Transformer temperature fluid field computational grid adaptive subdivision method

Abstract

The invention provides a transformer temperature fluid field computational grid adaptive subdivision method, which comprises the following steps of: establishing a transformer three-dimensional entity model, carrying out initial grid subdivision, and calculating the distribution of numerical discrete errors in grid units according to an error calculation formula, the method comprises the following steps of: selecting a grid unit error size, rearranging unit numbers according to the grid unit error size, selecting to-be-encrypted units in a fluid domain according to a proportion, performing local encryption processing on the selected units, performing node distribution optimization, forming a new grid, replacing an old grid with the new grid, resolving each node field quantity, and the like. According to the method, error calculation, unit marking and grid refinement processes are repeatedly executed, and a self-adaption process is ended when a calculation result of an observation point meets requirements, so that self-adaption encryption of the grid is realized, and the transformer grid subdivision number which can ensure the solving precision and does not greatly reduce the solving efficiency is determined.

Description

Technical Field

[0001] The invention relates to the technical field of transformer digitization, in particular to a transformer temperature fluid field calculation grid adaptive partitioning method. Background Art

[0002] As the virtual layer in the transformer digital twin technology architecture, the transformer multi-physics field numerical calculation is one of the key technologies to realize the digital transformation of the transformer. In order to accurately obtain the temperature field distribution inside the transformer and provide guidance for the structural optimization of the transformer, numerical calculation methods are widely used in the transformer temperature fluid field calculation. The transformer temperature fluid field calculation is a multi-degree-of-freedom, strongly nonlinear numerical coupling solution problem. The number of grids has an important influence on the solution accuracy. Too few grids cannot guarantee the solution accuracy, while too many grids will lead to too large a solution amount and reduce the solution efficiency. Therefore, an effective transformer overall mesh division control strategy in the transformer multi-physics field numerical calculation is of great significance to ensure the calculation efficiency and accuracy.

[0003] Chinese invention patent CN117574791A discloses a transformer temperature fluid field grid optimization method and system, including: obtaining the transformer structure and establishing a three-dimensional model, respectively performing coarse sectioning and fine sectioning on the model, and obtaining an initial grid of the coarse section model and a fine grid of the fine section model. Performing temperature-fluid field calculation on the initial grid and the fine grid respectively, comparing the calculation results to obtain a relative error of the result, and when the relative error of the result is greater than a set threshold, locally encrypting the initial grid of the coarse section model. The locally encrypted calculation result is cycled through the temperature-fluid field calculation, and the local encryption is cycled until the relative error of the result meets the requirement, and the grid scheme is determined.

[0004] However, when selecting the area to be encrypted, the above-mentioned grid optimization method compares the calculation results of the coarse-section model and the fine-section model, draws a two-dimensional temperature field result difference distribution cloud map, and selects the area where the calculated difference is greater than the set threshold for encryption based on the obtained temperature field result difference distribution cloud map. This method can select areas with large differences based on the two-dimensional difference cloud map, but in the complex three-dimensional fluid field of the transformer, the two-dimensional difference cloud map cannot fully reflect it. At the same time, this encryption decision is relatively rough and can only be encrypted based on macroscopic differences, which may lead to excessive encryption in some local areas and insufficient encryption in other subtle areas with large local errors, which ultimately affects the calculation accuracy and efficiency.

[0005] Therefore, the applicant proposed a formula for calculating the unit error in the transformer temperature fluid field calculation process based on an adaptive mesh subdivision control method based on error analysis, and then specifically extracted the units with large numerical calculation errors in the transformer temperature fluid field calculation process, thereby realizing adaptive encryption of the mesh and determining the number of transformer mesh subdivisions that can ensure the solution accuracy without significantly reducing the solution efficiency. Summary of the invention

[0006] The present invention proposes a global navigation satellite system measurement device, which solves the problems in the prior art. The technical solution of the present invention is implemented as follows: a transformer temperature fluid field calculation grid adaptive partitioning method is provided, comprising the following steps:

[0007] Step S1: Establish a three-dimensional solid model of the transformer and perform initial meshing, and calculate the temperature fluid field distribution of the model under the initial mesh;

[0008] Step S2: Calculate the distribution of numerical discrete errors in grid cells according to the error calculation formula, and rearrange the cell numbers according to the size of the grid cell errors;

[0009] Step S3: Select the cells to be encrypted in the fluid domain in proportion, perform local encryption processing on the selected cells and optimize the node distribution to form a new grid;

[0010] Step S4: Resolve the field quantities of each node by replacing the old grid with the new grid; repeat the error calculation, unit marking and grid refinement process in steps S2 to S3, and end the adaptive process when the calculation results of the observation points meet the requirements.

