes the refined enhancement of the electromagnetic field coarse grid data of the GIS equipment by combining convolution operations, residual connections and sigmoid activation functions. This method not only improves the detail expression of the grid data, but also can adaptively adjust the enhancement strategy according to the original grid data through the self-adjusting mechanism, thereby ensuring the accuracy and adaptability of the enhanced grid structure.

[0106] Furthermore, the function of the self-adjusting module is to perform fine grid structure enhancement. GIS equipment electromagnetic field coarse grid data Input the self-adjusting module in the enhancement module to obtain the fine grid enhancement grid structure characteristics , the formula is:

[0107]

[0108] In the formula, Enhance the mesh structure features for fine meshes, represents the grid-enhanced convolution operation, represents the fine-tuning convolution operation, is the coarse grid structure data, Supplementary information for the grid size of the coarse grid structure data, is the deviation vector of the coarse grid structure data, is the residual link, is the activation function.

[0109] As a further preferred technical solution, Figure 3 As shown, the differential convolution module includes differential convolution layers connected in sequence , Size Integration Layer , self-attention mechanism layer And the third convolutional layer Conv3, the third convolutional layer Conv3 is connected with Activation function;

[0110] The coarse grid field strength data is used as the differential convolution layer The input of the differential convolution layer Used to calculate field intensity characteristics of different sizes of the coarse grid field intensity data by using a differential algorithm;

[0111] The size integration layer Used to sum the field intensity features of different sizes calculated by the differential convolution layer to obtain a reshaped field intensity feature;

[0112] The reshaped field strength features are passed through the self-attention mechanism layer , after the third convolutional layer Conv3, the fine grid enhanced field strength features are output.

[0113] It should be noted that this embodiment designs a differential convolution module for accurate extraction of field intensity data. The differential convolution module can accurately extract electromagnetic field intensity features through specially designed differential convolution operations, effectively simulate the physical distribution of the electromagnetic field, and combine the self-attention mechanism. This module can highlight important field intensity features and improve the extraction accuracy of field intensity data and the attention concentration of the model. And by extracting and summing features with convolution kernels of different sizes, it is possible to capture the multi-scale features of electromagnetic field data, which is particularly important for understanding and simulating the complex changes of electromagnetic fields. This method improves the generalization ability of the model, enabling it to better adapt to changes in electromagnetic fields of different scales.

[0114] Specifically, this embodiment needs to accurately map the physical principles of the differential algorithm to the design of the convolution kernel, converting these continuous physical laws into convolution operations suitable for discrete data. Physical phenomena in electromagnetic field simulation usually exhibit different characteristics at different scales. If a convolution kernel is designed to capture multi-scale physical characteristics at the same time, it is necessary to comprehensively consider local and global information, which is not easy to achieve in numerical methods (differentials). Because traditional differential methods are usually more commonly used in two-dimensional or one-dimensional space, these methods need to be extended to three-dimensional space in three-dimensional GIS electromagnetic field simulation. In electromagnetic field simulation, physical processes are often independent of direction. The purpose of this embodiment is to design a rotationally invariant convolution kernel that can give a consistent physical response in any direction.

[0115] This embodiment first designs a basic convolution kernel based on the idea of ​​differential calculation, then applies the two-dimensional convolution kernel to the three-dimensional field strength data according to the needs of three-dimensional GIS electromagnetic field simulation, transforms the dimension of the two-dimensional differential convolution kernel, obtains the three-dimensional differential convolution kernel using the dimension transformation kernel, and uses the three-dimensional differential convolution kernel to perform a convolution operation with the original coarse grid data to obtain the gradient information of the coarse grid field strength. Then, the original coarse grid field strength data is upsampled to obtain the sampled fine grid field strength data, and the gradient information of the coarse grid field strength obtained by three-dimensional difference is used to refine and enhance the sampled fine grid field strength data to obtain preliminary fine grid data.

[0116] Compared with the process of obtaining the traditional convolution kernel following network training, the three-dimensional differential convolution kernel proposed in this embodiment is based on the differential algorithm (multiple formats). According to the differential algorithm, the calculation method of obtaining the central data point data according to the data points on the front and back steps (differential steps) of the x and y axes is used to abstract the basic convolution kernel. This convolution kernel can realize similar differential of the data on each step (step in the convolution) in the convolution operation, but the traditional differential is only used in two dimensions. This embodiment also needs to convert the abstract convolution kernel in dimension to adapt to the three-dimensional GIS electromagnetic simulation task.

