A converter transformer temperature field cloud map reconstruction method and system

Abstract

The application discloses a converter transformer temperature field cloud picture reconstruction method and system, and relates to the technical field of converter transformer design. The method comprises the following steps: obtaining the load rate, the harmonic distortion rate, the basic temperature field, the measuring point coordinates and the measuring point temperature sequence of the converter transformer by using a basic fast finite element temperature field model simulation; establishing the spatial mapping between the measured measuring points and the simulation grid nodes according to the measuring point coordinates and the measuring point temperature sequence; determining the steady-state temperature deviation according to the deviation between the measured temperature and the basic simulation temperature, and constructing a spatial error matrix by radial basis function interpolation based on the steady-state temperature deviation; determining the working condition correction coefficient according to the load rate and the harmonic distortion rate; and reconstructing the temperature field cloud picture based on the basic temperature field, the spatial error matrix and the working condition correction coefficient to obtain a corrected temperature field cloud picture. The application can obtain a temperature field cloud picture closer to the actual operating state of the converter transformer.

Description

Technical Field

[0001] This invention relates to the field of converter transformer design technology, and in particular to a method and system for reconstructing temperature field cloud maps of converter transformers. Background Technology

[0002] Temperature field contour maps are fundamental data for converter transformer design verification, condition assessment, and insulation life analysis. Conventional fast finite element models typically use the fundamental load as the main heat source input, which can quickly provide the overall temperature rise distribution. However, for converter transformers subjected to high-order harmonic currents, local leakage flux losses, and heat concentration phenomena in the end structure under DC transmission scenarios, basic fast finite element models often cannot accurately describe all these phenomena within minutes of computation. Simply relying on the temperature measured at the measurement point can only reflect the local location and cannot obtain a global contour map of the windings, oil channels, and insulation structure. Furthermore, although basic fast simulations have high computational speed, they tend to underestimate local hot spots caused by harmonics. While directly using a full-frequency electromagnetic-fluid-thermal strongly coupled transient model offers high accuracy, the model is large in scale, has complex boundary couplings, and a long convergence process, making it unsuitable for frequent operating condition assessments and rapid on-site maintenance verification. Summary of the Invention

[0003] To address the shortcomings of existing technologies, the present invention aims to provide a method and system for reconstructing temperature field cloud maps of converter transformers. By using a basic temperature field cloud map to provide a spatial skeleton, measured deviations to provide error calibration, interpolation matrices to provide global compensation, and operating condition coefficients to provide operational modulation, a balance is achieved between computational efficiency and local accuracy, thereby obtaining a temperature field cloud map that more closely approximates the actual operating state of a converter transformer.

[0004] To achieve the above objectives, the present invention is implemented through the following technical solution: The first aspect of this invention provides a method for reconstructing temperature field cloud maps of a converter transformer, comprising the following steps: The load rate, harmonic distortion rate, base temperature field, measuring point coordinates, and measuring point temperature sequence of the converter transformer were obtained by using a basic fast finite element temperature field model for simulation. Establish a spatial mapping between measured points and simulation grid nodes based on the measured point coordinates and temperature sequence; The steady-state temperature deviation is determined based on the difference between the measured temperature and the basic simulation temperature, and a spatial error matrix is ​​constructed based on the steady-state temperature deviation through radial basis function interpolation. Determine the operating condition correction factor based on the load rate and harmonic distortion rate; The temperature field cloud map is reconstructed based on the basic temperature field, spatial error matrix, and operating condition correction coefficient to obtain the corrected temperature field cloud map.

[0005] Furthermore, the specific steps for obtaining the converter transformer load rate, harmonic distortion rate, base temperature field, measuring point coordinates, and measuring point temperature sequence using the basic fast finite element temperature field model simulation are as follows: Obtain the geometric structure, winding region, oil passage location, material thermal properties, oil flow boundary, temperature boundary, rated current, operating current, and harmonic current of the converter transformer. An experimental platform for the combined electrothermal effect based on a basic rapid finite element temperature field model was built, and basic simulations were performed.

[0006] Furthermore, the basic rapid finite element temperature field model retains the spatial relationship of the winding ends, conductor transposition area, oil channel turning area, insulation structure near the pressure plate, and leakage magnetic concentration area. For some chamfers, bolt holes, or assembly gaps, equivalent treatment is performed without changing the main heat path.

[0007] Furthermore, the load factor is determined by the ratio of the effective value of the actual operating current to the effective value of the rated current, and the harmonic distortion rate is determined by the effective values ​​of each harmonic current and the effective value of the fundamental current.

[0008] Furthermore, the electrothermal combined effect test platform includes at least a programmable AC constant current power supply, a harmonic injection unit, an oil flow circulation temperature control chamber, a multi-channel fiber optic temperature demodulator, a sample clamping and insulating oil pretreatment device, and a simulation workstation. The programmable AC constant current power supply and harmonic injection unit are used to apply different load rates and harmonic distortion rates; the oil flow circulation temperature control chamber is used to maintain the inlet oil temperature and temperature boundary; the multi-channel fiber optic temperature demodulator is used to collect the temperature of the pre-embedded measuring points; and the simulation workstation is used to establish a basic fast finite element temperature field model that is geometrically consistent with the test sample.

[0009] Furthermore, the specific steps for establishing the spatial mapping between measured points and simulation mesh nodes based on the measured point coordinates and temperature sequences are as follows: The nearest neighbor rule is used to determine the simulation grid node that is closest to the measured point, and the basic simulation temperature of the simulation grid node is extracted from the basic temperature field cloud map.

[0010] Furthermore, after ensuring that the rate of change of deviation does not exceed the steady-state determination threshold within multiple consecutive sampling intervals, the steady-state temperature deviation is obtained by averaging the dynamic temperature deviations at multiple sampling times at the end.

[0011] Furthermore, by restoring the corresponding reference steady-state temperature deviation at each reference measurement point using the spatial error matrix, the interpolation weights of each measurement point are solved.

[0012] Furthermore, the radial basis functions are inverse quadratic radial basis functions or verified Gaussian radial basis functions.

[0013] Furthermore, the operating condition correction factors include load correction terms and harmonic correction terms. The load correction term is determined by a nonlinear fitting function of the load rate, and the harmonic correction term is determined by the deviation of the harmonic distortion rate from the reference harmonic distortion rate.

[0014] Furthermore, the measured temperature of the verification measurement point is used to calculate the temperature error, hot spot error, and operating condition extrapolation degree of the measurement point. Based on the preset error threshold, it is determined whether the corrected temperature field cloud map can be output as a high-confidence result. When the temperature error, hot spot error, and operating condition extrapolation degree of the measurement point exceed the preset error threshold, the corresponding operating condition is marked as an operating condition to be verified, and offline test sampling, simulation recalculation, or parameter recalibration is triggered.