[0011] As a preferred technical solution, in step S1, initial meshing is performed by combining mapping meshing with free meshing, and the temperature fluid field distribution of the model under the initial mesh is calculated based on the finite volume method.

[0012] As a preferred technical solution, in step S1, a three-dimensional transformer model coupled with multiple physical fields is established, wherein the three-dimensional transformer model includes an iron core, windings, structural parts, oil channels and a radiator. In the initial mesh division, scanning mapping division is adopted for the regular structures including transformer windings, insulation and oil channels, free division is adopted for the irregular structures including transformer casing, transformer oil and internal structural parts of the transformer, and hierarchical encryption division is performed on the fluid boundary layer part where the transformer oil contacts the casing wall.

[0013] As a preferred technical solution, in step S2, the error estimation formula is determined to be Eindicator, where:

[0014]

[0015] The second-order difference formula shown in formula (2) is used to estimate it, thereby further obtaining the error estimation formula Eindicator shown in formula (3);

[0016]

[0017] In the formula, △ represents the first-order difference operation, and △2 represents the second-order difference operation;

[0018]

[0019] Under three-dimensional conditions, the error index Eindicator is as shown in the formula, where Ve is the volume of the adjacent unit;

[0020] The fluid flow rate in the x, y and z directions is used as the error control quantity, and the grid is adaptively encrypted three times. After the three adaptive encryptions, the required final temperature fluid field calculation grid is output. In the first adaptive process, the fluid flow rate u in the x direction is used as the focus quantity, and the error calculation formula Eindicator(u) with the fluid flow rate u as the variable is calculated, and then the unit order is rearranged according to the error size.

[0021] As a preferred technical solution, in step S3, each time encryption is performed, the cells to be encrypted in the fluid domain are selected at a ratio of 5% of the current number of grids, and when selecting the encrypted grids of the fluid domain, the grid cells in the boundary layer of the fluid domain are avoided; the local encryption processing of the selected cells includes the following steps: 1) reducing the subdivision size of the selected cells and the cells near the selected cells; 2) re-subdividing to achieve the purpose of encryption; 3) optimizing the node distribution to form a new grid.

[0022] As a preferred technical solution, the first grid adaptive refinement specifically includes the following steps:

[0023] Step a: Taking the fluid flow velocity u in the x direction as the variable of interest, calculate the error calculation formula Eindicator(u) of the fluid with the fluid flow velocity u as the variable;

[0024] Step b: Then rearrange the order of the units according to the error size;

[0025] Step c: Select the cells to be encrypted in the fluid domain at a ratio of 5% of the current number of grids, and perform grid encryption and node distribution optimization.

[0026] Step d: Solve the new grid based on the finite volume method and calculate the node error Eindicator(u) of the new grid. If the error is less than the given value, the adaptive encryption ends. Otherwise, continue to select the unit with larger error for grid encryption until the error meets the requirement, and finally form the adaptive mesh.

[0027] As a preferred technical solution, step S4 specifically includes the following steps: using the newly formed adaptive mesh to replace the old mesh to re-solve the field quantities of each node, and taking the fluid flow velocity u in the y direction and the fluid flow velocity w in the z direction as the quantities of interest, calculating the error calculation formulas Eindicator(v) and Eindicator(w), and repeating Figure 8 The error calculation, unit marking and mesh refinement process in the process finally obtains the third adaptive encrypted mesh and intercepts the overall mesh section in the middle of the transformer.