[0117] Take five-point difference as an example. Figure 4 As shown, the differential convolution layer The design is as follows:

[0118] Design the basic convolution kernel based on the five-point difference algorithm :

[0119]

[0120] Considering the data enhancement for 3D GIS electromagnetic field simulation, the dimensionality transformation kernel Get the three-dimensional difference convolution kernel :

[0121]

[0122]

[0123] Among them, ⊙ represents the corresponding multiplication of elements.

[0124] Using 3D differential convolution kernel Extraction of coarse grid field intensity gradient information (extracting coarse grid electromagnetic field gradient changes to guide fine grid data generation):

[0125]

[0126] Among them, ∗ represents the convolution operation, For application The gradient information data obtained later;

[0127] The coarse grid field intensity data is enhanced by upsampling to obtain the upsampled fine grid field intensity data U (obtaining refined coarse grid data):

[0128]

[0129] in, The upsampling convolution kernel is obtained with network training;

[0130] Combined with the coarse grid field intensity gradient information Enhanced acquisition of preliminary fine-grid field strength data

[0131] .

[0132] This embodiment directly uses the form of the differential algorithm to define the convolution kernel, ensuring that the physical principles can be accurately mapped to the convolution operation. This is achieved by taking the core calculation formula of the differential algorithm and converting it into the form of the convolution kernel, thereby ensuring the correct application of continuous physical laws to discrete data. In addition, this embodiment takes scale transformation into account when designing the differential convolution kernel, allowing the electromagnetic field to be analyzed at different resolutions. By adjusting the size and shape of the convolution kernel, different physical resolutions can be simulated to capture multi-scale features.

[0133] By extending the two-dimensional convolution kernel to three-dimensional space through a certain transformation, the key step of applying the traditional differential method to the three-dimensional GIS electromagnetic field simulation is that this dimensional transformation takes into account the physical effects in each direction in the three-dimensional space. The rotational invariance problem can be solved by designing a symmetrical three-dimensional convolution kernel. A symmetrical kernel will produce a consistent response in all directions, which can solve the isotropy problem in electromagnetic field simulation.

[0134] It should be noted that the special differential convolution operation designed in this implementation is based on a method for solving numerical partial differential equations, such as the five-point difference method, which is often used to approximate the solution of partial differential equations in numerical analysis. In deep learning, traditional convolution layers are usually used to extract features of input data, while special differential convolution kernels can be designed to be more suitable for the simulation of specific physical phenomena, such as electromagnetic field simulation. The design of such convolution kernels is usually based on the discrete approximation of the second-order derivative of the function space in the differential algorithm, so that the network can capture the physical characteristics of the data while learning the data.

[0135] Compared with traditional convolutional layers, differential convolution can maintain high-quality data while ensuring that the data conforms to specific physical laws, making the data generated by the network more physically credible. This combination makes the model perform better when dealing with problems that require precise physical modeling, such as electromagnetic field simulation and climate change simulation.

[0136] Furthermore, the function of the differential convolution module set in this embodiment is to obtain accurate fine grid field strength data, and use the differential convolution module to Coarse grid field strength data of electromagnetic field of GIS equipment Processing is performed to obtain the fine grid enhanced field strength characteristics , the formula is:

[0137]

[0138]

[0139]

[0140] In the formula, is the field intensity feature generated by differential convolution, is the differential convolution operation, is the coarse grid field strength data, is the field strength characteristic after reorganization, Indicates the application of convolution kernels of different sizes. Enhance the field strength characteristics for the fine grid, For the self-attention mechanism operation, is the convolution operation, is the activation function.

[0141] As a further preferred technical solution, the convolution kernel of the differential convolution layer adopts any one of a five-point differential convolution kernel, a weighted differential convolution kernel, a multi-scale differential convolution kernel, a directional differential convolution kernel, a nine-point differential convolution kernel, and a mixed mode differential convolution.

[0142] It should be noted that those skilled in the art may also select other differential algorithms to be combined with the deep learning network according to actual application requirements, and this embodiment does not specifically limit this.

[0143] Specifically, when the convolution kernel of the differential convolution layer adopts a five-point differential convolution kernel, a five-point differential convolution module is formed. The field intensity characteristics generated by the five-point differential convolution are:

[0144]

[0145] Among them, the calculation formula of the field strength feature generated by the five-point differential convolution module can be expanded to:

[0146]

[0147]

[0148]

[0149] In the formula, is the five-point difference convolution kernel, are trainable weights, represents a variable step size, , , , , Represents training parameters.

[0150] Specifically, when the convolution kernel of the differential convolution layer adopts the weighted differential convolution kernel, a weighted differential convolution module is formed. Different from the five-point differential convolution module, the trainable weights are:

[0151]

[0152] In the formula, , , , Represents training parameters.

