Oil-immersed transformer hot-spot temperature prediction method, system, device and storage medium

Abstract

The application discloses an oil-immersed transformer hot spot temperature prediction method, system, device and storage medium, and comprises the following steps: a dynamic simulation model is established according to the multi-physical field thermal state of the oil-immersed transformer, and dynamic thermal simulation is performed on the dynamic simulation model under a preset operating condition to obtain simulation operation data; a hot spot temperature mapping model is established, the simulation operation data is taken as a training data set to train the hot spot temperature mapping model, and the hot spot temperature mapping model comprises a gated recurrent unit and a deep neural network; real-time operation data of the oil-immersed transformer is acquired, the real-time operation data is input into the trained hot spot temperature mapping model for temperature prediction, and the winding hot spot temperature of the oil-immersed transformer is obtained. The intelligent learning model is established by using the transformer data obtained by the multi-physical field simulation, more accurate transformer hot spot temperature can be obtained, and the accuracy of the hot spot temperature inversion is improved by taking the key temperature measurement area of the oil tank shell as an important parameter.

Description

Technical Field

[0001] This invention relates to the field of oil-immersed transformer technology, and in particular to a method, system, device, and storage medium for predicting hot spot temperature in oil-immersed transformers. Background Technology

[0002] Oil-immersed power transformers are crucial equipment in power grid transmission and transformation systems, and their operational safety and reliability directly impact the safety and stability of the power grid. For transformers, winding hot spot temperature is a vital operational indicator, determining their load capacity and directly affecting the degree of insulation aging. Accurate measurement of winding temperature is of paramount importance for accurately assessing transformer load capacity and service life.

[0003] Due to the high internal electric field, large current, and harsh and complex operating conditions of transformers, the hot spot temperature of transformer windings is difficult to measure directly by installing temperature sensors. Instead, it is mainly assessed and predicted through indirect calculation methods. Among existing methods for assessing transformer winding hot spot temperature, the empirical thermal model in the IEC 60076-7 standard guideline is a commonly used method. This model mainly sums the results of two differential equations, considering both the temperature rise of the hot spot before the oil flow change and the impact of the oil flow change on the hot spot temperature. In actual online monitoring, the above differential equations can be discretized into difference equations for rapid solution, thereby achieving real-time dynamic assessment of the hot spot temperature.

[0004] Calculating winding hot spot temperatures using IEC empirical thermal models requires numerous parameters. While some parameters can be accurately obtained through factory tests and hot-running tests, others, such as winding index, oil index, and thermal constant, are difficult to obtain precisely. Often, only rough estimates can be made by consulting IEC guidelines based on the transformer capacity and cooling type, resulting in discrepancies with the actual transformer values. Therefore, the accuracy of using IEC empirical thermal models to calculate hot spot temperatures in practical applications is limited. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention provides a method, system, device, and storage medium for predicting hot spot temperatures in oil-immersed transformers. It utilizes a large amount of dynamic data on the transformer's operating status obtained through multi-physics simulation to establish an intelligent learning model, enabling dynamic tracking and online monitoring of transformer hot spot temperatures. Furthermore, by analyzing the transformer's oil flow lines, it selects key temperature measurement areas on the tank casing as important parameters for hot spot temperature inversion, thereby improving the accuracy of hot spot temperature inversion.

[0006] In a first aspect, embodiments of the present invention provide a method for predicting the hot spot temperature of an oil-immersed transformer, comprising:

[0007] Based on the multi-physics thermal state of the oil-immersed transformer, a dynamic simulation model of the oil-immersed transformer is established, and dynamic thermal simulation is performed on the dynamic simulation model under preset operating conditions to obtain the simulation operation data of the oil-immersed transformer.

[0008] A hot spot temperature mapping model for the oil-immersed transformer is established, and the simulation operation data is used as a training dataset to train the hot spot temperature mapping model. The hot spot temperature mapping model includes a gated recurrent unit and a deep neural network.

[0009] The real-time operating data of the oil-immersed transformer is obtained, and the real-time operating data is input into the trained hot spot temperature mapping model for temperature prediction to obtain the winding hot spot temperature of the oil-immersed transformer. The real-time operating data includes load current, ambient temperature, top oil temperature, and average temperature of the top of the casing on both sides of the oil tank.

[0010] Furthermore, the step of establishing a dynamic simulation model of the oil-immersed transformer based on its multi-physics thermal state includes:

[0011] Based on the temperature field and fluid field of the oil-immersed transformer, a dynamic simulation model of the oil-immersed transformer is established; the temperature field includes core loss and winding loss, and the fluid field includes internal heat conduction, oil convection, air convection, and thermal radiation of the winding core;

[0012] The core loss is obtained from the factory test of the oil-immersed transformer.