[0028] Compared with the existing technology, this solution has the following beneficial effects: by proposing a unit error calculation formula in the transformer temperature fluid field calculation process, the units with large numerical calculation errors in the transformer temperature fluid field calculation process can be specifically extracted, thereby realizing adaptive mesh encryption. Compared with the above-mentioned encryption method based on the macro difference distribution map, the error calculation formula provides a more accurate error assessment, which can more effectively improve the accuracy of the calculation results, and can fully cover the error distribution in three-dimensional space. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0030] Figure 1 A method flow chart of a method for adaptively dividing a transformer temperature fluid field calculation grid according to the present invention;

[0031] Figure 2 It is a schematic diagram of the overall grid of the internal structure of the transformer of the present invention;

[0032] Figure 3 A schematic diagram of the transformer winding mapping and mesh division of the present invention;

[0033] Figure 4 A schematic diagram of hierarchical encrypted meshing of a transformer fluid domain boundary layer of a transformer temperature fluid field calculation mesh adaptive meshing method of the present invention;

[0034] Figure 5 It is a flow chart for solving the temperature fluid field control equation of the present invention;

[0035] Figure 6 A schematic diagram of a one-dimensional finite volume method grid discretization method of the present invention;

[0036] Figure 7This is a schematic diagram of the encryption of the selected unit of the present invention;

[0037] Figure 8 A schematic diagram of a selected unit in the first grid self-adaptation process of the present invention;

[0038] Fig. 9 This is a schematic diagram of encrypted units in the first grid self-adaptation process of the present invention;

[0039] Fig.10 This is a specific flow chart of the first adaptive encryption of the present invention;

[0040] Fig.11 is the initial grid distribution diagram of the present invention;

[0041] Fig.12 This is a schematic diagram of the first adaptive encryption of the present invention;

[0042] Fig.13 This is a schematic diagram of the second adaptive encryption of the present invention;

[0043] Fig.14 This is a schematic diagram of the third adaptive encryption of the present invention;

[0044] Fig.15 This is the temperature field distribution diagram of the transformer housing of the present invention;

[0045] Fig.16 This is the temperature distribution diagram of the low voltage winding of the present invention;

[0046] Fig.17 This is the temperature distribution diagram of the high voltage winding of the present invention. DETAILED DESCRIPTION

[0047] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0048] Reference Figure 1 , a transformer temperature fluid field calculation grid adaptive partitioning method is invented, which mainly includes the following steps:

[0049] S1 establishes a three-dimensional solid model of the transformer, uses a combination of mapping and free meshing to perform initial meshing, and calculates the temperature and fluid field distribution of the model under the initial mesh based on the finite volume method;

[0050] S2 calculates the distribution of numerical discrete errors in the unit according to the error calculation formula, and rearranges the unit numbers according to the size of the unit error;

[0051] S3 selects cells to be encrypted in the fluid domain at a ratio of 5% of the current number of grids, performs local encryption processing on the selected cells and optimizes node distribution to form a new grid;

[0052] S4 replaces the old grid with the new grid to re-solve the field quantities of each node, repeats the error calculation, unit marking and grid refinement process in S2~S3, and ends the adaptive process when the calculation results of the observation points meet the requirements.

[0053] The following describes the invention in conjunction with specific embodiments.

[0054] Step 1:

[0055] See also Figure 2 to Figure 4 , a three-dimensional transformer model with multi-physical field coupling is established, and the three-dimensional transformer model includes an iron core, windings, structural parts, oil channels and radiators. In the initial meshing of this patent, scanning mapping is used for regular structures such as transformer windings, insulation, and oil channels, and free meshing is used for irregular structures such as transformer casing, transformer oil, and internal structural parts of the transformer. In order to obtain an initial mesh with good mesh quality, hierarchical encryption is performed in the fluid boundary layer where the transformer oil contacts the casing wall.

[0056] The solution to the convective heat transfer process of transformer oil follows three basic laws of physics, including the law of conservation of mass, the law of conservation of momentum, and the law of conservation of energy. It is described by equations (1) to (3):

[0057]

[0058] In the formula, f is the external force per unit volume of fluid; ρ is the fluid density; v is the fluid velocity vector; p is the fluid pressure; η is the fluid dynamic viscosity; e is the fluid internal energy; T is the fluid temperature; cp is the fluid specific heat capacity; k is the fluid thermal conductivity; Φ is the fluid heat source; Sh is the part of the fluid mechanical energy converted into thermal energy under the action of viscous force.

[0059] The finite volume method discretizes the field to be solved into a finite number of non-overlapping control volumes, and integrates and discretizes the differential control equation for each control volume, and then derives a set of linear equations based on the field node variables and calculates and solves them. This patent uses the finite volume method principle to solve the above control equations and obtain the temperature field and fluid field distribution in the transformer.