[0153] Specifically, when the convolution kernel of the differential convolution layer adopts a multi-scale differential convolution kernel, a multi-scale differential convolution module is formed. Different from the five-point differential convolution module, if a small-scale differential convolution kernel is used, the trainable weights are:

[0154]

[0155] If a large-scale convolution kernel is used, the trainable weights are:

[0156]

[0157] In the formula, represents a small-size convolution kernel, Represents a large-size convolution kernel.

[0158] Specifically, when the convolution kernel of the differential convolution layer adopts a directional differential convolution kernel, a directional differential convolution module is formed, which is different from the five-point differential convolution module in that:

[0159]

[0160] In the formula, represents the directional difference convolution kernel, Represents a horizontal element. Represents a vertical element.

[0161] Specifically, when the convolution kernel of the differential convolution layer adopts a nine-point differential convolution kernel, a nine-point differential convolution module is formed, which is different from the five-point differential convolution module in that:

[0162]

[0163] In the formula, Represents the nine-point difference convolution kernel.

[0164] Specifically, when the convolution kernel of the differential convolution layer adopts a mixed mode differential convolution kernel, a mixed mode differential convolution module is formed, which is different from the five-point differential convolution module in that:

[0165]

[0166] In the formula, represents the mixed differential convolution kernel, represents a small-size convolution kernel, represents the directional difference convolution kernel, represents a large-size convolution kernel, , , Represents the scale factor.

[0167] It should be understood that those skilled in the art can explore the combination of the following networks with the proposed differential algorithm according to specific application requirements and objectives: RNN, attention mechanism, graph neural network, Transformer, convolutional neural network, generative adversarial network GAN, etc., which are not specifically limited in this embodiment.

[0168] Taking the Generative Adversarial Network (GAN) as an example, the use of the Generative Adversarial Network (GAN) is to generate high-quality data through an adversarial process. The Generative Adversarial Network consists of two parts: a generator (generating new data instances) and a discriminator (evaluating the authenticity of the data). In this embodiment, the differential method is used in the convolutional network part in this generative adversarial network framework, mainly combined with the convolutional layer of the generator network to achieve spatial feature enhancement of electromagnetic field data, which is used to enhance the feature extraction and generation quality of the data, and to help the model learn and simulate the distribution of electromagnetic field data more accurately.

[0169] It should be noted that the computing framework of the differential convolution module proposed in this embodiment combines strict physical rules (such as the five-point difference method) with advanced deep learning technology, which not only ensures the physical authenticity of the simulation results, but also utilizes the advantages of deep learning in data processing and feature extraction, thereby improving the reliability and application value of the simulation results.

[0170] Furthermore, this embodiment optimizes the computational efficiency and accuracy of the simulation by combining the self-adjusting module and the differential convolution module. The self-adjusting module reduces unnecessary calculations through intelligent enhancement strategies, while the differential convolution module ensures high-precision extraction of field strength data. The two work together to achieve efficient and accurate electromagnetic field simulation. In general, the design and implementation of the self-adjusting module and the differential convolution module greatly improve the accuracy, efficiency and applicability of GIS electromagnetic field simulation through intelligent enhancement, precise feature extraction and efficient calculation methods.

[0171] As a further preferred technical solution, Figure 5 As shown, the structural similarity network includes a first branch network, a second branch network, a second addition operation and a first multilayer perceptron, the outputs of the first branch network and the second branch network are both connected to the second addition operation, and the output of the second addition operation is connected to the first multilayer perceptron , the multilayer perceptron Followed by an activation function .

[0172] Specifically, the first branch network includes a convolutional neural network layer CNN and a batch normalization operation connected in sequence. , the regularization operation Then there is an activation function ;

[0173] The second branch network includes a second multilayer perceptron , the second multi-layer perceptron Then there is an activation function .

[0174] Furthermore, the function of the structural similarity network is to match the enhanced grid data with the field strength data for the first The fine grid structure characteristics of the electromagnetic field of GIS equipment with coarse grid number Enhanced field strength characteristics with fine grids The field strength and structural characteristics of:

[0175]

[0176]

[0177]

[0178]

[0179] In the formula, For the Extraction information of fine grid structure features corresponding to the coarse grid number of electromagnetic field of GIS equipment; For the Extraction information of the fine grid-enhanced field strength characteristics corresponding to the coarse grid number of the electromagnetic field of the GIS equipment; and Respectively represent the fine grid enhanced grid structure feature extraction information Enhanced field strength feature extraction information with fine grid The modulation weight of It is the comprehensive feature of information extraction of fine grid enhanced grid structure feature and fine grid enhanced field strength feature; For similarity characteristics; is the batch normalization operation, is the convolution operation, is a multi-layer perceptron of the structural similarity module, It is the combined characteristic of field strength and structure.