[0013] The winding loss is calculated using the following formula:

[0014] P cu =P d +P e

[0015]

[0016]

[0017] In the formula, P d and P e P represents DC loss and eddy current loss, respectively. d_rated and P e_rated These represent the DC loss and eddy current loss under rated current, respectively; M represents the load factor; and T` represents the winding hot spot temperature.

[0018] The internal heat conduction of the winding core is calculated using the following formula:

[0019]

[0020] In the formula, Q1 represents heat flow, Δ represents the Laplace operator, T` represents the temperature difference between surfaces, and R represents thermal resistance;

[0021] The oil convection is calculated using the following formula:

[0022]

[0023] In the formula, t represents time, ρ represents fluid density, u represents fluid velocity, p and F represent pressure and volume force, μ and C represent fluid dynamic viscosity and specific heat capacity, T represents fluid temperature, k represents thermal conductivity, Q2 represents heat source, K represents viscous force, ▽ represents Hamiltonian operator, Δ represents Laplace operator, and I represents identity matrix;

[0024] The air-side convection is calculated using the following formula:

[0025] Q3=Ah(T s -T a )

[0026] In the formula, Q3 represents the convective heat loss, A represents the surface area, h represents the convective heat transfer coefficient, and T represents the heat transfer coefficient. s and T a These represent the outer surface temperature of the oil tank of an oil-immersed transformer and the ambient temperature, respectively.

[0027] The thermal radiation is calculated using the following formula:

[0028]

[0029] In the formula, Q4 represents the thermal radiation radiated from the box wall into the air, η represents the Stefan-Boltzmann constant, and E represents the surface emissivity.

[0030] Furthermore, the step of performing dynamic thermal simulation on the dynamic simulation model under preset operating conditions to obtain the simulation operating data of the oil-immersed transformer includes:

[0031] Based on the preset operating conditions of the oil-immersed transformer, the simulated load current and simulated ambient temperature of the oil-immersed transformer are set.

[0032] Dynamic thermal simulation is performed on the dynamic simulation model based on the simulated load current and the simulated ambient temperature to obtain the simulated top oil temperature, simulated winding hot spot temperature, and simulated average temperature of the top of the shells on both sides of the oil-immersed transformer.

[0033] Furthermore, the gated recurrent unit includes an input layer and a reset update layer, and the deep neural network includes a combination layer, a first hidden layer, a second hidden layer, and an output layer;

[0034] The input layer, the combination layer, the first hidden layer, and the second hidden layer each have 6 neurons, the reset and update layer has 6 time steps, and the output layer has 1 neuron.

[0035] Further, the step of acquiring the real-time operating data of the oil-immersed transformer, inputting the operating data into the trained hot spot temperature mapping model for temperature prediction, and obtaining the winding hot spot temperature of the oil-immersed transformer includes:

[0036] The historical winding hot spot temperature and real-time operating data of the oil-immersed transformer are obtained. The historical winding hot spot temperature is input into the input layer and then subjected to gated loop through the reset and update layer to obtain the first output matrix.

[0037] The first output matrix and the real-time running data are concatenated to obtain the first input matrix;

[0038] The first input matrix is ​​input into the combined layer, and linear calculations are performed sequentially through the first hidden layer and the second hidden layer. The winding hot spot temperature of the oil-immersed transformer is output through the output layer.

[0039] Further, the step of concatenating the first output matrix and the real-time running data to obtain the first input matrix includes:

[0040] The first output matrix and the real-time running data are concatenated using the concat function to obtain the first input matrix; wherein, the real-time running data includes a first sub-running data and a second sub-running data, the first sub-running data being the current running data of the oil-immersed transformer, and the second sub-running data being the previous running data of the oil-immersed transformer;

[0041] The first input matrix is ​​represented by the following formula:

[0042] X' = ​​concat(H, F)

[0043] In the formula, H represents the first output matrix, and F represents the real-time running data.

[0044] Furthermore, the first sub-operational data includes the first load current, ambient temperature, first top oil temperature, and average temperature of the top of the casing on both sides of the oil tank; the second sub-operational data includes the second load current and the second top oil temperature.

[0045] Secondly, embodiments of the present invention provide a hot spot temperature prediction system for an oil-immersed transformer, comprising:

[0046] The dynamic simulation model is used to establish a dynamic simulation model of the oil-immersed transformer based on the multi-physics thermal state of the oil-immersed transformer, and to perform dynamic thermal simulation on the dynamic simulation model under preset operating conditions to obtain the simulation operation data of the oil-immersed transformer.

[0047] A hotspot temperature mapping model is used to establish a hotspot temperature mapping model for the oil-immersed transformer. The simulation operation data is used as a training dataset to train the hotspot temperature mapping model. The hotspot temperature mapping model includes a gated recurrent unit and a deep neural network.