[0060] The finite volume method is used to calculate the temperature fluid field of the transformer. When the transformer is under rated load, the ambient temperature is 25°C and there is no wind speed outside, the loss loading values ​​of various parts of the transformer are shown in Table 1, and the loading of the heat transfer coefficient due to the external boundary radiation is shown in Table 2.

[0061] Table 1 Loss loading value of each part of transformer under rated load (W)

[0062]

[0063] Table 2 Transformer outer boundary radiation heat transfer coefficient (W·m-2·K-1)

[0064]

[0065] The overall solution process is as shown in 5.

[0066] Step 2:

[0067] Numerical calculation is the process of approximating the actual field distribution through discrete interpolation. Therefore, the error can be considered as the difference between the interpolation function f in the region and the actual distribution g function. Figure 6 The grid discretization method of the finite volume method under one-dimensional conditions is given. In the figure, w1 and w2 are called control volume boundaries, the area enclosed by the boundaries is called the control volume, P1, P2, and P3 are called central nodes of the control volume, and the field values ​​(temperature, velocity, and pressure) of the boundaries w1~w2 are obtained by interpolation of the central nodes P1~P3.

[0068] In the interval P1-P2, based on the value of point P1, the actual distribution function g(x) of the field variable value is expanded by the second order Taylor, and we can get:

[0069]

[0070] Where g'(P1) and g”(P1) are the first and second derivatives of the function g(x) at point P1, respectively.

[0071] In numerical calculation, when the linear interpolation function f(x) is used to interpolate the field values ​​of each point in the control volume, the error Er between the interpolation function f(x) and the true solution g(x) can be expressed as:

[0072]

[0073] Obviously, the above error is caused by the nonlinearity of the true distribution g(x). Therefore, the first and second terms in equation (5) are error transmission terms, and the third term is the internal cause of the error, reflecting the degree to which the true distribution g(x) deviates from linearity in the interpolation interval P1-P2. Since the error is transitive, the deviation of the degree of freedom calculation result is not the source of the error. Reducing the error of the computational grid should be aimed at reducing the error. Therefore, only the error generation term in equation (5) is retained to obtain the error index. The larger the error index, the larger the error generated at the node. The error index Eindicator of the corresponding control volume is obtained by integrating the error term in the control volume, that is:

[0074]

[0075] The second-order derivative term g”(P1) of the actual distribution function g(x) of the field is approximated by the second-order derivative term f”(P1) of the interpolation function f(x). Thus, the error estimation formula Eindicator shown in formula (7) is obtained.

[0076]

[0077] In addition, in order to obtain the second-order derivative of the true solution, the second-order difference formula shown in formula (8) is used to estimate it, thereby further obtaining the error estimation formula Eindicator shown in formula (9).

[0078]

[0079] In the formula, △ represents the first-order difference operation, and △2 represents the second-order difference operation.

[0080]

[0081] Under three-dimensional conditions, the error indicator Eindicator is shown in formula (10), where Ve is the adjacent unit volume.

[0082]

[0083] This scheme uses the fluid flow rate in the x, y, and z directions as the error control quantity, performs three grid adaptive encryption, and outputs the required final temperature fluid field calculation grid after the three adaptive encryptions. The first adaptive process uses the fluid flow rate u in the x direction as the concern quantity, and calculates the error calculation formula Eindicator(u) with the fluid flow rate u as the variable. Then rearrange the unit order according to the error size.

[0084] Step 3:

[0085] To avoid a substantial increase in the number of meshes during the encryption process, each time the meshes are encrypted, the cells to be encrypted in the fluid domain are selected at a ratio of 5% of the current number of meshes. At the same time, to avoid over-encryption of the boundary layer meshes, when selecting the encrypted meshes of the fluid domain, the mesh cells in the boundary layer of the fluid domain are avoided. The selected cells are locally encrypted. The encryption diagram is shown in the figure. Figure 7 As shown in the figure, the mesh size of the selected unit and the units near the selected unit is reduced, and the mesh is re-meshed to achieve the purpose of encryption. At the same time, the node distribution is optimized to form a new mesh.