[0015] A second aspect of the present invention provides a converter transformer temperature field cloud map reconstruction system, comprising: The basic simulation module is used to simulate and obtain the load rate, harmonic distortion rate, basic temperature field, measurement point coordinates, and measurement point temperature sequence of the converter transformer using the basic fast finite element temperature field model. The measuring point mesh mapping module is used to establish a spatial mapping between measured measuring points and simulated mesh nodes based on the measuring point coordinates and temperature sequence. The spatial error matrix generation module is used to determine the steady-state temperature deviation based on the deviation between the measured temperature and the basic simulation temperature, and to construct the spatial error matrix based on the steady-state temperature deviation through radial basis function interpolation. The operating condition coefficient fitting module is used to determine the operating condition correction coefficient based on the load rate and harmonic distortion rate. The cloud map reconstruction module is used to reconstruct the temperature field cloud map based on the basic temperature field, spatial error matrix, and operating condition correction coefficients to obtain a corrected temperature field cloud map.

[0016] A third aspect of the present invention provides a computer-readable storage medium storing a computer program adapted to be loaded by a processor and executed the steps in the converter transformer temperature field cloud map reconstruction method as described in the first aspect of the present invention.

[0017] A fourth aspect of the present invention provides a computer device comprising: A processor, adapted to execute computer programs; A computer-readable storage medium storing a computer program, which, when executed by the processor, implements the converter transformer temperature field cloud map reconstruction method as described in the first aspect of the present invention.

[0018] A fifth aspect of the present invention provides a computer program product or computer program comprising computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the steps in the converter transformer temperature field contour map reconstruction method as described in the first aspect of the present invention.

[0019] The above one or more technical solutions have the following beneficial effects: This invention discloses a method and system for reconstructing temperature field cloud maps of converter transformers. After the basic temperature field cloud map is output, a spatial error interpolation matrix covering the entire domain is constructed by utilizing the spatial deviation between the measured node temperature and the basic simulated node temperature. Then, the load rate and harmonic distortion rate are introduced to modulate the error amplitude, thereby obtaining a temperature field cloud map that is closer to the actual operating state without re-executing the full-frequency strongly coupled simulation.

[0020] Advantages of additional aspects of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

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

[0022] Figure 1 This is a flowchart of the method for reconstructing the temperature field cloud map of the converter transformer in Embodiment 1 of the present invention; Figure 2 This is a schematic diagram illustrating the fitting relationship between the error field correction coefficient and the load rate in Embodiment 1 of the present invention; Figure 3 This is a schematic diagram of the hardware structure of the temperature field cloud map reconstruction system in Embodiment 2 of the present invention. Detailed Implementation

[0023] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0024] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof. The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0025] Example 1: Embodiment 1 of the present invention provides a method for reconstructing the temperature field cloud map of a converter transformer, such as... Figure 1 As shown, it includes the following steps: S1: The load rate, harmonic distortion rate, base temperature field, measuring point coordinates, and measuring point temperature sequence of the converter transformer are obtained by simulating the basic fast finite element temperature field model.

[0026] S1.1: Obtain the geometric structure, winding region, oil passage location, material thermophysical parameters, oil flow boundary, temperature boundary, rated current, operating current, and harmonic current of the converter transformer.

[0027] In one specific implementation, the geometry, winding region, oil passage location, material thermal properties, oil flow boundary, temperature boundary, rated current, operating current, and harmonic currents of the converter transformer are obtained. In this embodiment, the load factor is determined by the ratio of the effective value of the actual operating current to the effective value of the rated current, and the harmonic distortion rate is determined by the effective values ​​of the harmonic currents and the fundamental current.

[0028] To standardize the mathematical representation of operating conditions, the load factor β is defined as follows: (1).

[0029] In the formula, β represents the load factor; This represents the effective value of the actual operating current. This indicates the effective value of the rated current. and Both are current quantities, and their ratio is used to eliminate the difference in rated values ​​between devices with different capacities.

[0030] The harmonic distortion rate H is further defined as follows: (2).

[0031] In the formula, H represents the harmonic distortion rate; Indicates the effective value of the fundamental current; This represents the effective value of the m-th harmonic current; m represents the harmonic number; M represents the highest harmonic number included in the analysis.

[0032] Organize the test conditions into a set of conditions G: (3).

[0033] In the formula, G represents the set of test conditions; This represents the p-th load condition; This represents the q-th harmonic distortion rate operating condition; P represents the number of load rate operating conditions; and Q represents the number of harmonic distortion rate operating conditions.

[0034] S1.2: Build an experimental platform for combined electrothermal and thermal effects based on a basic rapid finite element temperature field model and conduct basic simulations.

[0035] In one specific implementation, the basic fast finite element temperature field model retains the spatial relationships of the winding ends, conductor transposition areas, oil channel turning areas, insulation structures near the pressure plate, and areas of concentrated leakage flux. For chamfers, bolt holes, or assembly gaps that have a minor impact on the temperature field but significantly increase the mesh count, equivalent processing is performed without altering the main thermal path. The location and rationale for the equivalent processing should be recorded in the model version log. The electrothermal combined effect test platform includes at least a programmable AC constant current power supply, a harmonic injection unit, an oil flow circulation temperature control chamber, a multi-channel fiber optic temperature demodulator, and a simulation workstation. The programmable AC constant current power supply and harmonic injection unit are used to apply different load rates and harmonic distortion rates, respectively; the oil flow circulation temperature control chamber is used to maintain the inlet oil temperature and temperature boundaries; the multi-channel fiber optic temperature demodulator is used to collect the temperature of pre-embedded measuring points; and the simulation workstation is used to establish a basic fast finite element temperature field model that is geometrically consistent with the test sample.

[0036] The thermal properties of materials should be differentiated for copper conductors, insulating paperboard, pressure plates, iron cores, transformer oil, and structural components. If the thermal conductivity or specific heat capacity of a material changes significantly with temperature, segmented parameters or table-lookup parameters can be used; if measured parameters are lacking, equipment design data or verified standard parameters should be used, and compensation should be made in the subsequent error matrix through measured deviations.

[0037] The temperature boundary should at least describe the oil inlet temperature, oil outlet constraints, natural convection or forced oil circulation conditions, cooler on / off status, and ambient temperature. For online reconfiguration scenarios, the cooler status can be written as a condition label into the data packet to prevent different cooling conditions under the same β and H from being mistakenly merged.

[0038] The purpose of the basic fast finite element model is to provide a stable spatial temperature framework, not to completely replace the high-precision fully coupled model. Therefore, in this embodiment... The generation should emphasize the reliability of grid node coordinates, region affiliation, and hotspot candidate locations; as long as Maintaining a mapping relationship with the measured deviation, subsequent... This allows for compensation for local deviations.

[0039] The temperature measurement channels in the electrothermal combined action test platform should cover as much as possible the winding ends, oil inlet, oil outlet, near the core, areas where the insulation thickness changes, and existing hot spots. If the number of measurement points is limited by insulation safety or wiring space, they should be prioritized in locations with large temperature gradients, sensitive leakage flux loss, or high historical fault risk.