[0180] It should be noted that the functions of the structural similarity network designed in this embodiment are: (1) Preserving physical structure: The structural similarity module ensures that the enhanced data retains its physical accuracy while maintaining structural integrity and coherence by focusing on the structural properties of the data. This is particularly important in the field of simulation because the physical behavior of the electromagnetic field strongly depends on its spatial structural characteristics. (2) Improving the quality of data enhancement: This module helps to generate higher quality data, especially in restoration and super-resolution tasks, and can ensure that the details of the generated data have a high structural similarity with the original coarse grid data. (3) Enhanced feature representation: The structural similarity module provides an effective mechanism to enhance and significantly characterize key features in electromagnetic field data, especially in terms of field strength and grid structure, by fusing features extracted by CNN and MLP.

[0181] As a further preferred technical solution, the step S50: calculating GIS fine grid enhancement data based on the similarity feature specifically includes the following steps:

[0182] S51, calculating regularized coarse grid structure data, regularized coarse grid field intensity data, regularized fine grid enhanced grid structure characteristics and regularized fine grid enhanced field intensity characteristics based on the coarse grid structure data, the coarse grid field intensity data, the fine grid enhanced grid structure characteristics and the fine grid enhanced field intensity characteristics respectively;

[0183] It should be noted that this embodiment specifically uses a field strength consistency module to process the coarse grid structure data, the coarse grid field strength data, the fine grid enhanced grid structure features, the fine grid enhanced field strength features and the similarity features, so as to converge the difference between the fine grid enhanced data and the coarse grid data.

[0184] Use the following formula to calculate the Electromagnetic field fine grid enhancement grid structure characteristics of strip GIS equipment and fine grid enhanced field strength characteristics With GIS equipment electromagnetic field coarse grid data and coarse grid field strength data Fine grid enhanced data after comprehensive compensation :

[0185]

[0186]

[0187]

[0188]

[0189] In the formula, , , and They are regularized electromagnetic field fine grid enhanced grid structure characteristics, regularized fine grid enhanced field strength characteristics, regularized electromagnetic field coarse grid data and regularized coarse grid field strength data; is the matrix 2-norm.

[0190] S52, calculating grid structure consistency based on the regularized coarse grid structure data and the regularized fine grid enhanced grid structure features;

[0191] Specifically, the calculation formula for grid structure consistency is:

[0192]

[0193] In the formula, The grid structure is consistent.

[0194] S53, calculating grid field intensity consistency based on the regularized coarse grid field intensity data and the regularized fine grid enhanced field intensity characteristics;

[0195] Specifically, the calculation formula for grid field strength consistency is:

[0196]

[0197] In the formula, is the grid field strength consistency.

[0198] S54, calculating grid field strength loss compensation based on the grid structure consistency and the grid field strength consistency;

[0199] Specifically, the calculation formula for grid field strength loss compensation is:

[0200]

[0201] In the formula, Compensate for grid field strength loss.

[0202] S55, calculating fine grid enhancement structure data based on the fine grid enhancement grid structure characteristics, the grid field strength loss compensation and the similarity characteristics;

[0203] Specifically, the calculation formula for fine grid enhanced structure data is:

[0204]

[0205] In the formula, Enhance structural data for fine meshes.

[0206] S56, calculating fine grid enhanced field strength data based on the fine grid enhanced field strength characteristics, the grid field strength loss compensation and the similarity characteristics;

[0207] Specifically, the calculation formula for fine grid enhanced field strength data is:

[0208]

[0209] In the formula, Enhance field strength data for fine grids; and is the scale factor.

[0210] As a further preferred technical solution, the differential deep learning network model used in this embodiment is a pre-trained model that can be used to calculate the similarity between the grid structure and the grid field strength. Before the step S20: inputting the GIS electromagnetic field coarse grid simulation data set into the differential deep learning network model, the method further includes the following steps:

[0211] A GIS electromagnetic field coarse grid historical simulation data set is used, and the differential deep learning network model is trained using the historical simulation data set until the total loss function of the model is minimized, wherein the total loss function includes a constraint loss function based on Maxwell's equations and a field strength and structural consistency loss function, and the formula is expressed as:

[0212]

[0213] In the formula, is the total loss function, is the field strength and structure consistency loss function, is the Maxwell constraint loss function, and is the weight parameter.

[0214] It should be noted that this embodiment has carried out a multi-dimensional loss function design. By combining structural similarity loss, field strength consistency loss and physical constraint loss, multiple important characteristics of simulation data are comprehensively considered. The design of this multi-dimensional loss function is beneficial to ensure the accuracy of the simulation while being able to adjust the weights between different losses to adapt to specific simulation needs and optimization goals.