[0086] The first mesh adaptive encryption takes the fluid flow velocity u in the x direction as the variable, and calculates the error calculation formula Eindicator(u) of the fluid with the fluid flow velocity u as the variable. Then the order of the units is rearranged according to the error size. The units to be encrypted in the fluid domain are selected at a ratio of 5% of the current number of grids, and the mesh encryption and node distribution are optimized. The new mesh is solved based on the finite volume method, and the node error Eindicator(u) of the new mesh is calculated. If the error E is less than the given value r, the adaptive encryption ends. Otherwise, continue to select units with larger errors for mesh encryption until the error meets the requirements, and finally form the adaptive mesh for this time. The specific implementation process is as follows Fig.10 shown.

[0087] Step 4:

[0088] The newly formed adaptive mesh is used to replace the old mesh to re-solve the field quantities of each node, and the fluid velocity u in the y direction and the fluid velocity w in the z direction are taken as the quantities of interest, and the error calculation formulas Eindicator(v) and Eindicator(w) are calculated. Repeat Fig.10 The error calculation, unit marking and mesh refinement process in the third adaptive mesh refinement are finally obtained. The overall mesh section of the middle part of the transformer is intercepted. The mesh changes of the observed section in the third mesh adaptive refinement are as follows: Figure 11 to Figure 3 shown.

[0089] The top oil temperature of the transformer and the hot spot temperature of the winding are two state quantities that the transformer focuses on observing during actual operation. Therefore, when studying the transformer grid adaptive encryption algorithm, this patent selects the top oil temperature of the transformer and the hot spot temperature of the winding as observation points, and calculates the temperature changes of the observation points before and after grid refinement. The adaptive process ends when the calculation results of the observation points meet the requirements.

[0090] The calculation results of the top oil temperature and the hot spot temperature of the winding during the grid adaptation process are shown in Table 3. Taking the calculation results under the condition of 5.118 million grid divisions as the benchmark value, when the number of grid divisions is 1.785 million, the difference between the hot spot temperature and the calculated temperature values ​​of each temperature monitoring point and the benchmark value exceeds 1°C, and the error is large. After three adaptive grid encryption divisions, the hot spot temperature and the calculated temperature values ​​of each temperature monitoring point are no more than 0.5°C from the benchmark value.

[0091] The overall grid after the third adaptive encryption is 65.6% of the number of grids under global encryption. A computer with a main frequency of 2.20GHz and 16 threads is used for parallel calculation. The single steady-state calculation time under the global encrypted grid is 253min, while the single steady-state calculation time using the grid after the third adaptive encryption is 162min, and the calculation time is reduced by 36.0%.

[0092] Table 3 Temperature changes of observation points during adaptive encryption

[0093]

[0094] The temperature fluid field is calculated using a grid that has been adaptively refined three times. The transformer is under rated load, the ambient temperature is 25°C and there is no wind. The transformer shell temperature distribution is calculated as follows: Fig.10 , Transformer shell temperature field distribution Fig.15 As shown, the high and low voltage winding temperatures are Figure 16-17 shown.

[0095] Compared with the existing technology, this solution has the following innovations:

[0096] (1) It is revealed that in the numerical calculation of variable pressure and temperature fluid field, the error between the field function and the linear interpolation function is mainly composed of three terms, the first two are error transfer terms, and the third is the internal cause of the error. Based on the internal cause of the error, a unit error calculation formula Eindicator is proposed.

[0097] (2) A method for adaptive mesh generation for transformer temperature fluid field calculation is proposed. The fluid flow velocity in the x, y, and z directions is used as the error control variable, and the mesh is adaptively encrypted three times. The overall mesh after the third adaptive encryption is 65.6% of the number of meshes under global encryption, and the calculation time is reduced by 36.0%.

[0098] By proposing a calculation formula for the unit error in the transformer temperature fluid field calculation process, the units with large numerical calculation errors in the transformer temperature fluid field calculation process can be extracted in a targeted manner, thereby realizing adaptive mesh encryption. Compared with the above-mentioned encryption method based on the macro difference distribution map, the error calculation formula provides a more accurate error assessment, which can more effectively improve the accuracy of the calculation results, and can fully cover the error distribution in three-dimensional space.

