A method for locating a turn-to-turn short circuit thermal fault of an oil-immersed transformer winding based on temperature distribution characteristics of an outer wall of a box

Abstract

The present application relates to a kind of oil-immersed transformer winding interturn short-circuit thermal fault positioning method based on box outer wall temperature distribution characteristics, belong to oil-immersed power transformer condition monitoring and fault diagnosis technical field.Through establishing three-dimensional fluid-heat coupling model, additional heat source is applied to candidate fault area to carry out forward simulation, and the dynamic thermal response database between internal heat source and box outer wall temperature response is constructed;Box outer wall temperature is collected and healthy temperature variation model is established, and abnormal temperature residual is extracted;Combined with dynamic thermal response matrix, non-negative sparse inversion is used to calculate the additional heat source intensity of each candidate area, and the suspected interturn short-circuit fault position is determined.The present application can realize the online positioning of small interturn short-circuit thermal fault without internal temperature measurement, improve early fault detection and positioning accuracy.

Description

Technical Field

[0001] This invention belongs to the field of condition monitoring and fault diagnosis technology for oil-immersed power transformers. Specifically, it relates to a method for locating inter-turn short-circuit thermal faults in oil-immersed transformer windings based on the temperature distribution characteristics of the outer wall of the transformer tank. More specifically, it relates to a method for locating minute inter-turn short-circuit thermal faults in oil-immersed transformer windings by establishing a three-dimensional fluid-heat transfer coupling model, performing forward simulation of additional heat sources in different candidate inter-turn short-circuit fault regions of the windings, constructing a dynamic thermal response database between internal additional heat sources and the temperature response of the outer wall of the transformer tank, and eliminating the influence of normal temperature rise and inverting non-negative sparse heat sources based on the measured temperature of the outer wall of the transformer tank, thereby realizing the location of minute inter-turn short-circuit thermal faults in oil-immersed transformer windings. Background Technology

[0002] Oil-immersed power transformers are crucial equipment in power systems, responsible for voltage transformation, power transmission, and power distribution. Their operational reliability directly impacts the safe and stable operation of the power transmission and distribution system. Transformers operate in a multi-physical environment influenced by electromagnetic, thermal, mechanical, and insulation fields over extended periods. The winding insulation may gradually deteriorate due to factors such as electromagnetic vibration, mechanical stress, localized overheating, material aging, and manufacturing defects.

[0003] Inter-turn short circuit is one of the more typical internal faults in oil-immersed transformers. When the insulation between some adjacent turns is damaged, a local short circuit loop is formed. Especially in the early stage of inter-turn short circuit, the number of turns involved in the short circuit is small, and the abnormal circulating current and additional losses generated in the fault area are relatively small compared to the overall transformer load and normal copper losses. Therefore, the changes in transformer terminal current, voltage, and total losses may still be within the normal operating fluctuation range.

[0004] Existing fault diagnosis methods for oil-immersed transformers typically rely on terminal current, terminal voltage, dissolved gases in the oil, partial discharge, vibration, or acoustic signals for condition assessment. However, for minor inter-turn faults caused by short circuits in a small number of turns, these characteristic parameters are often insufficiently apparent in the early stages of the fault. The fault usually needs to develop to a certain extent before it can be effectively identified, thus failing to meet the need for timely location of early-stage minor inter-turn short-circuit faults.

[0005] On the other hand, since the transformer windings are located inside a sealed oil tank, it is difficult to directly place a large number of temperature sensors inside the windings during actual operation. In contrast, installing temperature sensors on the outer wall of the tank is less difficult to implement and can continuously obtain transformer surface temperature information without damaging the internal insulation and sealing structure of the transformer.

[0006] However, the localized abnormal heat generated inside the windings needs to pass through multiple heat transfer links, including the winding insulation, transformer oil, and the outer wall of the enclosure, before it can be transferred to the outer wall. This heat transfer process is simultaneously affected by solid heat conduction, natural or forced circulation of the insulating oil, radiator heat exchange, and changes in external environmental conditions. As the internal heat source propagates outward, it undergoes spatial diffusion, spatial migration, and time delay. Therefore, a temperature rise at a certain location on the outer wall of the enclosure cannot be simply equated to a fault in the corresponding internal winding.