[0040] The sampling process should record a uniform timestamp for each temperature channel, each current channel, and each cooling status signal. If the current and temperature sampling frequencies are different, the original sampling sequence should be retained, and the interpolation method should be recorded after resampling. This way, in the event of a failure to determine steady state, it can be traced back to whether the actual thermal process is unstable or whether the fluctuation is caused by a data synchronization error.

[0041] In this embodiment, the output of the current step should not only save a single contour map, but also the boundary conditions, model mesh, material thermal property parameters, and operating condition labels required to calculate the contour map. When calibrating the same device again subsequently, the validated basic fast finite element model version can be directly called, avoiding result drift caused by repeated modeling.

[0042] At the data acquisition level, measured temperature Should be related to operating current Fundamental current Harmonic currents Oil temperature, ambient temperature, and cooler status are recorded synchronously. If different data sources have different sampling frequencies, they can be resampled at a uniform time step, with the current sampling time as the main time axis. If there are short-term data loss in the temperature measurement channel, adjacent stable points can be used for interpolation, but the interpolation interval should not be included in the selection of steady-state deviation samples.

[0043] At the basic simulation level The mesh does not need to be completely consistent with the high-precision fully coupled model, but it should cover the main heat conduction paths such as windings, core, oil channels, insulation components, and temperature boundaries. For mesh-refined regions, the node numbers and cell affiliations should be retained during subsequent measurement point-mesh mapping to avoid erroneously averaging local temperature gradients near hot spots onto the coarse mesh region.

[0044] To illustrate the physical difference between the real temperature field and the fundamental temperature field, consider the real temperature field with harmonic-added heat sources. This can be represented as the following control relationship: (4).

[0045] In the formula, ρ represents the material density; c represents the material specific heat capacity; Represents the actual temperature field; t represents time; λ represents the thermal conductivity of the material; ∇ represents the gradient operator; This indicates the heat source term caused by the fundamental current; Indicates the harmonic additional heat source term; This indicates a convective heat dissipation term.

[0046] Correspondingly, when the basic fast finite element model only considers the fundamental wave load, it can be expressed in the following form.

[0047] (5).

[0048] In the formula, This represents the basic temperature field, i.e., the temperature field contour map output by the basic fast finite element model.

[0049] In obtaining Next, a node integrity check should be performed on the basic temperature field contour map. This check includes verifying whether node coordinates are missing, whether isolated grids exist in hotspot candidate regions, whether temperature units are consistent, and whether the contour map matrix dimensions correspond one-to-one with the node numbers. If the basic temperature field contour map itself has breaks or duplicate nodes, then subsequent checks should be performed. Even if the interpolation is correct, local misalignment may still occur during the cloud map reconstruction stage.

[0050] The basic temperature field cloud map should also be mapped to the equipment structure area, for example, dividing the nodes into areas such as windings, core, oil ducts, insulation components, and shell. Area mapping is not a necessary condition for formula calculation, but it helps to set interpolation constraints in subsequent steps and determine whether the hot spots fall in reasonable locations in the final verification stage.

[0051] If the basic fast finite element model already includes some harmonic losses or localized additional losses, its scope should be clearly defined in the model description to avoid [further issues]. The same heat source is repeatedly compensated in this embodiment. The output of the basic fast model is represented uniformly, without being written as the true temperature field, and without being compared with the corrected one. Mixed use.

[0052] S2: Establish a spatial mapping between the measured points and the simulation grid nodes based on the measured point coordinates and the measured point temperature sequence.

[0053] In one specific implementation, this embodiment uses the nearest neighbor rule to determine the simulation grid node that is closest to the measured point, and extracts the basic simulation temperature of the simulation grid node from the basic temperature field cloud map.

[0054] To ensure point-by-point comparison between measured temperatures and baseline simulation temperatures, a spatial mapping between measurement points and mesh nodes needs to be established. Let the position vector of the i-th measured point be: (6).

[0055] In the formula, This represents the position vector of the i-th measured point; These represent the coordinates of the measuring point in three spatial directions; the superscript T indicates vector transpose.

[0056] The nearest neighbor rule is used to determine the simulation grid nodes corresponding to the measured points.

[0057] (7) In the formula, This represents the simulation mesh node number that is closest to the i-th measured point; This represents the position vector of the i-th measured point; This represents the position vector of the j-th simulation mesh node; This represents the index of the node that minimizes the distance; This represents the Euclidean distance.

[0058] After mapping is completed, the basic simulation temperature corresponding to the i-th measurement point can be extracted from the basic temperature field.

[0059] (8).

[0060] In the formula, This represents the basic simulation temperature at the location corresponding to the i-th measuring point; Represents the fundamental temperature field; represents the coordinates of the simulation grid node corresponding to the i-th measurement point; t represents the sampling time; β represents the load rate.

[0061] During the mapping process, the coordinate origin, coordinate axis direction, and length unit must be consistent with the basic rapid finite element model. If the measured points come from 3D scanning, design drawings, or manual records, they should first be converted to the simulation model coordinate system before performing nearest neighbor node matching. For measured points located near oil passages or insulation interfaces, it should also be verified whether their corresponding nodes fall in the correct physical region.

[0062] When the distance between a measured point and the nearest simulation node exceeds the preset spatial tolerance, it is not advisable to directly include it in the interpolation constraints. This measured point can be retained as a point to be verified, or its actual installation location can be manually confirmed and then remapped. The spatial tolerance can be determined comprehensively based on the mesh size, the measured point installation error, and the scale of the hotspot region.

[0063] If multiple measured points are mapped to the same simulation node, it should be determined whether to merge them based on the location of the measured points, channel quality, and thermophysical region. When merging, the average value, confidence-weighted average, or priority should be given to retaining hotspot-sensitive measured points. However, it should be ensured that the deviation sample after merging still corresponds to a unique spatial location.

[0064] If a measuring point is located at the boundary of two thermophysical regions, such as the boundary between the winding end and the oil passage, a local neighborhood weighted mapping can be used instead of simply using the nearest neighbor node. This alternative method still belongs to the spatial mapping step, as long as its final output... And it is used for subsequent deviation calculations without changing the core approach of this invention.

[0065] In practical implementation, measurement points should not only be arranged in areas with gentle temperature changes, but should also cover winding ends, oil passage bends, insulation layers, areas with concentrated leakage flux, and typical cooling paths. If the coordinates of the measurement points do not completely coincide with the simulation nodes, nearest neighbor mapping can be used first, followed by distance-weighted correction for neighboring units. If there is sensor drift or communication anomalies, timestamp alignment, outlier removal, and unit unification should be performed before deviation calculation.

[0066] After the measurement point mesh mapping is completed, measurement point confidence markers should also be established. For measurement points that have long-term drift, hysteresis response, or whose change trends are significantly inconsistent with those of neighboring measurement points, they should not be directly used as interpolation constraint points; instead, they can be used as verification points or low-weight reference points. This can prevent individual abnormal measurement points from forming unreasonable local error peaks in radial basis interpolation.