[0048] The hot spot temperature prediction module is used to acquire the real-time operating data of the oil-immersed transformer, input the real-time operating data into the trained hot spot temperature mapping model to predict the temperature, and obtain the winding hot spot temperature of the oil-immersed transformer. The real-time operating data includes load current, ambient temperature, top oil temperature, and average temperature of the top of the casing on both sides of the oil tank.

[0049] Thirdly, embodiments of the present invention also provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described method.

[0050] Fourthly, embodiments of the present invention also provide a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the above-described method.

[0051] This invention provides a method, system, device, and storage medium for predicting hot spot temperature of oil-immersed transformers. Compared with existing technologies, this invention utilizes data on the dynamic changes in the operating status of transformers obtained through multi-physics simulation to establish an intelligent learning model, which dynamically tracks and monitors the hot spot temperature of the transformer online. This enables more accurate prediction of the transformer hot spot temperature under dynamic conditions. Furthermore, by analyzing the oil flow lines of the transformer, key temperature measurement areas on the tank shell are selected as important parameters for hot spot temperature inversion, further improving the accuracy of hot spot temperature inversion. Attached Figure Description

[0052] Figure 1 This is a flowchart illustrating the hot spot temperature prediction method for oil-immersed transformers in an embodiment of the present invention.

[0053] Figure 2 This is a schematic diagram of the simulation results of the transformer oil streamline trajectory in an embodiment of the present invention;

[0054] Figure 3 This is a schematic diagram of the hotspot temperature mapping model in an embodiment of the present invention;

[0055] Figure 4 yes Figure 2 Schematic diagram of the middle layer 2;

[0056] Figure 5 This is a schematic diagram of the structure of the dynamic simulation model established in the embodiments of the present invention;

[0057] Figure 6 This is a schematic diagram of data obtained through dynamic simulation model in an embodiment of the present invention;

[0058] Figure 7 This is a schematic diagram of the hot spot temperature prediction system for an oil-immersed transformer in an embodiment of the present invention;

[0059] Figure 8 This is an internal structural diagram of the computer device in an embodiment of the present invention. Detailed Implementation

[0060] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0061] Please see Figure 1 The first embodiment of the present invention proposes a method for predicting the hot spot temperature of an oil-immersed transformer, comprising steps S10 to S30:

[0062] Step S10: Based on the multi-physics thermal state of the oil-immersed transformer, establish a dynamic simulation model of the oil-immersed transformer, and perform dynamic thermal simulation on the dynamic simulation model under preset operating conditions to obtain the simulation operation data of the oil-immersed transformer.

[0063] In existing methods for calculating winding hot spot temperatures using IEC empirical thermal models, the required parameters such as winding index, oil index, and thermal constant are difficult to obtain accurately. Often, only rough estimates can be made by consulting IEC guidelines based on the transformer capacity and cooling type, resulting in discrepancies with the actual transformer values. To address the limited accuracy of using IEC empirical thermal models for hot spot temperature calculations in practical applications, this invention establishes a dynamic simulation model of an oil-immersed transformer to simulate more accurate parameter data for subsequent model training.

[0064] The dynamic simulation model of this invention is constructed based on the multi-physics thermal state of an oil-immersed transformer. The simulation model includes multiple components such as windings, core, oil, tank shell, and heat sink. The tank and heat sink shell are constructed with multiple layers of materials, with paint layers on the inside and outside and iron plate layers in the middle. This can greatly reduce the number of meshes while realistically simulating actual heat transfer, thereby improving the calculation speed. At the same time, the oil channel support bars between the high and low windings are taken into account to form heat dissipation oil channels. The material property parameters of the simulation model are set according to the relevant material parameters of the actual transformer, so that the simulation results of hot spot temperature are more consistent with reality.

[0065] The dynamic simulation model of this invention involves physical fields including temperature and fluid fields. The temperature field mainly consists of core loss and winding loss, which are the primary sources of heat inside the transformer. Core loss under rated operation can be directly obtained from the transformer's factory test, while winding loss is affected by the load current. The winding loss P... cu The following formula can be used for calculation:

[0066] P cu =P d +P e

[0067]

[0068] In the formula, P d and P e P represents DC loss and eddy current loss, respectively. d_rated and P e_rated These represent the DC loss and eddy current loss under rated current, respectively. M represents the load factor, and T` represents the winding hot spot temperature.

[0069] The fluid field mainly considers the heat transfer process of the transformer, which can be divided into three stages: heat conduction inside the windings and core, oil convection, and air convection and thermal radiation. Among these, heat conduction inside the windings and core follows the basic laws of heat transfer, namely:

[0070]

[0071] In the formula, Q1 represents heat flow, Δ represents the Laplace operator, T` represents the temperature difference between surfaces, and R represents thermal resistance.