[0007] Furthermore, during actual operation, changes in load, ambient temperature, and the operating status of cooling devices such as fans and oil pumps can all cause changes in the overall temperature field of the outer wall of the enclosure. If the internal fault of the winding is judged directly based on the absolute temperature of the outer wall of the enclosure or the temperature of a single measuring point, it is easy to mistake temperature fluctuations caused by changes in normal operating conditions for internal thermal faults, and it may also cause weak local abnormal temperature rises to be masked by normal overall temperature rises.

[0008] Therefore, it is necessary to establish a quantitative relationship between the location and intensity of local heat sources inside the winding and the temperature changes at multiple points on the outer wall of the enclosure. By establishing an offline dynamic thermal response model between the inner and outer walls and eliminating the influence of normal load, ambient temperature and cooling status on the outer wall temperature of the enclosure online, the location and intensity of additional internal heat sources can be inverted based on the spatial distribution of the remaining abnormal temperature, thereby improving the early identification capability and spatial positioning accuracy of minor inter-turn short circuit faults. Summary of the Invention

[0009] The purpose of this invention is to provide a method for locating inter-turn short-circuit thermal faults in oil-immersed transformer windings based on the temperature distribution characteristics of the outer wall of the transformer casing. This method addresses the problems in the prior art, such as the lack of obvious electrical characteristics of minor inter-turn short-circuit faults, the difficulty in directly measuring the internal temperature of the windings, the susceptibility of the outer wall temperature to changes in load and environment, and the absence of a simple one-to-one correspondence between the outer wall temperature and the location of the internal fault.

[0010] This invention first establishes a three-dimensional fluid-heat coupling model using transformer structural parameters, material parameters, and operating parameters. Reference additional heat sources are sequentially applied to different candidate inter-turn short-circuit fault regions. Through forward temperature field simulation, the dynamic temperature response of the outer wall of the transformer tank corresponding to different internal heat source locations and intensities is obtained, forming a dynamic thermal response database. Subsequently, the placement of temperature sensors on the outer wall of the transformer tank is determined based on the differences in the external wall temperature responses corresponding to different candidate fault regions. During online operation, a healthy operating temperature change model is established based on healthy operating temperature data to eliminate the overall temperature rise caused by load changes, ambient temperature changes, and cooling state changes, obtaining abnormal temperature residuals. Finally, the dynamic thermal response matrix is ​​matched according to the current operating state, and the additional heat source intensity corresponding to each candidate fault region is calculated through non-negative sparse inversion, thereby identifying the main suspected inter-turn short-circuit thermal fault regions.

[0011] Specifically, the present invention includes the following steps.

[0012] S1. Establish a three-dimensional fluid-heat coupling model of the three-phase oil-immersed transformer to be tested.

[0013] The three-dimensional fluid-heat coupling model includes at least the iron core, A-phase winding, B-phase winding, C-phase winding, solid insulation structure, transformer oil, outer wall of the tank, and heat dissipation structure, and is used to describe the solid heat conduction inside the transformer, the heat transfer of the insulating oil flow, and the heat transfer process between the outer wall of the tank and the environment.

[0014] For the solid region formed by the core, windings, solid insulation structure, and outer wall of the enclosure, the transient heat transfer equation is expressed as:

[0015] in, Indicates the density of a solid region; Indicates the specific heat capacity of a solid region; Indicates the thermal conductivity of the solid region; Indicates the temperature of the solid region; This represents the volumetric heat source density of the solid region under normal operating conditions. N Indicates the number of candidate inter-turn short-circuit fault regions; ) indicates the first j Additional heat generation power of each candidate inter-turn short-circuit fault region; Indicates the first j Normalized spatial heat source distribution function for candidate inter-turn short-circuit fault regions; Indicates spatial location.

[0016] For the transformer oil region, its flow and heat transfer equation is expressed as:

[0017] in, Indicates the density of transformer oil; Indicates the specific heat capacity of transformer oil; Indicates the thermal conductivity of transformer oil; Indicates the transformer oil temperature; Indicates the transformer oil flow velocity; This indicates the volumetric heat source density of the transformer oil region under normal operating conditions.

