Method, device and equipment for determining stress distribution of transformer winding and storage medium

Abstract

The application relates to a stress distribution determination method, device and equipment of a transformer winding and a storage medium. The method comprises the following steps: in the case that a target transformer is detected to be faulty, the hoop stress and bending stress of a target winding of the target transformer are acquired, and the short-circuit current of the target transformer is acquired; the short-circuit current, the hoop stress and the bending stress are input into a distribution determination model to obtain the target stress distribution of different regions in the target winding. The method can improve the determination efficiency of the stress distribution of the transformer winding.

Description

Technical Field

[0001] This application relates to the field of transformer technology, and in particular to a method, apparatus, device, and storage medium for determining the stress distribution of transformer windings. Background Technology

[0002] With the continuous expansion of power system scale and the continuous improvement of voltage levels, the operational stability of transformers is crucial to the safety of power systems. When a transformer experiences a short-circuit fault, its internal windings are prone to deformation, misalignment, instability, or even collapse under the transient electromagnetic force caused by the short circuit. Therefore, timely and accurate acquisition of stress field data of transformer windings under short-circuit conditions is of great significance for assessing the transformer's short-circuit withstand capability and for fault early warning.

[0003] Traditional techniques rely on finite element simulation to model the electromagnetic and structural field distributions under different short-circuit conditions in order to calculate the stress field data of the transformer's internal windings.

[0004] However, this method consumes a lot of time and computing power for a single simulation, and a single simulation can only be applied to the current short-circuit condition. Therefore, this method of determining stress distribution has technical problems of low efficiency and difficulty in meeting real-time requirements. Summary of the Invention

[0005] Therefore, it is necessary to provide a method, apparatus, device, and storage medium for determining the stress distribution of transformer windings, which can improve the efficiency of determining the stress distribution of transformer windings and meet real-time requirements.

[0006] In a first aspect, this application provides a method for determining the stress distribution of a transformer winding, including:

[0007] In the event that a fault is detected in the target transformer, the circumferential pressure and bending stress of the target winding of the target transformer, as well as the short-circuit current of the target transformer, are obtained.

[0008] The short-circuit current, circumferential pressure, and bending stress are input into the distribution determination model to obtain the target stress distribution in different regions of the target winding.

[0009] In one embodiment, short-circuit current, circumferential pressure, and bending stress are input into a distribution determination model to obtain the target stress distribution in different regions of the target winding. This includes: inputting short-circuit current, circumferential pressure, and bending stress into the distribution determination model to obtain a target stress distribution image corresponding to the target winding; wherein, the pixel value of each pixel in the target stress distribution image is used to characterize the stress of the corresponding region of the pixel in the target winding.

[0010] In one embodiment, the distribution determination model is trained by: acquiring different training samples; wherein the short-circuit conditions of the sample transformers corresponding to the different training samples are not exactly the same, each training sample includes the sample short-circuit current of any sample transformer, the sample circumferential pressure and sample bending stress of the sample winding of the sample transformer, and the simulated stress distribution of the sample winding; the simulated stress distribution in each training sample is determined by fault simulation of the transformer short-circuit simulation model under the short-circuit condition corresponding to the training sample; using different training samples, the initial model is trained to obtain the distribution determination model.

[0011] In one embodiment, different training samples are used to train an initial model to obtain a distribution-determined model, including: inputting different training samples into the initial model to obtain the predicted stress distribution corresponding to each training sample; determining a first loss value of the initial model based on the predicted stress distribution and simulated stress distribution of each training sample; if the first loss value does not meet a preset convergence condition, determining different candidate hyperparameters and inputting different training samples into an intermediate model corresponding to each candidate hyperparameter to obtain a second loss value of each intermediate model; adjusting the parameters of the initial model based on the candidate hyperparameter corresponding to the intermediate model with the smallest second loss value, and returning to execute the operation of inputting different training samples into the initial model until the first loss value meets the preset convergence condition to obtain a distribution-determined model.

[0012] In one embodiment, obtaining the circumferential pressure and bending stress of the target winding of the target transformer includes: obtaining the material parameters and rated electrical parameters of the target transformer; determining the magnetic parameters of the target winding subjected to a fault based on the material parameters, rated electrical parameters, and short-circuit current; and determining the circumferential pressure and bending stress of the target winding of the target transformer based on the magnetic parameters.