[0072] Oil convection circulation is the main pathway for heat transfer inside the transformer. It transfers heat from the windings and core surface to the tank surface. The oil flow distribution characteristics satisfy the Navier-Stokes equations, namely:

[0073]

[0074] In the formula, t represents time, ρ represents fluid density, u represents fluid velocity, p and F represent pressure and volume force, μ and C represent fluid dynamic viscosity and specific heat capacity, T represents fluid temperature, k represents thermal conductivity, Q2 represents heat source, K represents viscous force, ▽ represents Hamiltonian operator, Δ represents Laplace operator, and I represents identity matrix;

[0075] The transformer oil tank and heat sink exchange heat with the outside air through convection and thermal radiation. The heat dissipation through air convection satisfies the following equation:

[0076] Q3=Ah(T s -T a )

[0077] In the formula, Q3 represents the convective heat loss, A represents the surface area, h represents the convective heat transfer coefficient, and T represents the heat transfer coefficient. s and T a These represent the outer surface temperature of the oil tank of an oil-immersed transformer and the ambient temperature, respectively.

[0078] According to the law of radiation, the heat radiated from the box wall into the air satisfies the following governing equation:

[0079]

[0080] In the formula, Q4 represents the thermal radiation radiated from the box wall into the air, η represents the Stefan-Boltzmann constant, and E represents the surface emissivity, where η = 5.67 × 10⁻⁸ W / (m²·K⁴).

[0081] After establishing the dynamic simulation model, the temperature distribution inside the transformer tank can be obtained by iteratively solving the above equations in the finite element simulation, thereby determining the location and temperature of the transformer hot spots.

[0082] To ensure more accurate prediction of hotspot temperatures, the selection of parameters is crucial. Therefore, this invention employs simulation analysis of the streamline trajectory inside the fuel tank, and the simulation results are as follows: Figure 2 As shown, the main flow paths of transformer oil windings in the tank after temperature rise can be summarized based on the oil flow trajectory diagram. Taking an oil-immersed self-cooled ONAN transformer as an example, the analysis results show that the temperature rise at the top of the tank casing and the connection point at the top of the heat sink is relatively high, and these areas carry a large amount of winding temperature information, showing a strong correlation with the hot spot temperature. Therefore, selecting the temperature regions at the top of both sides of the casing as the key areas for casing temperature measurement, and using them as the main parameters for hot spot temperature inversion, can improve the accuracy of the prediction results. These parameters can be obtained using an infrared thermometer.

[0083] After establishing the dynamic simulation model of the oil-immersed transformer, the dynamic thermal simulation under different operating conditions can be performed to obtain the transient change law of the winding hot spot temperature under load variation and different ambient temperatures. The specific steps are as follows:

[0084] Step S101: Based on the preset operating conditions of the oil-immersed transformer, set the simulated load current and simulated ambient temperature of the oil-immersed transformer.

[0085] Step S102: Perform dynamic thermal simulation on the dynamic simulation model based on the simulated load current and the simulated ambient temperature to obtain the simulated top oil temperature, simulated winding hot spot temperature and the average temperature of the top of the shells on both sides of the simulated oil tank of the oil-immersed transformer.

[0086] First, the operating conditions for the rise and fall of the transformer load current are set, generally including the main load range of normal transformer operation, such as 0.2 times the rated current to 1.0 times the rated current. This divides the load range into multiple different load variation ranges, thus covering the main operating conditions of the transformer. In this embodiment of the invention, the load range is divided into 24 ranges, expressed as multiples of the rated current, namely: 0.2 to 0.3, 0.3 to 0.4, 0.4 to 0.5, 0.5 to 0.6, 0.6 to 0.7, 0.7 to 0.8, 0.8 to 0.9, 0.9 to 1.0, and 0.2 to 0.4, 0.3 to 0.5, 0.4 to 0.6, 0.5 to 0.7, 0.6 to 0.8, 0.7 to 0.9, 0.8 to 1.0, 0.2 to 0.5, 0.3 to 0.6, 0.4 to 0.7, 0.5 to 0.8, 0.6 to 0.9, 0.7 to 1.0. It should be noted that the simulation data settings given in this embodiment are only a preferred method and not a specific limitation. The specific simulation data can be flexibly set according to the actual situation.

[0087] Then set the boundary conditions of the dynamic simulation model, i.e., the ambient temperature. Select a range of load current variation from the above ranges and set the simulation time step to 1 minute. For example, select the range where the load current increases from 0.8 times the rated current to 1.0 times the rated current. Calculate the winding loss before and after the load current change. Then substitute the loss value into the dynamic simulation model. First, simulate and run the calculation for 5 hours under a load factor of 0.8, and then simulate and run the calculation for 5 hours under a load factor of 1. Record and save the load current, ambient temperature, and the top oil temperature, winding hot spot temperature, and average temperature values ​​of the top of the two sides of the tank obtained from the simulation every minute.