[0018] The additional heat sources are spatially distributed according to the candidate inter-turn short-circuit fault regions, and satisfy the following:

[0019] in, Indicates the first j One candidate inter-turn short-circuit fault region; The unit is watts, which characterizes the total additional heat generation power of the candidate inter-turn short-circuit fault region; The unit is per cubic meter; This represents the additional volumetric heat source density applied to the heat transfer control equation.

[0020] For transformer oil flow, the oil velocity field can be solved using the incompressible flow continuity equation and the momentum conservation equation. For natural oil circulation operation, the buoyancy effect caused by temperature difference can be approximated using temperature-dependent density or Boussinesq.

[0021] Based on the actual structure of the transformer under test, further set the initial conditions, the coupling boundary conditions between the solid region and the transformer oil, the contact boundary conditions between different solid components, and the heat transfer boundary conditions between the outer wall of the tank and the external environment.

[0022] At the initial moment of the simulation, the initial temperature of each solid region and the transformer oil region can be set to the stable temperature field under the corresponding operating conditions; if stable temperature field data is lacking, it can be set to the ambient temperature or the actual measured initial shutdown temperature.

[0023] Temperature continuity and normal heat flow continuity conditions are set at the interfaces where the windings, core, solid insulation structure, and inner wall of the tank come into contact with transformer oil.

[0024] For the contact interface between different solid components, when the contact thermal resistance is negligible, temperature continuity and heat flow continuity conditions are adopted; when insulation layers, assembly gaps or contact thermal resistance need to be considered, the interface heat transfer relationship is set according to the corresponding contact thermal resistance.

[0025] The outer wall of the enclosure is subject to convective heat transfer boundary conditions between itself and the environment.

[0026] in, Indicates the thermal conductivity of the outer wall of the box; Indicates the temperature of the outer wall of the enclosure; Indicates ambient temperature; This represents the equivalent convective heat transfer coefficient between the outer wall of the enclosure and the environment. This represents the unit normal vector of the interface.

[0027] When the influence of environmental radiative heat transfer on the temperature response of the outer wall of the tank is not negligible, the influence of radiative heat transfer is further considered in the above-mentioned outer wall heat transfer boundary. The parameters used include the emissivity of the outer surface of the tank, the equivalent environmental radiative temperature, and the Stefan-Boltzmann constant.

[0028] When an oil-immersed transformer is equipped with a radiator, fan, or oil pump, an equivalent heat transfer boundary is set according to the corresponding cooling method. Natural cooling uses natural convection heat transfer parameters, fan operation uses corresponding forced convection heat transfer parameters, and oil pump operation adjusts the transformer oil flow and heat transfer parameters according to the oil flow velocity. The heat transfer coefficients, oil flow velocities, and environmental parameters corresponding to different cooling states form separate sets of operating state parameters.

[0029] S2 divides the candidate inter-turn short-circuit fault region of the three-phase winding and performs forward simulation of additional heat sources.

[0030] Candidate inter-turn short-circuit fault regions are divided according to the spatial positions of phase A, phase B, and phase C windings within the transformer. In one embodiment, each phase winding is divided axially into an upper region, a middle region, and a lower region, thereby forming nine candidate inter-turn short-circuit fault regions.

[0031] When it is necessary to further improve the radial fault location capability, the corresponding candidate fault area can be further divided into the inner winding area, the middle winding area, and the outer winding area along the winding radial direction, thereby increasing the number of candidate fault areas.

[0032] In each candidate inter-turn short-circuit fault region, the same or normalized reference additional heat power is applied. Transient fluid-heat transfer forward simulations are carried out under multiple representative load conditions, ambient temperature, and cooling conditions to obtain the internal temperature field and the outer wall temperature field of the enclosure under healthy conditions and after applying additional heat sources to different candidate inter-turn short-circuit fault regions.

[0033] S3 normalizes the temperature response obtained from the forward simulation and establishes a dynamic thermal response database.

[0034] First, the additional heat source in all candidate inter-turn short-circuit fault areas is set to zero to obtain the internal temperature field and the external wall temperature field of the enclosure under healthy operating conditions.