[0215] Furthermore, the formula of the field strength and structure consistency loss function is expressed as:

[0216]

[0217]

[0218]

[0219] In the formula, represents the gradient obtained by applying the difference method, is the structural similarity loss, is the field strength consistency loss, and is the scale factor, Enhance the mesh structure features for fine meshes, is the coarse grid structure data, To enhance the field strength characteristics for fine grids, is the coarse grid field strength data, is the 2-norm of the vector.

[0220] It should be noted that this embodiment ensures that the enhanced electromagnetic field data is highly consistent with the original data in structure and field strength through structural similarity loss and field strength consistency loss. This consistency not only promotes the convergence of simulation data, but also ensures the physical authenticity of simulation results and improves the reliability of simulation.

[0221] And by introducing the scaling factor, a flexible loss function adjustment mechanism is provided. This mechanism can not only adjust the importance of different parts of the loss function according to specific tasks, but also improve the generalization ability and adaptability of the model when processing different types of electromagnetic field data.

[0222] Furthermore, the formula of the constraint loss function based on Maxwell equations is expressed as:

[0223]

[0224]

[0225]

[0226]

[0227]

[0228]

[0229] In the formula, is the magnetic field component, is the Gaussian law loss, is Gauss's magnetic law loss, is the loss due to Faraday's law of electromagnetic induction, is Ampere's law loss, The proportionality constant, , , , is the scale factor, represents the curl, is the vector 2 norm; is the magnetic permeability of vacuum, is the permittivity of vacuum, It is the fine grid enhanced field strength data.

[0230] It should be noted that the physical constraint loss function ensures that the enhanced electromagnetic field data strictly follows the basic laws of electromagnetism through the key components of Maxwell's equations - Gauss's law, Gauss's law of magnetism, Faraday's law of electromagnetic induction, and Ampere's law. This constraint makes the electromagnetic field data output by the model not only numerically accurate, but also strict in physical laws, which increases the scientificity and practicality of the model.

[0231] It should be noted that when the differential method is combined with the generative adversarial network to form a differentially enhanced generative adversarial network, the differential convolution kernel is directly integrated into the generator architecture, thereby introducing differential enhancement of spatial features. The adversarial here is mainly reflected in the loss of the model, that is, when using a generative adversarial network, the total loss function of the model includes not only the physical constraint loss function and the field strength and structural consistency loss function, but also the adversarial loss part.

[0232] In summary, the loss function designed in this embodiment combines the advantages of traditional physical models and deep learning. Through precise mathematical expressions and physical constraints, it helps to optimize the stability and convergence speed of the algorithm and improve training efficiency. The total loss function combines electromagnetic field theory and deep learning technology, which not only enhances the model's ability to process electromagnetic field simulation data, but also ensures the accuracy and physical authenticity of the simulation results, providing an effective optimization strategy for electromagnetic field simulation.

[0233] In addition, if Figure 6 As shown, the second embodiment of the present invention proposes a GIS electromagnetic field coarse grid data enhancement system based on differential GAN, and the system includes:

[0234] The data acquisition module 10 is used to acquire a GIS electromagnetic field coarse grid simulation data set, wherein the GIS electromagnetic field coarse grid simulation data set includes coarse grid structure data and coarse grid field strength data;

[0235] The data enhancement module 20 is used to process the GIS electromagnetic field coarse grid simulation data set using the deployed differential deep learning network model, wherein the differential deep learning network model includes an enhancement network and a structural similarity network connected in sequence, and the enhancement network includes a self-adjusting module and a differential convolution module; the coarse grid structure data and the coarse grid field strength data are calculated using the self-adjusting module and the differential convolution module respectively to obtain fine grid enhanced grid structure features and fine grid enhanced field strength features; the fine grid enhanced grid structure features and the fine grid enhanced field strength features are matched using the structural similarity network to calculate similarity features;

[0236] The data calculation module 30 is used to calculate GIS fine grid enhanced data based on the similarity features.

[0237] It should be noted that the differential deep learning network model in this embodiment is the same as that recorded in the above-mentioned first embodiment, and will not be repeated here.