[0067] S3: Determine the steady-state temperature deviation based on the deviation between the measured temperature and the basic simulation temperature, and construct the spatial error matrix based on the steady-state temperature deviation through radial basis function interpolation.

[0068] S3.1: Determine the steady-state temperature deviation based on the deviation between the measured temperature and the basic simulation temperature.

[0069] In one specific implementation, this embodiment obtains the steady-state temperature deviation by averaging the dynamic temperature deviations at multiple sampling times at the end after the rate of change of deviation within multiple consecutive sampling intervals does not exceed the steady-state determination threshold.

[0070] This step belongs to the node deviation extraction stage. Its purpose is not to directly replace the global cloud map with measured point temperatures, but rather to extract the deviation from the measured temperatures. With basic simulation temperature Extract stable, interpolable, and calibrable deviation samples from the intervals. Assume the operating condition is... The dynamic temperature deviation of the i-th measuring point at time t is as follows: (9).

[0071] In the formula, This indicates that the i-th measuring point is at time t and operating condition Dynamic temperature deviation; This represents the measured temperature at the i-th measuring point; This indicates the basic simulation temperature at the location corresponding to the measuring point.

[0072] The key to this step is not obtaining the temperature difference at a specific moment, but rather obtaining the steady-state deviation that represents the current thermal equilibrium state. Converter transformers have significant thermal inertia; after changes in current or cooling conditions, the temperature at the measuring point may continue to rise or fall for a considerable period of time. If the instantaneous deviation is directly taken, the transient process will be mistakenly written into the spatial error matrix.

[0073] Steady-state determination should consider both the rate of change of deviation and the trend of change of the original temperature. If the rate of change of deviation is small but the original temperature still drifts slowly with the oil temperature, it should be judged whether a quasi-steady state has been reached in conjunction with the temperature boundary. If the rate of change of deviation changes in the same direction at multiple measuring points at the same time, it may be an overall trend caused by changes in operating conditions and should not be used as a stable error sample.

[0074] For missing measurement points caused by short-term communication interruptions, interpolation can be performed on the original temperature sequence, but the interpolation interval should not enter the steady-state averaging window. This ensures that the final part participates in the calculation of equation (12). It comes from real samples, not spurious samples filled in by interpolation.

[0075] If a certain measuring point A persistently negative value indicates that the basic rapid finite element model may overestimate the temperature at that location. This negative deviation should not be mechanically deleted, as it helps the error matrix describe the positive and negative distribution of spatial deviations; it should only be checked at abnormal measurement points when the negative deviation clearly contradicts adjacent measurement points, equipment structure, and temperature boundaries.

[0076] The steady-state deviation sample output in this step should include the measurement point number, measurement point coordinates, corresponding simulation node number, steady-state deviation value, sample weight, and quality label. Subsequent steps will only read the quality-controlled sample set to avoid the interpolation algorithm directly accessing the unfiltered original temperature sequence.

[0077] Specifically, to determine whether the deviation has entered a stable phase, the rate of change of deviation is defined. as follows.

[0078] (10).

[0079] In the formula, This represents the rate of change of the deviation at the i-th measuring point; This represents the k-th sampling time. Indicates the previous sampling time; This represents the dynamic temperature deviation. Equation (10) is designed under the same operating conditions with fixed β and H.

[0080] When all L consecutive sampling intervals meet the following conditions, the deviation of the corresponding measuring point is determined to have reached a steady state.

[0081] (11).

[0082] In the formula, ε represents the steady-state determination threshold; L represents the number of continuous sampling intervals; and K represents the current last sampling time number. Indicates the sampling interval number involved in the steady-state determination; The deviation change rate is expressed in equation (11) in square form, indicating that the deviation change rate does not exceed the threshold.

[0083] After satisfying the steady-state criterion, the average deviation of the last L consecutive sampling times is taken as the steady-state temperature deviation of the measuring point under the current operating conditions.

[0084] (12).

[0085] In the formula, The steady-state temperature deviation at the i-th measuring point is represented by L; the meanings of L, K, and ℓ follow those of equation (11).

[0086] Under the same operating conditions, the average steady-state temperature deviation of all measuring points is defined as: (13).

[0087] In the formula, The horizontal line indicates the average steady-state deviation of all measuring points under the current operating condition; n represents the total number of measuring points; i represents the measuring point number.

[0088] To describe spatial variability, the spatial dispersion of steady-state deviation can be further defined.

[0089] (14).

[0090] In the formula, This indicates the spatial dispersion of the steady-state deviation; This represents the difference between the steady-state deviation of the i-th measuring point and the average steady-state deviation; n represents the total number of measuring points.

[0091] In multi-condition calibration, the same measuring point may show a smaller deviation under low-harmonic conditions, but a significantly larger deviation under high-harmonic conditions. This phenomenon indicates that... Spatial form and The amplitude modulation needs to coexist: Responsible for indicating where the deviation occurs. This indicates the degree to which the deviation should be amplified under the current operating conditions.

[0092] To highlight local hot spot deviations, a maximum steady-state temperature deviation is defined.

[0093] (15).

[0094] In the formula, This represents the maximum steady-state deviation among all measuring points under the current operating conditions; max indicates taking the maximum value.

[0095] Further define the hotspot deviation ratio.

[0096] (16). In the formula, Indicates the hotspot deviation ratio; Indicates the maximum steady-state deviation; The horizontal line above indicates the average steady-state deviation.

[0097] By combining the statistics given in equations (13) to (16), it can be determined whether spatial compensation is needed. If the average steady-state deviation is large and A smaller value indicates an overall temperature shift in the basic fast finite element model, which can be corrected using a smoother error matrix; if... and If the value is large, it indicates that the local hot spot deviation is significant, and spatial interpolation rather than a single proportional coefficient must be used for correction.

[0098] The above deviation analysis can generate error samples that simultaneously include time, operating condition, and spatial dimensions. For 100% load rate and strong harmonic operating conditions, the winding end measuring points... Typically, the temperature rises rapidly over time before stabilizing, with its steady-state value significantly higher than that of the fundamental control group. In the oil passage region, the deviation may manifest as an overall increase in oil temperature and localized flow lag. These phenomena indicate that the additional heat source from harmonics affects both the hot spots of the solid winding and the heat-carrying state of the oil flow, necessitating spatial error compensation rather than single-point correction.

[0099] S3.2: Based on the steady-state temperature deviation, a spatial error matrix is ​​constructed by radial basis function interpolation.

[0100] In one specific implementation, this step belongs to the spatial error matrix construction stage. A set of reference operating conditions is selected. As the reference operating condition for the error field, conditions with significant thermal effects and stable test data are preferentially selected, such as those with high harmonic distortion rates under rated load or short-time overload. Under this reference condition, the steady-state deviations of all measuring points constitute interpolation samples, used to generate a continuous error matrix across the entire temperature field space. Let the spatial coordinate vector of any location to be determined be: (17).

[0101] In the formula, r represents the three-dimensional coordinate vector of the position to be determined; x, y and z represent the coordinate values ​​of the spatial position in the three coordinate directions, respectively; the superscript T indicates the vector transpose.