[0088] The next step is to modify the ambient temperature of the transformer. For example, in this embodiment, the ambient temperature can be set to 10°, 20°, 30° and 40° respectively. Repeat the above steps until the simulation of all operating conditions under all ambient temperatures, including the 24 ranges of load current increase and the 24 ranges of load current decrease, is completed. The simulation data under all operating conditions is obtained, namely the simulated load current, ambient temperature, top oil temperature, winding hot spot temperature and the average temperature of the top of the casing on both sides of the oil tank. This simulation data will be used as the model dataset for subsequent applications.

[0089] Step S20: Establish a hot spot temperature mapping model for the oil-immersed transformer, and use the simulation running data as a training dataset to train the hot spot temperature mapping model. The hot spot temperature mapping model includes a gated recurrent unit and a deep neural network.

[0090] Step S30: Obtain the real-time operating data of the oil-immersed transformer, input the real-time operating data into the trained hot spot temperature mapping model for temperature prediction, and obtain the winding hot spot temperature of the oil-immersed transformer. The real-time operating data includes load current, ambient temperature, top oil temperature, and average temperature of the top of the casing on both sides of the oil tank.

[0091] The network model established in this invention, reflecting the mapping relationship between measurable parameters and hotspot temperatures, is constructed using gated recurrent units combined with a deep neural network. For the established hotspot temperature mapping model, the aforementioned simulation data is used as the training dataset for training. Since the model has the same structure and dimensions as the actual transformer and considers in detail the heat conduction inside the windings, oil convection, air-side heat convection, and surface heat radiation processes of the tank, it can accurately calculate the temperature distribution at various locations in the tank and the hotspot temperature variation curve over time. Therefore, the transient temperature change data of the transformer under different load variations and different ambient temperatures obtained from simulation calculations are also accurate and reliable. Using this data to train the network model to estimate the hotspot temperature has high accuracy. The following section, in conjunction with the appendix... Figure 3 The structure of the hotspot temperature mapping model and the data processing procedure are explained.

[0092] The hotspot temperature mapping model of the present invention is composed of a gated recurrent unit and a deep neural network. The gated recurrent unit includes an input layer and a reset update layer, namely layer 1 and layer 2. The deep neural network includes a combination layer, a first hidden layer, a second hidden layer and an output layer, namely layer 3, layer 4, layer 5 and layer 6.

[0093] The reason for using a gated recurrent unit (ROU) is to solve the gradient vanishing or exploding problem in standard RNNs while preserving long-term information of the sequence. In this embodiment, the input data of the gated recurrent unit is the historical winding hot spot temperature of the oil-immersed transformer. Since the input layer has 6 neurons, the 6 historical winding hot spot temperatures are used to form an input matrix X, which is then input into the input layer. After passing through the gated loop of the reset and update gates in the reset and update layer, the first output matrix H is obtained. Please refer to [link to relevant documentation]. Figure 4 The time step in the reset update layer is 6, and the output matrix at step t can be represented as: h t =GRU(h t-1 ,X t ), t∈[1,6], X t This represents the element at time step t in the input matrix X. The specific data processing in the reset and update layer can be referred to the general data processing process of GRU, and will not be explained in detail here.

[0094] In addition to the output matrix of the gated recurrent unit, the input data of the combined layer of the deep neural network also includes other operating data of the oil-immersed transformer, including the current and previous load current, the current and previous top oil temperature, the average temperature of the top of the tank casings on both sides, and the ambient temperature. These data can be obtained through dynamic monitoring of the transformer. Since the load current and top oil temperature data are relatively easy to obtain in real time, more accurate measured values ​​can be substituted into them. However, the average temperature of the top of the tank casings on both sides and the ambient temperature data are obtained less frequently, and in practice, interpolation is more often used to obtain intermediate values, resulting in relatively larger input errors. Therefore, the load current and top oil temperature data are used from two consecutive data sets to increase the input weight, while the average temperature of the top of the tank casings on both sides and the ambient temperature data are only used from the current data set. That is, by inputting more accurate data values, the accuracy of the prediction results can be improved.

[0095] In this embodiment, the running data can be represented by matrix F. Then, the concat function is used to combine the output matrix of the gated recurrent unit and the running data matrix F to obtain the input matrix X` of the combined layer, i.e., X` = concat(H,F). The first and second hidden layers of the deep neural network are both fully connected layers with 6 neurons each. The output matrices after linear calculation are H`1 and H`2, respectively. Finally, Y is output by the output layer. p The winding hotspot temperature, used for the final inversion prediction, involves data processing by a deep neural network, including the activation function and classifier, which can be referenced from conventional DNN network models and will not be elaborated further here. The hotspot temperature mapping model established in this invention, combining a gated recurrent unit and a deep neural network, can capture dependencies with large time step distances in the time series, thereby improving the accuracy of hotspot temperature prediction when hotspot temperatures are dynamically changing.