[0099] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for adaptive grid partitioning of transformer temperature fluid field calculation, characterized in that: The following steps are involved: Step S1: Establish a three-dimensional solid model of the transformer and perform initial meshing, and calculate the temperature fluid field distribution of the model under the initial mesh; Step S2: Calculate the distribution of numerical discrete errors in grid cells according to the error calculation formula, and rearrange the cell numbers according to the size of the grid cell errors; Step S3: Select the cells to be encrypted in the fluid domain in proportion, perform local encryption processing on the selected cells and optimize the node distribution to form a new grid; Step S4: Resolve the field quantities of each node by replacing the old grid with the new grid; repeat the error calculation, unit marking and grid refinement process in steps S2 to S3, and end the adaptive process when the calculation results of the observation points meet the requirements.

2. A transformer temperature fluid field calculation grid adaptive partitioning method according to claim 1, characterized in that: In the step S1, the initial grid division is performed by combining mapping division and free division, and the temperature fluid field distribution of the model under the initial grid is calculated based on the finite volume method.

3. A transformer temperature fluid field calculation grid adaptive partitioning method as claimed in claim 2, characterized in that: In the step S1, a three-dimensional transformer model coupled with multiple physical fields is established, wherein the three-dimensional transformer model includes an iron core, windings, structural parts, oil channels and a radiator. In the initial meshing, scanning mapping is used for the regular structures including the transformer windings, insulation and oil channels, free meshing is used for the irregular structures including the transformer casing, transformer oil and internal structural parts of the transformer, and hierarchical encryption is performed on the fluid boundary layer portion where the transformer oil contacts the casing wall.

4. The method for adaptive grid generation for transformer temperature fluid field calculation according to claim 1, characterized in that: In step S2, the error estimation formula is determined as E indicator ,in, The second-order difference formula shown in formula (2) is used to estimate it, and then the error estimation formula E shown in formula (3) is obtained. indicator ; In the formula, △ represents the first-order difference operation, and △2 represents the second-order difference operation; Under three-dimensional conditions, the error index E indicator As shown in the formula, Ve is the adjacent unit volume; The fluid flow rate in the x, y and z directions is used as the error control quantity, and the grid is adaptively encrypted three times. After the three adaptive encryptions, the required final temperature fluid field calculation grid is output. In the first adaptive process, the fluid flow rate u in the x direction is used as the focus quantity, and the error calculation formula Eindicator(u) with the fluid flow rate u as the variable is calculated, and then the unit order is rearranged according to the error size.

5. The method for adaptive grid generation for transformer temperature fluid field calculation according to claim 1, characterized in that: In step S3, each time encryption is performed, cells to be encrypted in the fluid domain are selected at a ratio of 5% of the current number of grids, and when selecting the encrypted grids of the fluid domain, grid cells in the boundary layer of the fluid domain are avoided; The local encryption process for the selected unit includes the following steps: 1) reducing the segmentation size of the selected unit and the units near the selected unit; 2) re-dividing to achieve the purpose of encryption; 3) optimizing the node distribution to form a new grid.

6. A transformer temperature fluid field calculation grid adaptive partitioning method as claimed in claim 5, characterized in that: The first mesh adaptive refinement specifically includes the following steps: Step a: Taking the fluid flow velocity u in the x direction as the variable of interest, calculate the error calculation formula Eindicator(u) of the fluid with the fluid flow velocity u as the variable; Step b: Then rearrange the order of the units according to the error size; Step c: Select the cells to be encrypted in the fluid domain at a ratio of 5% of the current number of grids, and perform grid encryption and node distribution optimization. Step d: Solve the new grid based on the finite volume method and calculate the node error Eindicator(u) of the new grid. If the error is less than the given value, the adaptive encryption ends. Otherwise, continue to select the unit with larger error for grid encryption until the error meets the requirement, and finally form the adaptive mesh.

7. The method for adaptive grid generation for transformer temperature fluid field calculation according to claim 1, characterized in that: Step S4 specifically includes the following steps: replacing the old grid with the newly formed adaptive mesh, resolving the field quantities of each node, and taking the fluid flow velocity u in the y direction and the fluid flow velocity w in the z direction as the quantities of interest, calculating the error calculation formulas Eindicator(v) and Eindicator(w), repeating the error calculation, unit marking and mesh refinement process in Figure 8, and finally obtaining the third adaptive encrypted grid, intercepting the overall mesh section in the middle of the transformer.

Images