[0102] The spatial error field based on radial basis functions can be written as: (18).

[0103] In the formula, This represents the spatial error compensation value at position r; The interpolation weights corresponding to the i-th measurement point are represented by φ; φ represents the radial basis function. This represents the coordinates of the i-th measured point; n represents the total number of measured points participating in the interpolation. This represents the Euclidean distance.

[0104] The radial basis function in this embodiment is an inverse quadratic radial basis function or a verified Gaussian radial basis function.

[0105] (19).

[0106] In the formula, σ represents the radial basis function value at a distance of d; d represents the Euclidean distance between the location to be determined and the measuring point; σ represents the kernel width parameter. Equation (19) uses an inverse quadratic radial basis function.

[0107] This embodiment solves for the interpolation weights of each measuring point by restoring the corresponding steady-state temperature deviation of the spatial error matrix at each reference measuring point location.

[0108] (20).

[0109] In the formula, This represents the value of the error field at the j-th measurement point; This represents the steady-state deviation of the j-th measuring point under the reference working condition; j represents the measuring point number; and n represents the total number of measuring points.

[0110] In this step, the kernel width σ of the radial basis functions is affected by... The smoothness is an important parameter. If σ is too small, it will cause the error matrix to form spikes near the measurement point, resulting in unreasonable isolated high-temperature spots in the contour map; if σ is too large, it will cause local hot spots to be over-smoothed, weakening the risk identification ability of the winding ends and oil passage turning areas.

[0111] σ can be determined through cross-validation: some measurement points are temporarily used as validation points, and the remaining measurement points are used to construct... Then compare the prediction errors at the validation points. Select σ, which minimizes the validation error and ensures continuous cloud pattern, as the kernel width parameter for the current device or the current region.

[0112] When the measurement points are unevenly distributed, different interpolation strategies can be set according to the region. Higher sample weights or smaller kernel widths are used for the winding ends, leakage flux concentration areas, and oil passage transition areas to retain local deviations; smoother interpolation is used for the outer area of ​​the oil tank with a small temperature gradient to avoid amplifying the measurement point noise.

[0113] Should be with Use the same node order or the same spatial index. If the basic temperature field contour map is stored in matrix form, then It should also be stored as a matrix of the same size; if the basic temperature field contour map is stored in the form of a node table, then The node number should be stored as the primary key. This consistency is a prerequisite for the subsequent point-by-point stacking of equation (22).

[0114] In the boundary region, the radial basis function may introduce extrapolation errors due to the lack of measurement points on the outer side. This can be addressed by setting virtual boundary points, reducing the confidence level of boundary extrapolation, or limiting the correction amplitude in the boundary region. These measures do not change the underlying mechanism. The definition is simply to improve the stability of boundary cloud maps in engineering implementation.

[0115] In engineering implementation, The calculation can be performed point-by-point according to the grid nodes of the basic temperature field contour map and stored as a... Error matrices of the same dimension can also be calculated separately for regions such as windings, oil channels, and insulation structures. If the number of measurement points is small, priority should be given to ensuring that measurement points in hotspot-sensitive areas and structural transition areas participate in interpolation; if the number of measurement points is large, regularization or kernel width cross-validation can be used to avoid error field oscillations caused by local measurement point noise.

[0116] S4: Determine the operating condition correction coefficient based on the load rate and harmonic distortion rate.

[0117] In one specific implementation, Spatial error matrix Typically, the data is constructed under reference operating conditions, but actual operating conditions may have different β and H values. To avoid re-performing full interpolation for each operating condition, this invention introduces a condition correction coefficient kc to characterize the modulation effect of the operating condition on the error amplitude. In this embodiment, the condition correction coefficient includes a load correction term and a harmonic correction term. The load correction term is determined by a nonlinear fitting function of the load rate, and the harmonic correction term is determined by the deviation of the harmonic distortion rate from the reference harmonic distortion rate.

[0118] Specifically, kc is first determined using a nonlinear fitting method based on the load rate. When the harmonic distortion rate H deviates from the reference operating condition, the influence of H on the correction amplitude is reflected by selecting the reference error matrix, calibrating sample grouping, or using two-dimensional lookup table parameters, thereby avoiding multiple contradictory meanings for the same symbol.

[0119] The operating condition correction factor kc can be fitted using a nonlinear exponential form.

[0120] .

[0121] In the formula, kc represents the correction coefficient; β represents the load rate; A, B, C, and D all represent equipment characteristic fitting constants; and exp represents the natural exponential function. A, B, C, and D are only used for the coefficient fitting relationship under this operating condition.

[0122] Figure 2 The fitting relationship between the error field correction coefficient and the load rate is shown. Figure 2 It can be seen that kc increases non-linearly with increasing load rate, indicating that in the high load range, the underestimated error amplitude in the basic simulation does not increase linearly. Therefore, when reconstructing multi-condition contour maps, a fixed compensation ratio cannot be simply used; instead, the corresponding kc value or fitting function should be called according to the range in which β is located.

[0123] When H deviates significantly from the reference harmonic distortion rate, multiple harmonic level kc calibration samples can be established under the same β, or... This can be written as a two-dimensional lookup table function determined by both the load term and the harmonic term. Regardless of whether a fitting function or a lookup table function is used, k is only used for modulation. The amplitude remains unchanged. The spatial grid structure remains unchanged. The basic constraints that reflect steady-state deviations at the reference measurement point.

[0124] In this step The calibration should use the same as Check the independent verification conditions. If the fitted function has a small error within the existing conditions, but deviates significantly in adjacent conditions, it indicates that the function form of kc is too rigid or the sample coverage is insufficient. The calibration sample should be increased or a segmented lookup table method should be used.

[0125] When the load factor β is low, the overall heat source of the converter transformer is small, and the impact of local harmonic additional losses on the cloud pattern may not be significant; when β is close to the rated value or short-term overload, the additional losses and the basic copper loss work together, and kc may show a nonlinear growth. Equation (21) adopts an exponential relationship precisely to express this nonlinear trend.

[0126] When H varies over a wide range, it can be... The terms can be broken down into load calibration terms and harmonic correction terms, or multiple β-fit curves can be established by grouping by H. Regardless of the implementation method, this invention maintains the meaning of kc as the operating condition correction coefficient and no longer writes kc as other physical quantities.

[0127] S5: Reconstruct the temperature field cloud map based on the basic temperature field, spatial error matrix, and operating condition correction coefficient to obtain the corrected temperature field cloud map.

[0128] In one specific implementation, the corrected temperature field cloud map of this embodiment is equal to the sum of the base temperature field cloud map and the spatial error matrix modulated by the operating condition correction coefficient.

[0129] (twenty two).

[0130] In the formula, This represents the corrected temperature field output. The base temperature field is represented by r; the spatial location is represented by β; the load factor is represented by H; the harmonic distortion rate is represented by kc; and the correction factor is represented by kc. This represents the spatial error matrix.