[0238] As a further preferred technical solution, the data calculation module 30 specifically includes:

[0239] A regularization unit, used for calculating regularized coarse grid structure data, regularized coarse grid field intensity data, regularized fine grid enhanced grid structure characteristics and regularized fine grid enhanced field intensity characteristics based on the coarse grid structure data, the coarse grid field intensity data, the fine grid enhanced grid structure characteristics and the fine grid enhanced field intensity characteristics respectively;

[0240] A structural consistency calculation unit, used for calculating grid structural consistency based on the regularized coarse grid structural data and the regularized fine grid enhanced grid structural features;

[0241] A field intensity consistency calculation unit, used to calculate the grid field intensity consistency based on the regularized coarse grid field intensity data and the regularized fine grid enhanced field intensity characteristics;

[0242] A field strength loss compensation calculation unit, used for calculating grid field strength loss compensation based on the grid structure consistency and the grid field strength consistency;

[0243] A first fine grid enhancement data calculation unit, configured to calculate fine grid enhancement structure data based on the fine grid enhancement grid structure feature, the grid field strength loss compensation and the similarity feature;

[0244] The second fine grid enhanced data calculation unit calculates fine grid enhanced field strength data based on the fine grid enhanced field strength feature, the grid field strength loss compensation and the similarity feature.

[0245] As a further preferred technical solution, the system further includes a model training module for:

[0246] A GIS electromagnetic field coarse grid historical simulation data set is used, and the differential deep learning network model is trained using the historical simulation data set until the total loss function of the model is minimized, wherein the total loss function includes a constraint loss function based on Maxwell's equations and a field strength and structural consistency loss function, and the formula is expressed as:

[0247]

[0248] In the formula, is the total loss function, is the field strength and structure consistency loss function, is the Maxwell constraint loss function, and is the weight parameter.

[0249] It should be noted that the specific representation of the constraint loss function based on Maxwell's equations and the field strength and structure consistency loss function can be found in the introduction of the first embodiment above.

[0250] It should be noted that other embodiments or implementation methods of the GIS electromagnetic field coarse grid data enhancement system based on differential network described in the present invention can refer to the above-mentioned method embodiments and will not be repeated here.

[0251] In addition, the third embodiment of the present invention further proposes a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the GIS electromagnetic field coarse simulation data enhancement method based on differential fusion neural network as described in the first embodiment above is implemented.

[0252] It should be noted that the logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in combination with these instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in combination with these instruction execution systems, devices or apparatuses. More specific examples (non-exhaustive list) of computer-readable media include the following: an electrical connection portion with one or more wirings (electronic device), a portable computer disk box (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium and then editing, interpreting or processing in other suitable ways if necessary, and then stored in a computer memory.

[0253] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0254] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0255] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In the description of the present invention, the meaning of "plurality" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0256] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations of the present invention. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present invention.

Claims

1. A GIS electromagnetic field rough simulation data enhancement method based on differential fusion neural network, characterized in that: The method comprises: Acquire a GIS electromagnetic field coarse grid simulation data set, wherein the GIS electromagnetic field coarse grid simulation data set includes coarse grid structure data and coarse grid field strength data; Input the GIS electromagnetic field coarse grid simulation data set into a differential deep learning network model, wherein the differential deep learning network model includes an enhanced network and a structural similarity network connected in sequence, the enhanced network includes a self-adjusting module and a differential convolution module, wherein the structural similarity network includes a first branch network, a second branch network, a second addition operation and a first multi-layer perceptron, the outputs of the first branch network and the second branch network are both connected to the second addition operation, the output of the second addition operation is connected to the first multi-layer perceptron, and the multi-layer perceptron is followed by an activation function; the first branch network includes a convolutional neural network layer CNN and a batch regularization operation connected in sequence, the regularization operation is followed by an activation function, the second branch network includes a second multi-layer perceptron, and the second multi-layer perceptron is followed by an activation function; The self-adjusting module and the differential convolution module are used to calculate the coarse grid structure data and the coarse grid field strength data respectively to obtain fine grid enhanced grid structure characteristics and fine grid enhanced field strength characteristics; Using the structural similarity network to match the fine grid enhanced grid structure feature and the fine grid enhanced field intensity feature, and calculating the similarity feature, including: using the first branch network to extract the structural feature information of the fine grid enhanced grid structure feature, using the second branch network to extract the field intensity feature information of the fine grid enhanced field intensity feature, using the modulation weights corresponding to the feature information of the fine grid enhanced grid structure feature and the feature information of the fine grid enhanced field intensity feature, calculating the field intensity and structure combination feature, and calculating the similarity feature based on the field intensity and structure combination feature; Based on the similarity features, GIS fine grid enhancement data is calculated.