[0013] In one embodiment, the bending stress of the target winding includes radial bending stress and axial bending stress of the winding.

[0014] Secondly, this application also provides a device for determining the stress distribution of a transformer winding, comprising:

[0015] The acquisition module is used to acquire the circumferential pressure and bending stress of the target winding of the target transformer, as well as the short-circuit current of the target transformer, when a fault is detected in the target transformer.

[0016] The determination module is used to input short-circuit current, circumferential pressure, and bending stress into the distribution determination model to obtain the target stress distribution in different regions of the target winding.

[0017] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the various method embodiments of the first aspect described above.

[0018] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps in the method embodiments of the first aspect described above.

[0019] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps in the method embodiments of the first aspect described above.

[0020] The aforementioned method, apparatus, equipment, and storage medium for determining the stress distribution of transformer windings, upon detecting a fault in the target transformer, first acquires the circumferential pressure and bending stress of the target winding, as well as the short-circuit current of the target transformer. Then, the short-circuit current, circumferential pressure, and bending stress are input into the stress distribution determination model to obtain the target stress distribution in different regions of the target winding. This eliminates the need to repeatedly construct complex magnetic-force coupling models and perform iterative solutions to simulate the electromagnetic and structural field distributions under different short-circuit conditions. Thus, while ensuring the accuracy of the obtained target stress distribution, computational resources are effectively saved, enabling rapid perception of the transformer winding stress distribution and providing reliable data support for real-time transformer status monitoring and fault early warning. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a diagram illustrating the application environment of a method for determining the stress distribution of transformer windings in one embodiment.

[0023] Figure 2 This is a flowchart illustrating a method for determining the stress distribution of a transformer winding in one embodiment;

[0024] Figure 3 This is a schematic diagram of the process for determining circumferential pressure and bending stress in one embodiment;

[0025] Figure 4 This is a flowchart illustrating the process of training a distribution-determining model in one embodiment;

[0026] Figure 5 This is a flowchart illustrating the process of training the distribution determination model in another embodiment;

[0027] Figure 6 This is a flowchart illustrating a method for determining the stress distribution of a transformer winding in another embodiment;

[0028] Figure 7 This is a structural block diagram of a device for determining the stress distribution of a transformer winding in one embodiment;

[0029] Figure 8 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0030] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0031] It should be noted that the terms "first," "second," etc., used in this application can be used to describe various elements, but these elements are not limited by these terms. These terms are only used to distinguish the first element from the second element. The terms "comprising" and "having," and any variations thereof, used in this application, are intended to cover non-exclusive inclusion. The term "multiple" used in this application refers to two or more. The term "and / or" used in this application refers to one of the embodiments or any combination of multiple embodiments.

[0032] The method for determining the stress distribution of transformer windings provided in this application embodiment can be applied to, for example... Figure 1 In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be integrated onto server 104, or it can be located in the cloud or on other network servers. Terminal 102 can be, but is not limited to, a device or equipment for collecting the circumferential pressure and bending stress of transformer windings, as well as the short-circuit current of the transformer. Terminal 102 sends the collected circumferential pressure and bending stress of the transformer windings, as well as the short-circuit current of the transformer, to the data storage system for storage. This allows server 104 to directly read the circumferential pressure and bending stress of the transformer windings and the short-circuit current of the transformer from the data storage system when determining the stress distribution of the transformer windings. Server 104 can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services.

[0033] In one exemplary embodiment, such as Figure 2 As shown, a method for determining the stress distribution of a transformer winding is provided, which can be applied to... Figure 1 Taking server 104 as an example, the following steps are included:

[0034] S201, in the event that a fault is detected in the target transformer, acquire the circumferential pressure and bending stress of the target winding of the target transformer, as well as the short-circuit current of the target transformer.

[0035] Among them, the short-circuit current is the inrush current flowing through the target winding of the target transformer when a fault occurs, reflecting the severity of the fault and the magnitude of the excitation of the magnetic parameters of the target winding. The circumferential pressure is the pressure exerted on the target winding along the circumferential direction under the action of electromagnetic force, reflecting the degree of compression of the target winding. Bending stress reflects the degree of bending deformation of the target winding under the action of electromagnetic force and can include multiple dimensions.