[0131] Specifically, when executing formula (22), each node should be read. and And calculate kc based on the current β and H. If at a certain node If a problem is missing, out of bounds, or has low confidence, the system can use neighborhood completion or maintain the existing neighborhood. The original value is recorded, and the low confidence level of the node is also recorded in the output cloud map.

[0132] After outputting, the hotspot temperature, hotspot coordinates, and the structural region to which the hotspot belongs should be provided simultaneously. Outputting only a color cloud map is insufficient to support operation and maintenance decisions; hotspot coordinates and region labels can help engineers determine whether the hotspot is located at the end of the winding, at the bend of the oil passage, or at a change in insulation thickness.

[0133] Finally, the base temperature field, spatial error matrix, and operating condition correction coefficients are combined to obtain the corrected temperature field cloud map.

[0134] S6: Calculate the temperature error, hot spot error, and operating condition extrapolation degree of the measured point using the measured temperature of the verification point, and determine whether the corrected temperature field cloud map can be output as a high-confidence result based on the preset error threshold.

[0135] In one specific implementation, when the temperature error at the measuring point, the hot spot error, and the extrapolation degree of the operating condition exceed the preset error threshold, the corresponding operating condition is marked as an operating condition to be verified, and offline test sampling, simulation recalculation, or parameter recalibration is triggered.

[0136] Specifically, to ensure To ensure engineering reliability, some measuring points or operating conditions should be used as verification data, excluding them from the fitting of interpolation weights and operating condition coefficient parameters, and specifically used for verifying and correcting the results. During verification, the verification measuring points... and The extracted values ​​at the corresponding coordinates are compared, and the maximum error, root mean square error, and hotspot error are calculated. The root mean square error is only used as an overall accuracy indicator. The maximum error and temperature rise trend error of hotspot areas should also be calculated separately to avoid the overall average indicator masking the risk of local hotspots.

[0137] The verification process should also distinguish between measurement point temperature error and temperature field spatial distribution error. Measurement point temperature error is used to determine whether the value is close to the measured temperature; temperature field spatial distribution error is used to determine whether the hot spot region, temperature gradient direction, and high-temperature zone length conform to engineering understanding. If the measurement point temperature error is small but the hot spot position has shifted significantly, the measurement point-mesh mapping and kernel width σ should be checked first, rather than directly adjusting kc.

[0138] When the root mean square error, hotspot error, or extrapolation distance of the operating condition exceeds a preset threshold, the system should not directly output a high-confidence contour map, but should instead mark the operating condition as a condition requiring verification. Operating conditions requiring verification can proceed to offline testing for sample replenishment, high-precision simulation recalculation, or parameter recalibration; the original parameters should be retained on the online side. ,current , , Version number and input conditions are included to ensure the results are traceable.

[0139] The verification should include at least three types of indicators: the first type is the temperature error at the measuring point, used to judge... Is it close to the known measuring point? The second category is hotspot error, used to determine whether the location and range of high-temperature areas conform to structural understanding; the third category is the degree of operating condition extrapolation, used to determine whether the current β and H exceed the range of existing calibration samples.

[0140] If the temperature error at the measuring point is within acceptable limits but the hotspot morphology is clearly abnormal, the measuring point-grid mapping should be checked first. The core width; if the hotspot morphology is reasonable but the overall temperature is too high or too low, priority should be given to checking the core width; The calibration parameters; if both types of errors are abnormal at the same time, the basic rapid finite element model and experimental data should be checked back to S1.

[0141] High-confidence outputs, attention outputs, and outputs pending verification can be marked differently. High-confidence outputs can directly proceed to equipment condition assessment; attention outputs prompt engineers to check hotspot areas and extrapolation levels; outputs pending verification should not be used as the sole basis for insulation life assessment and should trigger offline recalculation or supplementary testing.

[0142] All verification results should be saved in conjunction with the model version, error matrix version, and load factor version. This way, if it is found later that the parameters of a certain version are not applicable, the load conditions and contour maps affected by that version can be traced back, avoiding the online system from using unverified historical parameters for a long time.

[0143] Without altering the core principles of this invention, the basic temperature field Output can be from finite element, finite volume, finite difference, or validated equivalent thermal network models; measured temperature The data can originate from fiber optic gratings, embedded thermocouples, oil temperature sensors, infrared inspection results, or high-precision recalculation results; the spatial interpolation function φ can be a Gaussian radial basis function, thin-plate spline, multiple quadratic functions, or a Kriging interpolation function; correction coefficients... Exponential, piecewise polynomial, lookup table interpolation, or regression models with monotonicity constraints can be used, as long as they still utilize the measurement point deviation for construction. And through the working condition correction factor Post-processing compensation is an equivalent implementation of the present invention.

[0144] From the data flow perspective, S1 outputs a basic data packet with a working condition number. The data packet includes at least β, H, The outputs of S2 and S3 are: measurement point coordinates, measurement point temperature sequence, and model version; the outputs of S2 and S3 are error matrix data packages that can be used for global compensation, which include at least the measurement point-grid mapping relationship, steady-state deviation, interpolation weights, and E(r); the outputs of S4, S5, and S6 are corrected cloud map data packages that can be directly used for engineering verification, which include at least the following: , Hotspot locations, error statistics, and confidence level markers.

[0145] The three data packets mentioned above are arranged in a unidirectional, progressive manner. Each data packet must be verified before it can be used in the calculation of the next. The next data packet must not overwrite the original measured values ​​and basic simulation values ​​in the previous data packet. This avoids confusion about data sources during subsequent parameter recalibration and ensures the stability of the meanings of the three core objects in the patented solution: "basic temperature field cloud map, error matrix, and operating condition coefficients."

[0146] During project implementation, if an abnormality is detected in the temperature measurement channel, the operating condition exceeds the calibration range, or the interpolation matrix is ​​ill-conditioned under a certain operating condition, the system can stop the high-confidence output and enter the pending review process.

[0147] The steps in this embodiment are not loosely parallel, but rather passed step-by-step through a data link: "operating condition parameters—basic temperature field cloud map—measuring point deviation—error matrix—operating condition modulation—result verification." The output of each stage serves as the input for the next stage, and the corresponding timestamp, operating condition number, model version, and measuring point number are retained to facilitate tracing the source of errors during subsequent verification.

[0148] This embodiment uses Figure 1 The three-stage main process outlined in the document serves as the general framework: the first stage corresponds to the construction and basic simulation of the electrothermal joint test platform; the second stage corresponds to the analysis of the node temperature deviation characteristics under harmonic effects; and the third stage corresponds to the correction of the spatial error matrix equation. The subsequent additions of measurement point mapping, steady-state identification, parameter calibration, system implementation, and version verification are all detailed implementation stages of the above main process.

[0149] Compared to methods that rely solely on basic fast finite element simulation, this invention utilizes the spatial error matrix... Compensating for local contour plot deviations caused by harmonic losses allows for a more accurate reconstruction of temperature distributions at winding ends, oil passage transition zones, and leakage flux concentration areas. Compared to methods that directly perform full-frequency strongly coupled simulations, this invention offers advantages over existing methods. By performing post-processing corrections on the basis of existing data, and avoiding repeated solutions to large-scale coupled equation systems, the reconstruction of complex working condition cloud maps can be shortened from several days to minutes or less.