[0096] The following example uses a 10kV experimental transformer with distributed fiber optic temperature sensors installed in its internal windings to experimentally verify the hotspot temperature prediction method provided in this invention. Please refer to [link / reference needed]. Figure 5 The multiphysics dynamic simulation model shown is based on a physical transformer. Taking the typical operating conditions of the transformer as an example, the transformer was operated at 0.5 times its rated current for 4 hours, then at a load factor of 1.0 for 2 hours, then at a load factor of 0.8 for 2 hours, and finally back to 0.5 times its rated current for 4 hours. After a large number of simulations in each load range, the load current, ambient temperature, top oil temperature, hot spot temperature, and average temperature of the top of the casing on both sides of the tank were recorded and saved every minute, resulting in a total of 115,200 sets of data. Typical data obtained from the simulation are as follows. Figure 6 As shown in the figure. Then, these simulation data are used as data samples to construct and train a hotspot temperature mapping model based on a GRU-DNN network. This hotspot temperature mapping model is then used to plot the winding hotspot temperature curves of the oil-immersed transformer, and compared with the hotspot temperature curves measured by actual fiber optic temperature sensors and those calculated using IEC. According to the experimental data, the root mean square error (RMSE) between the hotspot temperature curve plotted using the hotspot temperature mapping model provided by this invention and the hotspot temperature curve measured by the actual fiber optic temperature sensor is 0.55℃, with a maximum absolute error of 1.9℃. In contrast, the RMS error between the hotspot temperature curve calculated using IEC and the hotspot temperature curve measured by the actual fiber optic temperature sensor is 2.88℃, with a maximum absolute error of 6.4℃. Therefore, this method can improve the accuracy of online assessment of transformer hotspot temperature under dynamic load changes, providing effective support for accurately assessing the dynamic load capacity of transformers.

[0097] The hot spot temperature prediction method for oil-immersed transformers provided in this invention differs from traditional methods, which can only roughly estimate parameters based on transformer capacity and cooling type by consulting IEC guidelines, often resulting in deviations from the actual transformer values. This invention utilizes a large amount of dynamic data on transformer operating status changes obtained from multi-physics simulations to establish an intelligent learning model. This model not only enables dynamic tracking and online monitoring of transformer hot spot temperatures but also, because the transient temperature change data of transformers under different load variations and ambient temperature conditions obtained from simulation calculations are accurate and reliable, the prediction model trained using these data also has high accuracy in predicting hot spot temperatures.

[0098] Please see Figure 7 Based on the same inventive concept, the second embodiment of the present invention provides a hot spot temperature prediction system for an oil-immersed transformer, comprising:

[0099] The dynamic simulation model 10 is used to establish a dynamic simulation model of the oil-immersed transformer based on the multi-physics thermal state of the oil-immersed transformer, and to perform dynamic thermal simulation on the dynamic simulation model under preset operating conditions to obtain the simulation operation data of the oil-immersed transformer.

[0100] Hot spot temperature mapping model 20 is used to establish the hot spot temperature mapping model of the oil-immersed transformer. The simulation operation data is used as a training dataset to train the hot spot temperature mapping model. The hot spot temperature mapping model includes a gated recurrent unit and a deep neural network.

[0101] The hot spot temperature prediction module 30 is used to acquire the real-time operating data of the oil-immersed transformer, input the real-time operating data into the trained hot spot temperature mapping model for temperature prediction, and obtain the winding hot spot temperature of the oil-immersed transformer. The real-time operating data includes load current, ambient temperature, top oil temperature, and average temperature of the top of the casing on both sides of the oil tank.

[0102] The technical features and effects of the oil-immersed transformer hotspot temperature prediction system proposed in this embodiment of the invention are the same as those of the method proposed in this embodiment of the invention, and will not be repeated here. Each module in the above-mentioned oil-immersed transformer hotspot temperature prediction system can be implemented entirely or partially through software, hardware, or a combination thereof. Each module can be embedded in or independent of the processor in a computer device in hardware form, or it can be 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.

[0103] Please see Figure 8 The diagram illustrates the internal structure of a computer device in one embodiment. This computer device can specifically be a terminal or a server. The computer device includes a processor, memory, network interface, display, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The network interface of the computer device 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 predicting the hot spot temperature of an oil-immersed transformer. The display screen of the computer device can be a liquid crystal display (LCD) or an e-ink display. The input devices of the computer device can be a touch layer covering the display screen, buttons, a trackball, or a touchpad mounted on the computer device casing, or an external keyboard, touchpad, or mouse, etc.