[0036] Optionally, when the fault detection system of the target transformer detects a fault in the target transformer, it can send a fault alarm signal. Therefore, after receiving the fault alarm signal, the short-circuit current can be collected from the current sensor installed on the target transformer, and the circumferential pressure and bending stress of the target winding can be calculated based on the short-circuit current, the current fault condition, and the model of the target transformer.

[0037] Optionally, the bending stress of the target winding includes the radial bending stress and the axial bending stress of the winding.

[0038] Radial bending stress characterizes the stress generated when the conductor of the target winding bends in a plane perpendicular to the axial direction under the action of the radial component of the electromagnetic force, and is related to radial instability of the target winding (such as conductor bending or warping). Axial bending stress characterizes the stress generated when the conductor of the target winding bends along the height direction under the action of the axial component of the electromagnetic force, and is related to axial instability of the target winding (such as coil tilting or compression).

[0039] S202, short-circuit current, circumferential pressure and bending stress are input into the distribution determination model to obtain the target stress distribution in different regions of the target winding.

[0040] Optionally, the distribution determination model can be a neural network model. Specifically, the obtained short-circuit current, circumferential pressure, and bending stress can be input into the distribution determination model to perform feature extraction and spatial mapping of the short-circuit current, circumferential pressure, and bending stress, and output the set of stress values ​​corresponding to different regions in the target winding, i.e., the target stress distribution.

[0041] The aforementioned method for determining the stress distribution of transformer windings, upon detecting a fault in the target transformer, first acquires the circumferential pressure and bending stress of the target winding, as well as the short-circuit current. Then, the short-circuit current, circumferential pressure, and bending stress are input into the stress distribution determination model to obtain the target stress distribution in different regions of the target winding. This eliminates the need to repeatedly construct complex magnetic-force coupling models and perform iterative solutions to simulate the electromagnetic and structural field distributions under different short-circuit conditions. Thus, while ensuring the accuracy of the obtained target stress distribution, it effectively saves computational resources, enabling rapid perception of the transformer winding stress distribution and providing reliable data support for real-time transformer status monitoring and fault early warning.

[0042] Based on the above embodiments, in an exemplary embodiment, in S202 above, short-circuit current, circumferential pressure and bending stress are input into the distribution determination model to obtain the target stress distribution in different regions of the target winding, including: inputting short-circuit current, circumferential pressure and bending stress into the distribution determination model to obtain the target stress distribution image corresponding to the target winding.

[0043] In the target stress distribution image, the pixel value of each pixel is used to characterize the stress in the corresponding region of the target winding. The pixel value of each pixel in the target stress distribution image has a linear mapping relationship with the stress in the corresponding region of the target winding. For example, the larger the pixel value, the higher the stress in the corresponding region.

[0044] Optionally, the distribution determination model may include cascaded input layers, convolutional layers, and output layers. The convolutional layers can be constructed based on GRU (Gated Recurrent Unit). Specifically, the input layer is used to embed short-circuit current, circumferential pressure, and bending stress through numerical encoding. The convolutional layer is used to extract spatial features and scale fusion of the embedded short-circuit current, circumferential pressure, and bending stress to obtain target features. The output layer is used to perform regression processing on the target features output by the convolutional layer to map the target features into a target stress distribution image output.

[0045] In this embodiment, a target stress distribution image corresponding to the target winding is obtained by determining the distribution model of short-circuit current, circumferential pressure, and bending stress inputs. Pixel values ​​are then used to characterize the stress magnitude in the corresponding region of the target winding. This effectively presents a visual representation of the stress distribution in different regions of the target winding while ensuring the accuracy of the obtained target stress distribution. This facilitates rapid identification of stress concentration areas based on the target stress distribution image, thereby improving the engineering practicality of the target stress distribution.

[0046] Based on the above embodiments, in an exemplary embodiment, such as Figure 3As shown, obtaining the circumferential pressure and bending stress of the target winding of the target transformer in S201 above includes the following steps:

[0047] S301, obtain the material parameters and rated electrical parameters of the target transformer.

[0048] The material parameters of the target transformer include parameters related to the material and electromagnetic characteristics of the target components. The rated electrical parameters of the target transformer are the benchmark parameters related to electrical performance specified at the factory.