[0150] From a computational efficiency perspective, this invention transforms the localized temperature deviations caused by high-frequency harmonics into storable and retrievable parameters. and In the online phase, there is no need to repeatedly rebuild the full-frequency electromagnetic-thermal coupling model. This advantage is particularly evident for design verification scenarios that require scanning a large number of load factors and harmonic combinations.

[0151] In terms of accuracy, this invention does not rely solely on empirical correction coefficients, but rather utilizes the deviation of measured points to constrain the spatial error matrix. Measured information is then input... Then, the local hot spot deviation can be propagated to the global cloud map, so that the cloud map retains the spatial continuity of the basic simulation and absorbs the correction effect of the experimental data on the real equipment.

[0152] From a maintainability perspective, , and Managing these three components separately facilitates local updates when equipment structure, sensor layout, or operating conditions change. If only the operating condition coefficients need updating, there is no need to regenerate the base temperature field; if only new measurement point samples are added, only the error matrix needs to be updated.

[0153] Compared to methods that only perform single-point temperature correction, this invention utilizes radial basis functions to expand discrete measurement point deviations into a global error matrix, enabling the output of continuous contour maps instead of isolated measurement point values. Compared to simple, uniform temperature rise translation methods, this invention introduces… , and Spatial indicators can reflect the fact that errors are uneven at different locations, and prevent local hotspots from being masked by average errors.

[0154] This invention also has good engineering deployability. The fitting constants A, B, C, D, etc., are determined by the equipment calibration data and can be adjusted according to different capacities, different cooling structures, and different winding arrangements; , and All versions can be stored as traceable versions; the online reconstruction process does not directly modify the basic rapid finite element model, facilitating auditing, verification, and parameter rollback. These features enable this invention to serve both rapid solution comparison in the design phase and hotspot risk assessment in the operation and maintenance phase.

[0155] Example 2: Embodiment 2 of the present invention provides a converter transformer temperature field cloud map reconstruction system, such as... Figure 3 As shown, the system includes a data acquisition interface, a test calibration module, a basic simulation module, a measurement point grid mapping module, a spatial error matrix generation module, a working condition coefficient fitting module, a cloud map reconstruction module, a model and matrix storage, an alarm output interface, a cloud map output interface, and a version management module.

[0156] Specifically, it includes: The basic simulation module is used to simulate and obtain the load rate, harmonic distortion rate, basic temperature field, measurement point coordinates, and measurement point temperature sequence of the converter transformer using the basic fast finite element temperature field model. The measuring point mesh mapping module is used to establish a spatial mapping between measured measuring points and simulated mesh nodes based on the measuring point coordinates and temperature sequence. The spatial error matrix generation module is used to determine the steady-state temperature deviation based on the deviation between the measured temperature and the basic simulation temperature, and to construct the spatial error matrix based on the steady-state temperature deviation through radial basis function interpolation. The operating condition coefficient fitting module is used to determine the operating condition correction coefficient based on the load rate and harmonic distortion rate. The cloud map reconstruction module is used to reconstruct the temperature field cloud map based on the basic temperature field, spatial error matrix, and operating condition correction coefficients to obtain a corrected temperature field cloud map.

[0157] In terms of hardware deployment, the data acquisition interface can access operational data through a station-side monitoring system, experimental data acquisition instrument, edge gateway, or local database. The algorithm processing host does not need to be directly connected to all sensors; as long as it can obtain timestamped current, harmonic, temperature, and cooling status data, it can execute the method of this invention.

[0158] The algorithm processing host can be a CPU, GPU, NPU, or a combination thereof. The online reconstruction phase mainly performs matrix reading, function calculation, and node-by-node addition and multiplication, with a computational load far lower than that of full-frequency strongly coupled simulation. The experimental calibration module is used for sample updating and calibration. Its offline calibration phase has a large computational load and can be deployed on a simulation workstation or server.

[0159] The model and matrix memory should include , , Parameters and verification records are stored separately. This belongs to the basic fast finite element model output. It belongs to the error compensation model. It belongs to the operating condition modulation model. The three versions are independent but differ in output. Call the function by version combination.

[0160] The system should have a parameter publishing mechanism. After the offline side completes the addition of new test samples, interpolation matrix updates, and refitting of operating condition coefficients, it should undergo verification error checks before being published to the online side. The online side should only call the published version and not directly write calibration parameters, thereby reducing the risk of misoperation.

[0161] The alarm output interface can output results to station-side monitoring systems, design verification software, or operation and maintenance diagnostic platforms. The output content includes current operating conditions, hotspot temperature, hotspot coordinates, hotspot region, root mean square error, maximum error, operating condition extrapolation markers, and cloud map file path.

[0162] Figure 3 The data acquisition interface, algorithm processing host, model and matrix storage, and cloud map output interface in the system correspond to data acquisition in S1, algorithm processing in S2, S3, and S4, parameter storage in S3 and S4, and result output in S6, respectively.

[0163] The data acquisition interface receives operating current, harmonic current, oil temperature, ambient temperature, cooler status, and measuring point temperature. The test calibration module manages test data under different combinations of β and H. The basic simulation module outputs... The measurement point-mesh mapping module is used to perform... Calculate and extract The spatial error matrix generation module is used to solve for the interpolation weights. And generate The working condition coefficient fitting module is used to calibrate working condition coefficient parameters such as A, B, C, and D. The cloud map reconstruction module is used to execute equation (22) and output the results. .

[0164] The model and matrix memory should at least store the basic fast finite element model version, mesh node coordinates, and measurement point coordinates. , Parameters, Correction Results And verification error records. The version management module is used to ensure online calls... and All data comes from verified versions to prevent unverified parameters from being directly entered into the operation and maintenance system. The alarm output interface can output attention, warning, or alarm levels based on hotspot temperature, hotspot location, root mean square error, and degree of operating condition extrapolation.

[0165] During system implementation, the online side can be deployed on an industrial computer at the station, an edge server, or an algorithm processing host, with processors including CPUs, GPUs, or combinations thereof. Since online reconstruction primarily involves matrix reading, coefficient calculation, and node-by-node addition and multiplication, its computational complexity is far lower than that of full-frequency strongly coupled simulation, making it suitable for rapid scanning under multiple operating conditions and on-site operation and maintenance verification. The offline side is responsible for adding new test samples, high-precision simulation recalculation, and parameter version releases. The two sides are connected through a model and matrix memory.

[0166] Example 3: Embodiment 3 of the present invention provides a computer-readable storage medium storing a computer program adapted for loading by a processor and executing the steps in the converter transformer temperature field cloud map reconstruction method as described in Embodiment 1 of the present invention.