[0035] Subsequently, a reference additional heat power is applied to each candidate inter-turn short-circuit fault region, and the temperature change at different locations on the outer wall of the enclosure before and after the application of the reference additional heat source is calculated. The temperature change is then divided by the corresponding reference additional heat power to obtain the temperature response of the outer wall of the enclosure under unit power conditions.

[0036] time t The corresponding unit step thermal response matrix is ​​expressed as:

[0037] in, Indicates time t The unit step thermal response matrix; Indicates the firstj Temperature response vectors formed at different temperature measurement locations on the outer wall of the enclosure when a unit additional heat power is applied to a candidate inter-turn short-circuit fault region; N Indicates the number of candidate inter-turn short-circuit fault regions; t This indicates the response time after the application of an additional heat source.

[0038] Under discrete-time conditions, the unit step thermal response corresponding to different response times is processed according to the time series to obtain the dynamic thermal response characteristics of different response stages. Furthermore, the thermal response data corresponding to different loads, different ambient temperatures, different cooling states, different candidate fault regions, and different response times are stored to form a dynamic thermal response database between the internal candidate heat source location, the intensity of the additional heat source, and the temperature change of the outer wall of the enclosure.

[0039] S4. Determine the placement of temperature sensors based on the dynamic thermal response database and collect the temperature of the outer wall of the enclosure.

[0040] Multiple candidate temperature sensor positions are pre-set on the outer wall of the enclosure. Each candidate position covers at least the spatial projection area of ​​phase A winding, phase B winding and phase C winding on the outer wall of the enclosure, and covers the upper, middle and lower parts of the corresponding spatial projection area of ​​each phase winding.

[0041] Furthermore, candidate temperature measurement positions are spatially paired on the front and rear opposite sidewalls or the left and right opposite sidewalls of the fuel tank. The final temperature sensor position is selected based on the temperature response differences generated by different candidate inter-turn short-circuit fault areas at each candidate temperature measurement position in the dynamic thermal response database, so that the selected temperature measurement position has good temperature response differences and positioning resolution for different internal heat sources.

[0042] In a preferred embodiment, nine temperature measuring points are arranged on the upper, middle, and lower parts of the A-phase winding, B-phase winding, and C-phase winding, respectively, on the front and rear side walls of the fuel tank, thus forming eighteen temperature measuring points on the outer wall of the tank. The corresponding measuring points on the front and rear of the fuel tank form a spatial pairing relationship to reduce the impact of local disturbances in the temperature field on the fault location results.

[0043] In practical applications, the number and specific installation location of temperature sensors can be adjusted according to the transformer size, winding structure, tank shape, heat dissipation structure, and required fault location spatial resolution, and are not limited to eighteen temperature sensors.

[0044] During online operation, temperature sensors installed at various temperature measurement locations on the outer wall of the enclosure continuously collect temperature data of the outer wall, forming a temperature measurement vector.

[0045] S5. Establish a healthy temperature change model based on healthy operating temperature data and extract abnormal temperature residuals. During normal operation of an oil-immersed transformer, changes in load, ambient temperature, and the operating status of the cooling system can cause common or related temperature changes at multiple temperature measurement points on the outer wall of the transformer tank. To avoid the overall temperature rise caused by these normal operating factors masking the weak abnormal temperature information formed by the additional heat source of local inter-turn short circuits, this invention establishes a healthy common temperature change model based on the outer wall temperature data of the transformer tank collected during healthy operation.

[0046] The temperature measurement vectors of the outer wall of the enclosure at multiple sampling times under healthy operating conditions are combined to form a healthy temperature sample set. The healthy average temperature vector is determined based on the healthy temperature sample set. The average health temperature vector is subtracted from each health temperature sample to obtain a centered health temperature sample.

[0047] In one implementation, the centralized healthy temperature sample can be decomposed using principal component analysis, singular value decomposition, or other methods capable of extracting common temperature change patterns from multiple measurement points. The main common temperature change basis vectors are then selected based on a preset cumulative contribution rate or a preset number of basis vectors, forming a healthy common temperature change basis matrix. .

[0048] Set the weight matrix The weight matrix is ​​used to reflect the measurement reliability, noise level, or validity of the current data at different temperature measurement points on the outer wall of the enclosure. The weight matrix can be a diagonal matrix, with its diagonal elements being the weight coefficients of the corresponding temperature measurement points.