[0049] Optionally, rated electrical parameters may include rated voltage ratio, rated capacity, rated current on the high-voltage side, and rated current on the medium-voltage side. Target components may include an iron core, target winding, oil tank and support bars, pads, laminated paperboard, etc. Correspondingly, material parameters may include the elastic modulus, Poisson's ratio, and density of the target winding, as well as the elastic modulus, Poisson's ratio, and density of the pads.

[0050] Optionally, the material parameters and rated electrical parameters of the target transformer can be read directly from the data storage system.

[0051] S302, based on material parameters, rated electrical parameters and short-circuit current, determine the magnetic parameters of the target winding under fault conditions.

[0052] Optionally, the electromagnetic parameters may include magnetic flux density. The current density Js of the target winding can be determined based on the short-circuit current, material parameters and rated electrical parameters. Then, the vector magnetic potential A of the target winding under fault action can be solved according to the transient magnetic field equation shown in the following formula (1).

[0053] (1)

[0054] Where v is the magnetoresistive force and σ is the conductivity. For curl operator.

[0055] Optionally, the magnetic flux density B can be determined based on the calculated vector magnetic potential A, where B = ×A.

[0056] Furthermore, based on the structure of the target winding, a two-dimensional structural model corresponding to the target winding can be established, and the two-dimensional structural model can be meshed to obtain the magnetic flux density of each grid.

[0057] S303, based on magnetic parameters, determines the circumferential pressure and bending stress of the target winding of the target transformer.

[0058] Optionally, the radial bending stress corresponding to each grid can be calculated according to the following formula (2). .

[0059] (2)

[0060] in, Let I be the magnetic flux density of the m-th grid cell in the radial direction, I be the short-circuit current, and N be the number of coil turns of the target winding.

[0061] Accordingly, the axial bending stress corresponding to each grid can be calculated according to the following formula (3). .

[0062] (3)

[0063] in, Let be the magnetic flux density of the m-th grid cell along the axis.

[0064] Optionally, the radial bending stress of the target winding can be obtained from the radial bending stress corresponding to each grid cell in the radial direction, and the axial bending stress of the target winding can be obtained from the axial bending stress corresponding to each grid cell in the axial direction.

[0065] Furthermore, the annular pressure of the target winding can be calculated based on the axial bending stress and radial bending stress of the target winding.

[0066] In this embodiment, the material parameters and rated electrical parameters of the target transformer are first obtained. Then, the magnetic parameters of the magnetic field acting on the target winding are determined by combining the short-circuit current. Finally, the circumferential pressure and bending stress of the target winding are calculated based on the magnetic parameters. This eliminates the need for on-site measurement of complex dynamic electromagnetic parameters and for solving the circumferential pressure and bending stress of the winding using full-dimensional magnetic-force coupled finite element simulation technology. This reduces the difficulty of data acquisition and calculation while ensuring the reliability and accuracy of the calculated circumferential pressure and bending stress results, saving computational resources and time costs. It provides scientific and reliable data support for the accurate output of the target stress distribution of the target winding by the distribution determination model.

[0067] Based on the above embodiments, in an exemplary embodiment, such as Figure 4 As shown, the distribution determination model is trained through the following steps:

[0068] S401, obtain different training samples.

[0069] Among them, the short-circuit conditions of the sample transformers corresponding to different training samples are not exactly the same. Each training sample includes the sample short-circuit current of any sample transformer, the sample circumferential pressure and sample bending stress of the sample winding of the sample transformer, and the simulated stress distribution of the sample winding. The simulated stress distribution in each training sample is determined by fault simulation of the transformer short-circuit simulation model under the short-circuit condition corresponding to the training sample.

[0070] For example, finite element simulation can be used to perform fault simulation based on the transformer short-circuit simulation model under the short-circuit condition corresponding to the training sample, and obtain the simulated stress distribution in each training sample.

[0071] Optionally, identical short-circuit conditions indicate that the sample transformers have identical short-circuit types, load rates, and operating states. The short-circuit types can include three-phase short circuits, single-phase ground faults, and inter-turn short circuits. The operating states include high-voltage to low-voltage operation, high-voltage to medium-voltage operation, and medium-voltage to low-voltage operation. The load rate can be in the range of 5%-110%.

[0072] S402 uses different training samples to train the initial model and obtain a distribution-determined model.