[0104] Those skilled in the art will understand that Figure 8The 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 computing devices may include more or fewer components than shown in the diagram, or combine certain components, or have the same component arrangement.

[0105] Furthermore, embodiments of the present invention also propose a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described method.

[0106] Furthermore, embodiments of the present invention also propose a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the above-described method.

[0107] In summary, this invention provides a method, system, device, and storage medium for predicting hot spot temperature of an oil-immersed transformer. The method involves establishing a dynamic simulation model of the oil-immersed transformer based on its multi-physics thermal state, and performing dynamic thermal simulation on the model under preset operating conditions to obtain simulation operation data. A hot spot temperature mapping model is then established, and the simulation operation data is used as a training dataset to train the model. This model includes a gated recurrent unit and a deep neural network. Real-time operating data of the oil-immersed transformer is acquired and input into the trained hot spot temperature mapping model for temperature prediction, resulting in the winding hot spot temperature of the oil-immersed transformer. This real-time operating data includes load current, ambient temperature, top oil temperature, and the average temperature of the top of both sides of the tank casing. This invention utilizes data on the dynamic changes in the operating status of transformers obtained through multiphysics simulation to establish an intelligent learning model. This model dynamically tracks and monitors the hot spot temperature of the transformer online, enabling more accurate acquisition of the hot spot temperature under dynamic conditions. Furthermore, by analyzing the transformer's oil flow lines, key temperature measurement areas on the tank shell are selected as important parameters for hot spot temperature inversion, further improving the accuracy of hot spot temperature inversion.

[0108] The various embodiments in this specification are described in a progressive manner. For directly identical or similar parts of the embodiments, refer to each other. Each embodiment focuses on its differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments. It should be noted that the technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as the combination of these technical features does not contradict each other, it should be considered within the scope of this specification.

[0109] The embodiments described above are merely preferred embodiments of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various improvements and substitutions without departing from the technical principles of this invention, and these improvements and substitutions should also be considered within the scope of protection of this application. Therefore, the scope of protection of this patent application should be determined by the scope of the claims.

Claims

1. A method for predicting the hot spot temperature of an oil-immersed transformer, characterized in that, include: Based on the multi-physics thermal state of the oil-immersed transformer, a dynamic simulation model of the oil-immersed transformer is established, and dynamic thermal simulation is performed on the dynamic simulation model under preset operating conditions to obtain the simulation operation data of the oil-immersed transformer; the physical fields include temperature field and fluid field, the temperature field includes core loss and winding loss, and the fluid field includes internal heat conduction of the winding core, oil convection, air convection, and thermal radiation. A hot spot temperature mapping model for the oil-immersed transformer is established, and the simulation operation data is used as a training dataset to train the hot spot temperature mapping model. The hot spot temperature mapping model includes a gated recurrent unit and a deep neural network. The real-time operating data of the oil-immersed transformer is obtained, and the real-time operating data is input into the trained hot spot temperature mapping model for temperature prediction to obtain the winding hot spot temperature of the oil-immersed transformer. The real-time operating data includes load current, ambient temperature, top oil temperature, and average temperature of the top of the casing on both sides of the oil tank. The gated recurrent unit includes an input layer and a reset update layer, and the deep neural network includes a combination layer, a first hidden layer, a second hidden layer, and an output layer. The number of neurons in the input layer, the combination layer, the first hidden layer, and the second hidden layer is 6 each; the number of time steps in the reset and update layer is 6; and the number of neurons in the output layer is 1. The steps of acquiring the real-time operating data of the oil-immersed transformer, inputting the operating data into the trained hot spot temperature mapping model for temperature prediction, and obtaining the winding hot spot temperature of the oil-immersed transformer include: The historical winding hot spot temperature and real-time operating data of the oil-immersed transformer are obtained. The historical winding hot spot temperature is input into the input layer and then subjected to gated loop through the reset and update layer to obtain the first output matrix. The first output matrix and the real-time running data are concatenated to obtain the first input matrix; The first input matrix is ​​input into the combination layer, and linear calculations are performed sequentially through the first hidden layer and the second hidden layer. The winding hot spot temperature of the oil-immersed transformer is output through the output layer. The step of concatenating the first output matrix and the real-time running data to obtain the first input matrix includes: The first output matrix and the real-time running data are concatenated using the concat function to obtain the first input matrix; wherein, the real-time running data includes a first sub-running data and a second sub-running data, the first sub-running data being the current running data of the oil-immersed transformer, and the second sub-running data being the previous running data of the oil-immersed transformer; The first sub-operational data includes the first load current, ambient temperature, first top oil temperature, and average temperature of the top of the casing on both sides of the oil tank. The second sub-operational data includes the second load current and the second top oil temperature.