2. The GIS electromagnetic field rough simulation data enhancement method based on differential fusion neural network as claimed in claim 1 is characterized in that: The self-adjusting module includes a first convolutional layer and a second convolutional layer connected in sequence, and the output features of the first convolutional layer and the output features of the second convolutional layer are output to the activation function layer after a first addition operation; The coarse grid structure data is used as the input of the first convolutional layer, the grid size supplementary information of the coarse grid structure data is used as the input of the second convolutional layer, the bias vector of the coarse grid structure data is used as the input of the first addition operation, and the coarse grid structure data and the grid size supplementary information of the coarse grid structure data are output to the first addition operation via residual connection.

3. The GIS electromagnetic field rough simulation data enhancement method based on differential fusion neural network as claimed in claim 1 is characterized in that: The process formula for calculating the coarse grid structure data using the self-adjusting module is expressed as: In the formula, To enhance the grid structure features of the fine grid, Conv1() represents the grid enhancement convolution operation, Conv2() represents the fine-tuning convolution operation, and U k is the coarse grid structure data, h k Supplementary information for the grid size of the coarse grid structure data, is the deviation vector of the coarse grid structure data, Residual() is the residual link, and sigmoid is the activation function.

4. The GIS electromagnetic field rough simulation data enhancement method based on differential fusion neural network as claimed in claim 1 is characterized in that: The differential convolution module includes a differential convolution layer, a size integration layer, a self-attention mechanism layer and a third convolution layer connected in sequence, and an activation function is connected after the third convolution layer; The coarse grid field intensity data is used as the input of the differential convolution layer, and the differential convolution layer is used to calculate the field intensity features of different sizes of the coarse grid field intensity data by using a differential algorithm; The size integration layer is used to sum the field intensity features of different sizes calculated by the differential convolution layer to obtain the reorganized field intensity features.

5. The GIS electromagnetic field rough simulation data enhancement method based on differential fusion neural network as claimed in claim 1 is characterized in that: The process formula for calculating the coarse grid field strength data using the differential convolution module is expressed as: In the formula, is the field strength feature generated by differential convolution, Conv npt () is the differential convolution operation, I k is the coarse grid field strength data, is the reorganized field strength feature, Conv size () indicates the application of convolution kernels of different sizes. For the fine grid enhanced field strength feature, SelfAttention() is the self-attention mechanism operation, Conv() is the convolution operation, and ReLU is the activation function.

6. The GIS electromagnetic field rough simulation data enhancement method based on differential fusion neural network as claimed in claim 4 is characterized in that: The convolution kernel of the differential convolution layer adopts any one of a five-point differential convolution kernel, a weighted differential convolution kernel, a multi-scale differential convolution kernel, a directional differential convolution kernel, a nine-point differential convolution kernel, and a mixed mode differential convolution.

7. The GIS electromagnetic field rough simulation data enhancement method based on differential fusion neural network as claimed in claim 1 is characterized in that: The using the structural similarity network to match the fine grid enhanced grid structure feature and the fine grid enhanced field strength feature to calculate the similarity feature includes: The structural feature information of the fine grid enhanced grid structure feature is extracted by using the first branch network, and the formula is expressed as: In the formula, is the characteristic information of the fine grid enhanced grid structure feature described in the kth item, The fine grid is used to enhance the grid structure feature, CNN() is a stacked convolution operation, BatchNorm() is a batch normalization operation, and ReLU is an activation function; The second branch network is used to extract the field intensity feature information of the fine grid enhanced field intensity feature, and the formula is expressed as: In the formula, is the characteristic information of the fine grid enhanced field strength characteristics described in the kth item, Enhance the field strength feature for the fine grid, MLP() is the operation performed by the second multi-layer perceptron; The field strength and structure combination feature is calculated by using the modulation weights corresponding to the feature information of the fine grid enhanced grid structure feature and the feature information of the fine grid enhanced field strength feature. The formula is expressed as: In the formula, f k,combined is the field strength and structure combination feature, α and β represent the modulation weights corresponding to the feature information of the fine grid enhanced grid structure feature and the feature information of the fine grid enhanced field strength feature, respectively; Based on the field strength and structure combination characteristics, the similarity characteristics are calculated, and the formula is expressed as: f k,final =ReLU(MLP final (f k,combined )) In the formula, f k,final For the similarity feature, MLP final () is the operation performed by the first multi-layer perceptron.