[0073] For each training sample, the sample short-circuit current of the sample transformer, the sample circumferential pressure of the sample winding of the sample transformer, and the sample bending stress can be input into the initial model to obtain the initial stress distribution. Then, based on the initial stress distribution and simulated stress distribution corresponding to each training sample, the loss value of the initial model is calculated. If the loss value does not meet the predetermined convergence condition, the parameters of the initial model are adjusted, and a new loss value of the initial model is calculated until the loss value meets the preset convergence condition. The current initial model is then used as the distribution determination model.

[0074] In this embodiment, training samples covering different short-circuit conditions are acquired. The sample short-circuit current, sample winding circumferential pressure, and sample bending stress are used as model inputs. The simulated stress distribution obtained from the transformer short-circuit simulation model under the corresponding short-circuit condition is used as a label to train the initial model, resulting in a distribution-determining model. This eliminates the need to construct a separate stress prediction model for each short-circuit condition and avoids repeatedly performing time-consuming magnetic-force coupling simulation calculations during the actual inference stage. This effectively improves the model's generalization ability while ensuring the accuracy and physical rationality of the model's learning of the mapping law between short-circuit characteristics and winding stress distribution. This allows the model to adapt to the stress distribution prediction needs under different short-circuit conditions, thereby achieving efficient and universal prediction of transformer winding stress distribution under multiple conditions. This provides a reliable data source for real-time monitoring and fault early warning of the transformer's full-condition operation.

[0075] Based on the above embodiments, in an exemplary embodiment, such as Figure 5 As shown, S402 above includes the following steps:

[0076] S501, different training samples are input into the initial model to obtain the predicted stress distribution corresponding to different training samples.

[0077] Optionally, for each training sample, the sample short-circuit current of the sample transformer, the sample circumferential pressure of the sample winding of the sample transformer, and the sample bending stress can be input into the initial model to obtain the predicted stress distribution corresponding to the training sample.

[0078] S502, based on the predicted stress distribution and simulated stress distribution of each training sample, determines the first loss value of the initial model.

[0079] Optionally, for each training sample, a first intermediate loss value for stress distribution prediction based on the current hyperparameters of the initial model can be calculated according to the predicted stress distribution and simulated stress distribution corresponding to the training sample. This first intermediate loss value can be the result of weighted summation of the function value of the mean squared error loss function and the function value of the structural similarity index loss function.

[0080] Furthermore, the arithmetic mean of the first intermediate loss values ​​corresponding to each training sample can be used as the first loss value of the initial model.

[0081] S503, if the first loss value does not meet the preset convergence condition, different candidate hyperparameters are determined, and different training samples are input into the intermediate model corresponding to each candidate hyperparameter to obtain the second loss value of each intermediate model.

[0082] Optionally, the preset convergence conditions may include a first loss value greater than a preset threshold and / or the rate of change of the first loss value relative to the previous iteration being less than a preset rate of change.

[0083] Hyperparameters can include the number of convolutional layers in the model, kernel size, learning rate, batch size, weight ratio of the loss function, and number of channels in the regression head.

[0084] Optionally, if the first loss value does not meet the preset convergence condition, multiple sets of different initial candidate hyperparameters can be generated using the Tent mapping (piecewise linear mapping) to ensure that the generated multiple sets of initial candidate hyperparameters are uniformly distributed in the hyperparameter search space.

[0085] Alternatively, multiple sets of different initial candidate hyperparameters can be represented as matrix X in equation (4).

[0086] (4)

[0087] Where X is an initial candidate hyperparameter set consisting of multiple different candidate hyperparameters. Each row of the matrix contains a set of initial candidate hyperparameters, that is, X includes d initial candidate hyperparameters. For any set of initial candidate hyperparameters, each column corresponds to a hyperparameter of one dimension, that is, it includes a total of n dimensions of hyperparameters.

[0088] Furthermore, the fitness corresponding to each set of initial hyperparameters can be expressed as a matrix Fx as shown in equation (5).