2. The method for predicting the hot spot temperature of an oil-immersed transformer according to claim 1, characterized in that, The core loss is obtained from the factory test of the oil-immersed transformer; The winding loss is calculated using the following formula: P cu =P d +P e In the formula, P d and P e P represents DC loss and eddy current loss, respectively. d_rated and P e_rated These represent the DC loss and eddy current loss under rated current, respectively; M represents the load factor; and T` represents the winding hot spot temperature. The internal heat conduction of the winding core is calculated using the following formula: In the formula, Q1 represents heat flow, Δ represents the Laplace operator, T` represents the temperature difference between surfaces, and R represents thermal resistance; The oil convection is calculated using the following formula: In the formula, t represents time, ρ represents fluid density, u represents fluid velocity, p and F represent pressure and volume force, μ and C represent fluid dynamic viscosity and specific heat capacity, T represents fluid temperature, k represents thermal conductivity, Q2 represents heat source, K represents viscous force, ▽ represents Hamiltonian operator, Δ represents Laplace operator, and I represents identity matrix; The air-side convection is calculated using the following formula: Q3=Ah(T s -T a ) In the formula, Q3 represents the convective heat loss, A represents the surface area, h represents the convective heat transfer coefficient, and T represents the heat transfer coefficient. s and T a These represent the outer surface temperature of the oil tank of an oil-immersed transformer and the ambient temperature, respectively. The thermal radiation is calculated using the following formula: In the formula, Q4 represents the thermal radiation radiated from the box wall into the air, η represents the Stefan-Boltzmann constant, and E represents the surface emissivity.

3. The method for predicting the hot spot temperature of an oil-immersed transformer according to claim 1, characterized in that, The step of performing dynamic thermal simulation on the dynamic simulation model under preset operating conditions to obtain the simulation operating data of the oil-immersed transformer includes: Based on the preset operating conditions of the oil-immersed transformer, the simulated load current and simulated ambient temperature of the oil-immersed transformer are set. Dynamic thermal simulation is performed on the dynamic simulation model based on the simulated load current and the simulated ambient temperature to obtain the simulated top oil temperature, simulated winding hot spot temperature, and simulated average temperature of the top of the shells on both sides of the oil-immersed transformer.

4. The method for predicting the hot spot temperature of an oil-immersed transformer according to claim 1, characterized in that, The first input matrix is ​​represented by the following formula: X' = ​​concat(H, F) In the formula, H represents the first output matrix, and F represents the real-time running data.

5. A hot spot temperature prediction system for an oil-immersed transformer, characterized in that, The system is applied to the method as described in any one of claims 1 to 4, comprising: The dynamic simulation model is used to establish a dynamic simulation model of the oil-immersed transformer based on the multi-physics thermal state of the oil-immersed transformer, and to perform dynamic thermal simulation on the dynamic simulation model under preset operating conditions to obtain the simulation operation data of the oil-immersed transformer. A hotspot temperature mapping model is used to establish a hotspot temperature mapping model for the oil-immersed transformer. The simulation operation data is used as a training dataset to train the hotspot temperature mapping model. The hotspot temperature mapping model includes a gated recurrent unit and a deep neural network. The gated recurrent unit includes an input layer and a reset update layer, and the deep neural network includes a combination layer, a first hidden layer, a second hidden layer, and an output layer; the number of neurons in the input layer, the combination layer, the first hidden layer, and the second hidden layer is 6, the time step of the reset update layer is 6, and the number of neurons in the output layer is 1. The hot spot temperature prediction module is used to acquire the real-time operating data of the oil-immersed transformer, input the real-time operating data into the trained hot spot temperature mapping model to predict the temperature, and obtain the winding hot spot temperature of the oil-immersed transformer. The real-time operating data includes load current, ambient temperature, top oil temperature, and average temperature of the top of the casing on both sides of the oil tank. The step of acquiring the real-time operating data of the oil-immersed transformer, inputting the operating data into the trained hot spot temperature mapping model for temperature prediction, and obtaining the winding hot spot temperature of the oil-immersed transformer includes: The historical winding hot spot temperature and real-time operating data of the oil-immersed transformer are obtained. The historical winding hot spot temperature is input into the input layer and then subjected to gated loop through the reset and update layer to obtain the first output matrix. The first output matrix and the real-time running data are concatenated to obtain the first input matrix; The first input matrix is ​​input into the combined layer, and linear calculations are performed sequentially through the first hidden layer and the second hidden layer. The winding hot spot temperature of the oil-immersed transformer is output through the output layer.

6. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, 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 4.

7. 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 4.

Images