8. The GIS electromagnetic field rough simulation data enhancement method based on differential fusion neural network as claimed in claim 1 is characterized in that: The step of calculating GIS fine grid enhanced data based on the similarity feature includes: Calculating regularized coarse grid structure data, regularized coarse grid field intensity data, regularized fine grid enhanced grid structure characteristics, and regularized fine grid enhanced field intensity characteristics based on the coarse grid structure data, the coarse grid field intensity data, the fine grid enhanced grid structure characteristics, and the fine grid enhanced field intensity characteristics, respectively; Calculating grid structure consistency based on the regularized coarse grid structure data and the regularized fine grid enhanced grid structure features; Calculating grid field strength consistency based on the regularized coarse grid field strength data and the regularized fine grid enhanced field strength characteristics; Calculating grid field strength loss compensation based on the grid structure consistency and the grid field strength consistency; Calculating fine grid enhancement structure data based on the fine grid enhancement grid structure characteristics, the grid field strength loss compensation and the similarity characteristics; Based on the fine grid enhanced field strength characteristics, the grid field strength loss compensation and the similarity characteristics, fine grid enhanced field strength data is calculated.

9. The GIS electromagnetic field rough simulation data enhancement method based on differential fusion neural network as claimed in claim 1 is characterized in that: Before inputting the GIS electromagnetic field coarse grid simulation data set into the differential deep learning network model, the method further includes: A GIS electromagnetic field coarse grid historical simulation data set is used, and the differential deep learning network model is trained using the historical simulation data set until the total loss function of the model is minimized, wherein the total loss function includes a constraint loss function based on Maxwell's equations and a field strength and structural consistency loss function, and the formula is expressed as: L total =ηL consistency +ζL Maxwell Where, L total is the total loss function, L consistency is the field strength and structure consistency loss function, L Maxwell is the Maxwell constraint loss function, η and ζ are weight parameters.

10. The GIS electromagnetic field rough simulation data enhancement method based on differential fusion neural network as claimed in claim 9, characterized in that: The formula of the field strength and structure consistency loss function is expressed as: L consistency =σL struct +δL field In the formula, represents the gradient obtained by applying the difference method, L struct is the structural similarity loss, L field is the field strength consistency loss, σ and δ are proportional factors, To enhance the mesh structure characteristics of fine mesh, U k is the coarse grid structure data, To enhance the field strength characteristics of fine grid, I k is the coarse grid field strength data, ∥∥ 2 is the 2-norm of the vector.

11. The GIS electromagnetic field rough simulation data enhancement method based on differential fusion neural network according to claim 9 is characterized in that: The formula of the constraint loss function based on Maxwell equations is expressed as: L Maxwell =w1·L Gauss +w2·L Gauss-Mag +w3·L Faraday +w4·L Ampere In the formula, is the magnetic field component, L Gauss is Gaussian law loss, L Gauss-Mag is Gauss's magnetic law loss, L Faraday is the loss according to Faraday’s law of electromagnetic induction, L Ampere is the Ampere's law loss, k is the proportional constant, w1, w2, w3, w4 are proportional factors, represents the curl, ∥·∥ 2 is the vector 2-norm; μ0 is the magnetic permeability of vacuum, ∈0 is the permittivity of vacuum, It is the fine grid enhanced field strength data.

12. A GIS electromagnetic field coarse grid data enhancement system based on differential GAN, characterized in that: The system comprises: A data acquisition module is used to acquire a GIS electromagnetic field coarse grid simulation data set, wherein the GIS electromagnetic field coarse grid simulation data set includes coarse grid structure data and coarse grid field strength data; A data enhancement module is used to process the GIS electromagnetic field coarse grid simulation data set using the deployed differential deep learning network model, wherein the differential deep learning network model includes an enhancement network and a structural similarity network connected in sequence, and the enhancement network includes a self-adjusting module and a differential convolution module; the coarse grid structure data and the coarse grid field strength data are calculated using the self-adjusting module and the differential convolution module respectively to obtain fine grid enhanced grid structure features and fine grid enhanced field strength features; the fine grid enhanced grid structure features and the fine grid enhanced field strength features are matched using the structural similarity network to calculate similarity features, including extracting structural feature information of the fine grid enhanced grid structure features using the first branch network, extracting field strength feature information of the fine grid enhanced field strength features using the second branch network, calculating field strength and structure combination features using modulation weights corresponding to the feature information of the fine grid enhanced grid structure features and the feature information of the fine grid enhanced field strength features, and calculating the similarity features based on the field strength and structure combination features; A data calculation module, used for calculating GIS fine grid enhanced data based on the similarity features; The structural similarity network includes a first branch network, a second branch network, a second addition operation and a first multilayer perceptron, the outputs of the first branch network and the second branch network are both connected to the second addition operation, the output of the second addition operation is connected to the first multilayer perceptron, and the multilayer perceptron is followed by an activation function; the first branch network includes a convolutional neural network layer CNN and a batch normalization operation connected in sequence, the regularization operation is followed by an activation function, the second branch network includes a second multilayer perceptron, and the second multilayer perceptron is followed by an activation function.

13. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 11 is implemented.

Images