[0089] (5)

[0090] In this matrix Fx, each row represents the fitness of the corresponding group of initial candidate hyperparameters. Higher fitness indicates faster convergence when using the corresponding initial candidate hyperparameters for model parameter tuning. Based on this, discoverers, scouts, and followers can be selected from the initial candidate hyperparameter set according to the fitness of each group. Specifically, the initial candidate hyperparameters are sorted in descending order of fitness. All candidate hyperparameters in the sorted results before the first digit are considered discoverers, used for a wide-ranging search to update the initial candidate hyperparameters. All candidate hyperparameters in the sorted results between the second and first digits are considered scouts, used to avoid getting trapped in local optima during the hyperparameter search. The number of scouts accounts for 10% to 20% of the initial candidate hyperparameter set. All other candidate hyperparameters besides discoverers and followers are considered followers to improve search accuracy.

[0091] Optionally, the hyperparameters of the j-th dimension obtained in the (t+1)-th iteration during the hyperparameter search process of the i-th discoverer. It can be expressed as the following formula (6).

[0092] (6)

[0093] Where α is an adjustable parameter used to control the attenuation amplitude, and it is usually located in the (0,1) interval. Let be the maximum number of iterations for the i-th discoverer, Q be a random number following a normal distribution, and L be a 1*d matrix. The detection value is the value reported by the investigator. The smaller the detection value, the lower the risk of getting trapped in a local optimum. ST is the preset detection value threshold.

[0094] Optionally, the hyperparameters of the j-th dimension obtained in the (t+1)-th iteration during the hyperparameter search process of the i-th discoverer. It can be expressed as the following formula (6).

[0095] Optionally, the hyperparameters of the j-th dimension obtained by the p-th follower after the (t+1)-th iteration. It can be expressed as the following formula (7).

[0096] (7)

[0097] in, The follower with the lowest fitness. E represents the follower with the highest fitness, and is a random vector.

[0098] For scout updates, when the fitness of a scout hyperparameter combination is greater than the fitness of the globally optimal hyperparameter combination, the corresponding dimension value of that hyperparameter combination will be updated to the product of the corresponding dimension value of the globally optimal hyperparameter combination in the current iteration plus a random coefficient following a normal distribution and the absolute value of the difference between the current hyperparameter value and the globally optimal hyperparameter value. When the fitness of a scout hyperparameter combination is equal to the fitness of the globally optimal hyperparameter combination, the corresponding dimension value of that hyperparameter combination will be updated to the product of the current value plus a random coefficient and the absolute value of the difference between the current hyperparameter value and the globally worst hyperparameter value, divided by the quotient obtained by adding a very small positive number to the difference between the current fitness and the globally worst fitness. The very small positive number is used to avoid the mathematical error of the denominator being zero. Through the above update rules, the poorly performing scout hyperparameter combinations can be made to move closer to the globally optimal region, and the best performing scout hyperparameter combinations can undergo fine mutation near the optimal region, thereby improving the global search capability and convergence stability of the hyperparameter optimization process.

[0099] Based on this, the initial candidate hyperparameter set can be updated to obtain the final candidate hyperparameters. Furthermore, for each set of candidate hyperparameters, one or more corresponding intermediate models can be constructed. In addition, the sample short-circuit current of the sample transformer, the sample circumferential pressure of the sample winding of the sample transformer, and the sample bending stress of the sample transformer corresponding to different training samples can be input into the intermediate model to obtain the second loss value corresponding to the intermediate model.

[0100] S504, based on the candidate hyperparameters corresponding to the intermediate model with the smallest second loss value, adjust the parameters of the initial model, and return to execute the operation of inputting different training samples into the initial model until the first loss value meets the preset convergence condition, so as to obtain a distribution-determined model.

[0101] Optionally, the second loss value can also be the result of a weighted sum of the function values ​​of the mean squared error loss function and the structural similarity index loss function. Therefore, the larger the value of the second loss value, the lower the accuracy of the stress distribution prediction of the intermediate model corresponding to the second loss value.

[0102] Based on this, it can be determined whether the minimum second loss value meets the preset convergence condition. If the minimum second loss value meets the preset convergence condition, the candidate hyperparameters of the intermediate model corresponding to the minimum second loss value can be determined as the target hyperparameters. Based on the target hyperparameters, the parameters of the initial model are adjusted, and the operation of inputting different training samples into the initial model is returned until the first loss value meets the preset convergence condition to obtain the distribution determination model.

[0103] If the minimum second loss value does not meet the preset convergence condition, the process can be repeated to determine the candidate hyperparameters until the obtained minimum second loss value meets the preset convergence condition.