[0167] Example 4: Embodiment 4 of the present invention provides a computer device, the device comprising: A processor, adapted to execute computer programs; A computer-readable storage medium storing a computer program, which, when executed by the processor, implements the steps in the converter transformer temperature field cloud map reconstruction method as described in Embodiment 1 of the present invention.

[0168] Example 5: Embodiment 5 of the present invention provides a computer program product or computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the steps in the converter transformer temperature field cloud map reconstruction method as described in Embodiment 1 of the present invention.

[0169] The steps and methods involved in Examples 2, 3, 4 and 5 above correspond to those in Example 1. For specific implementation methods, please refer to the relevant description section of Example 1.

[0170] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed in this application can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0171] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. A computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the flow or function according to the embodiments of this application is generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in or transmitted through a computer-readable storage medium. The computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless means. The computer-readable storage medium can be any available medium that a computer can access or a data processing device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium, an optical medium, or a semiconductor medium, etc.

[0172] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for reconstructing temperature field cloud maps of a converter transformer, characterized in that, Includes the following steps: The load rate, harmonic distortion rate, base temperature field, measuring point coordinates, and measuring point temperature sequence of the converter transformer were obtained by using a basic fast finite element temperature field model for simulation. Establish a spatial mapping between measured points and simulation grid nodes based on the measured point coordinates and temperature sequence; The steady-state temperature deviation is determined based on the difference between the measured temperature and the basic simulation temperature, and a spatial error matrix is ​​constructed based on the steady-state temperature deviation through radial basis function interpolation. Determine the operating condition correction factor based on the load rate and harmonic distortion rate; The temperature field cloud map is reconstructed based on the basic temperature field, spatial error matrix, and operating condition correction coefficient to obtain the corrected temperature field cloud map.

2. The method for reconstructing the temperature field cloud map of a converter transformer as described in claim 1, characterized in that, The specific steps for obtaining the converter transformer load rate, harmonic distortion rate, base temperature field, measuring point coordinates, and measuring point temperature sequence using a basic fast finite element temperature field model are as follows: Obtain the geometric structure, winding region, oil passage location, material thermal properties, oil flow boundary, temperature boundary, rated current, operating current, and harmonic current of the converter transformer. An experimental platform for the combined electrothermal effect based on a basic rapid finite element temperature field model was built, and basic simulations were performed.

3. The method for reconstructing the temperature field cloud map of a converter transformer as described in claim 1, characterized in that, The basic rapid finite element temperature field model retains the spatial relationship of the winding ends, conductor transposition area, oil channel turning area, insulation structure near the pressure plate and leakage magnetic concentration area. For some chamfers, bolt holes or assembly gaps, equivalent treatment is performed without changing the main heat path.

4. The method for reconstructing the temperature field cloud map of a converter transformer as described in claim 1, characterized in that, The load factor is determined by the ratio of the effective value of the actual operating current to the effective value of the rated current, and the harmonic distortion rate is determined by the effective values ​​of each harmonic current and the effective value of the fundamental current.

5. The method for reconstructing the temperature field cloud map of a converter transformer as described in claim 1, characterized in that, The electrothermal combined effect test platform includes at least a programmable AC constant current power supply, a harmonic injection unit, an oil flow circulation temperature control chamber, a multi-channel fiber optic temperature demodulator, a sample clamping and insulating oil pretreatment device, and a simulation workstation. The programmable AC constant current power supply and harmonic injection unit are used to apply different load rates and harmonic distortion rates; the oil flow circulation temperature control chamber is used to maintain the inlet oil temperature and temperature boundary; the multi-channel fiber optic temperature demodulator is used to collect the temperature of the pre-embedded measuring points; and the simulation workstation is used to establish a basic fast finite element temperature field model that is geometrically consistent with the test sample.

6. The method for reconstructing the temperature field cloud map of a converter transformer as described in claim 1, characterized in that, The specific steps for establishing the spatial mapping between measured points and simulation mesh nodes based on the measured point coordinates and temperature sequences are as follows: The nearest neighbor rule is used to determine the simulation grid node that is closest to the measured point, and the basic simulation temperature of the simulation grid node is extracted from the basic temperature field cloud map.

7. The method for reconstructing the temperature field cloud map of a converter transformer as described in claim 1, characterized in that, The steady-state temperature deviation is obtained by averaging the dynamic temperature deviations at multiple sampling times after the rate of change of deviation within multiple consecutive sampling intervals does not exceed the steady-state determination threshold.

8. The method for reconstructing the temperature field cloud map of a converter transformer as described in claim 1, characterized in that, The interpolation weights for each measuring point are solved by restoring the corresponding steady-state temperature deviation of the spatial error matrix at each reference measuring point.

9. The method for reconstructing the temperature field cloud map of a converter transformer as described in claim 1, characterized in that, The radial basis functions are either inverse quadratic radial basis functions or verified Gaussian radial basis functions.

10. The method for reconstructing the temperature field cloud map of a converter transformer as described in claim 1, characterized in that, The operating condition correction factor includes a load correction term and a harmonic correction term. The load correction term is determined by a nonlinear fitting function of the load rate, and the harmonic correction term is determined by the deviation of the harmonic distortion rate from the reference harmonic distortion rate.

11. The method for reconstructing the temperature field cloud map of a converter transformer as described in claim 1, characterized in that, The measured temperature at the verification point is used to calculate the temperature error, hot spot error, and operating condition extrapolation degree. Based on the preset error threshold, it is determined whether the corrected temperature field cloud map can be output as a high-confidence result. When the temperature error, hot spot error, and operating condition extrapolation degree exceed the preset error threshold, the corresponding operating condition is marked as an operating condition to be verified, and offline test sampling, simulation recalculation, or parameter recalibration is triggered.

12. A converter transformer temperature field cloud map reconstruction system, characterized in that, include: The basic simulation module is used to simulate and obtain the load rate, harmonic distortion rate, basic temperature field, measurement point coordinates, and measurement point temperature sequence of the converter transformer using the basic fast finite element temperature field model. The measuring point mesh mapping module is used to establish a spatial mapping between measured measuring points and simulated mesh nodes based on the measuring point coordinates and temperature sequence. The spatial error matrix generation module is used to determine the steady-state temperature deviation based on the deviation between the measured temperature and the basic simulation temperature, and to construct the spatial error matrix based on the steady-state temperature deviation through radial basis function interpolation. The operating condition coefficient fitting module is used to determine the operating condition correction coefficient based on the load rate and harmonic distortion rate. The cloud map reconstruction module is used to reconstruct the temperature field cloud map based on the basic temperature field, spatial error matrix, and operating condition correction coefficients to obtain a corrected temperature field cloud map.

13. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the converter transformer temperature field cloud map reconstruction method as described in any one of claims 1-11.

14. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program adapted to be loaded by a processor and executed as described in any one of claims 1-11.

15. A computer device, characterized in that, include: A processor, adapted to execute computer programs; A computer-readable storage medium storing a computer program, which, when executed by the processor, implements the converter transformer temperature field cloud map reconstruction method as described in any one of claims 1-11.

Images