[0104] In this embodiment, the predicted stress distribution is obtained by inputting different training samples into the initial model. A first loss value is calculated based on the predicted results and the simulated stress distribution to quantify the accuracy of the initial model. When the first loss value does not meet the convergence condition, an intermediate model is constructed using multiple sets of candidate hyperparameters, and the hyperparameter with the smallest loss is selected to adjust the parameters of the initial model. This eliminates the need for manual subjective judgment of model training effectiveness and manual adjustment of hyperparameters. Thus, while ensuring the prediction accuracy of the model with a defined distribution, it effectively achieves automatic selection of model hyperparameters and standardization of the training process. This avoids prediction bias caused by poor hyperparameter adaptability, thereby accelerating model convergence, reducing iteration rounds and computational resource consumption, and providing a reliable training mechanism for quickly obtaining a high-performance winding stress distribution prediction model.

[0105] Figure 6 This is a flowchart illustrating a method for determining the stress distribution of a transformer winding in another embodiment. Based on the above embodiments, this embodiment provides an optional example of a method for determining the stress distribution of a transformer winding. (Combined with...) Figure 6 The specific implementation process is as follows:

[0106] S601, in the event of a fault detected in the target transformer, acquires the material parameters and rated electrical parameters of the target transformer, as well as the short-circuit current of the target transformer.

[0107] S602, based on material parameters, rated electrical parameters and short-circuit current, determine the magnetic parameters of the target winding under fault conditions.

[0108] S603, based on magnetic parameters, determines the circumferential pressure and bending stress of the target winding of the target transformer.

[0109] S604 inputs short-circuit current, circumferential pressure, and bending stress into the distribution determination model to obtain the target stress distribution image corresponding to the target winding.

[0110] Specifically, the distribution determination model is trained in the following way: different training samples are obtained; wherein, the short-circuit conditions of the sample transformers corresponding to different training samples are not exactly the same, each training sample includes the sample short-circuit current of any sample transformer, the sample circumferential pressure and sample bending stress of the sample winding of the sample transformer, and the simulated stress distribution of the sample winding; the simulated stress distribution in each training sample is determined by fault simulation of the transformer short-circuit simulation model under the short-circuit condition corresponding to the training sample; the initial model is trained using different training samples to obtain the distribution determination model. The initial model is trained using different training samples to obtain a distribution-determined model. This process includes: inputting different training samples into the initial model to obtain the predicted stress distribution corresponding to each training sample; determining the first loss value of the initial model based on the predicted stress distribution and simulated stress distribution of each training sample; determining different candidate hyperparameters if the first loss value does not meet the preset convergence condition, and inputting different training samples into the intermediate model corresponding to each candidate hyperparameter to obtain the second loss value of each intermediate model; adjusting the parameters of the initial model based on the candidate hyperparameter corresponding to the intermediate model with the smallest second loss value, and then returning to execute the operation of inputting different training samples into the initial model until the first loss value meets the preset convergence condition to obtain the distribution-determined model.

[0111] The specific processes of S601-S604 described above can be found in the description of the above method embodiments. Their implementation principles and technical effects are similar, and will not be repeated here.

[0112] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages in other steps. It is understood that the steps in different embodiments can be freely combined as needed, and all non-contradictory solutions formed by such combinations are within the scope of protection of this application.

[0113] Based on the same inventive concept, this application also provides a stress distribution determination device for transformer windings to implement the stress distribution determination method for transformer windings described above. The solution provided by this device is similar to the solution described in the above method; therefore, the specific limitations of one or more embodiments of the stress distribution determination device for transformer windings provided below can be found in the limitations of the stress distribution determination method for transformer windings described above, and will not be repeated here.

[0114] In one exemplary embodiment, such as Figure 7 As shown, a device for determining the stress distribution of a transformer winding is provided, comprising: an acquisition module 710 and a determination module 720, wherein:

[0115] The acquisition module 710 is used to acquire the circumferential pressure and bending stress of the target winding of the target transformer, as well as the short-circuit current of the target transformer, when a fault is detected in the target transformer.

[0116] The determination module 720 is used to input short-circuit current, circumferential pressure and bending stress into the distribution determination model to obtain the target stress distribution in different regions of the target winding.

[0117] In one embodiment, the acquisition module 710 is specifically used to acquire the material parameters and rated electrical parameters of the target transformer; determine the magnetic parameters of the target winding subjected to the fault based on the material parameters, rated electrical parameters and short-circuit current; and determine the circumferential pressure and bending stress of the target winding of the target transformer based on the magnetic parameters.

[0118] In one embodiment, the determining module 720 is specifically used to input short-circuit current, circumferential pressure and bending stress into the distribution determining model to obtain a target stress distribution image corresponding to the target winding; wherein, the pixel value of each pixel in the target stress distribution image is used to characterize the stress of the corresponding region of the pixel in the target winding.

[0119] The various modules in the aforementioned transformer winding stress distribution determination device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of a computer device in hardware form or independent of it, or stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each module.

[0120] In one exemplary embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 8As shown, the computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores the circumferential pressure and bending stress of the target winding of the target transformer, as well as the short-circuit current of the target transformer. The I / O interfaces are used for information exchange between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements a method for determining the stress distribution of a transformer winding.

[0121] Those skilled in the art will understand that Figure 8 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0122] In one embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above method embodiments.

[0123] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the steps in the above method embodiments.

[0124] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0125] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0126] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0127] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for determining the stress distribution of a transformer winding, characterized in that, The method includes: In the event that a fault is detected in the target transformer, the circumferential pressure and bending stress of the target winding of the target transformer, as well as the short-circuit current of the target transformer, are obtained. The short-circuit current, the circumferential pressure, and the bending stress are input into the distribution determination model to obtain the target stress distribution in different regions of the target winding.

2. The method according to claim 1, characterized in that, The step of inputting the short-circuit current, the circumferential pressure, and the bending stress into the distribution determination model to obtain the target stress distribution in different regions of the target winding includes: The short-circuit current, the circumferential pressure, and the bending stress are input into the distribution determination model to obtain the target stress distribution image corresponding to the target winding; In the target stress distribution image, the pixel value of each pixel is used to characterize the stress of the corresponding region of the pixel in the target winding.

3. The method according to claim 1 or 2, characterized in that, The distribution determination model is trained in the following manner: Different training samples are obtained; wherein the short-circuit conditions of the sample transformers corresponding to different training samples are not exactly the same. Each training sample includes the sample short-circuit current of any sample transformer, the sample circumferential pressure and sample bending stress of the sample winding of the sample transformer, and the simulated stress distribution of the sample winding; the simulated stress distribution in each training sample is determined by fault simulation of the transformer short-circuit simulation model under the short-circuit condition corresponding to the training sample. The initial model is trained using the different training samples to obtain a distribution-determined model.

4. The method according to claim 3, characterized in that, The step of training the initial model using the different training samples to obtain a distribution-determined model includes: The different training samples are input into the initial model to obtain the predicted stress distribution corresponding to each training sample; Based on the predicted stress distribution and simulated stress distribution of each training sample, the first loss value of the initial model is determined; If the first loss value does not meet the preset convergence condition, different candidate hyperparameters are determined, and the different training samples are input into the intermediate model corresponding to each candidate hyperparameter to obtain the second loss value of each intermediate model. Based on the candidate hyperparameters corresponding to the intermediate model with the smallest second loss value, the parameters of the initial model are adjusted, and the operation of inputting different training samples into the initial model is returned until the first loss value satisfies the preset convergence condition to obtain a distribution-determined model.

5. The method according to claim 1 or 2, characterized in that, The process of obtaining the circumferential pressure and bending stress of the target winding of the target transformer includes: Obtain the material parameters and rated electrical parameters of the target transformer; Based on the material parameters, the rated electrical parameters, and the short-circuit current, determine the magnetic parameters of the target winding subjected to the fault. Based on the magnetic parameters, the circumferential pressure and bending stress of the target winding of the target transformer are determined.

6. The method according to claim 1 or 2, characterized in that, The bending stress of the target winding includes the radial bending stress and the axial bending stress of the winding.

7. A device for determining the stress distribution of a transformer winding, characterized in that, The device includes: The acquisition module is used to acquire the circumferential pressure and bending stress of the target winding of the target transformer, as well as the short-circuit current of the target transformer, when a fault is detected in the target transformer. The determination module is used to input the short-circuit current, the circumferential pressure, and the bending stress into the distribution determination model to obtain the target stress distribution in different regions of the target winding.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images