[0049] Set time t The collected temperature measurement vector of the outer wall of the box is According to:

[0050] Obtain the projection coefficients of the current temperature data in the subspace of health public temperature change.

[0051] Further evidence:

[0052] Obtain the weighted orthogonal projection complement matrix corresponding to the subspace of health public temperature change.

[0053] Therefore, according to:

[0054] Obtain the residual of abnormal temperature.

[0055] The abnormal temperature residual mainly retains the local space temperature anomalies in the current outer wall temperature of the enclosure that cannot be explained by the healthy common temperature change model, thereby reducing the impact of load changes, ambient temperature changes, and normal cooling state changes on the location of inter-turn short-circuit thermal faults.

[0056] Furthermore, to ensure that the online abnormal temperature residual and the offline established dynamic thermal response model are in the same feature space, the dynamic thermal response matrix corresponding to the current operating state is... The transformation is performed using the same weighted orthogonal projection complement matrix as the measured temperature.

[0057] in, This represents the equivalent dynamic thermal response matrix in the residual space of abnormal temperatures.

[0058] S6, Match the dynamic thermal response matrix according to the current operating status, and establish an internal additional heat source inversion model. Based on the current cooling status of the transformer, a subset of operating status data corresponding to the current cooling status is determined from the dynamic thermal response database.

[0059] When the cooling state includes fan start / stop state, oil pump start / stop state, or different forced cooling levels, corresponding dynamic thermal response data subsets can be established for different discrete cooling states to avoid direct interpolation between cooling states with different physical properties.

[0060] After determining the data subset corresponding to the current cooling state, the corresponding operating state node is determined based on the current load and current ambient temperature. When the current load and ambient temperature match existing operating state nodes in the database, the corresponding dynamic thermal response matrix is ​​directly selected; when the current load or ambient temperature is between adjacent operating state nodes, interpolation is performed based on the dynamic thermal response matrices of adjacent operating state nodes to obtain the dynamic thermal response matrix corresponding to the current operating state. .

[0061] The weighted orthogonal projection complement matrix obtained using S5 Projecting the dynamic thermal response matrix:

[0062] This ensures that the dynamic thermal response matrix and the abnormal temperature residual are in the same orthogonal complementary space of healthy common temperature changes.

[0063] In containing K Within a set time window of each sampling moment, the abnormal temperature residuals corresponding to each sampling moment are sequentially combined to form the abnormal temperature residual of the time window:

[0064] Accordingly, the equivalent dynamic thermal response matrices corresponding to each sampling time are combined to form:

[0065] Therefore, the following is established:

[0066] in, This represents the additional heat source intensity vector for each candidate inter-turn short-circuit fault region. This represents the residual term resulting from model error, temperature measurement error, and unmodeled disturbances.

[0067] When the start time of a suspected inter-turn short-circuit fault is unknown, multiple candidate fault start times are set before the current analysis time window. For each candidate fault start time, the dynamic thermal response matrix corresponding to the response time is extracted from the dynamic thermal response database based on the time difference between each sampling time and the candidate fault start time, forming a candidate time window dynamic response matrix. Additional heat source inversion is performed and the temperature fitting error is calculated. The candidate fault start time with the smallest temperature fitting error or the highest location confidence is determined as the estimated fault start time, and the inversion result corresponding to the estimated fault start time is used as the fault location result.

[0068] S7 performs non-negative sparse inversion on the internal additional heat source and identifies the suspected fault location.

[0069] The inversion result of the additional heat source is obtained by the following formula:

[0070] in, This represents the vector of inversion results for the additional heat source intensity; This represents the additional heat source intensity vector for each candidate inter-turn short-circuit fault region; This represents the weight matrix corresponding to the set time window; This represents the sparse regularization parameter; Indicates the space smoothing regularization parameter; This indicates the abnormal temperature residual within the set time window; This represents the dynamic response matrix within a set time window; This represents the spatial adjacency difference matrix or the graph Laplace matrix.

[0071] Among them, the non-negativity constraint is used to ensure that the obtained additional heat source intensity meets the physical condition that the actual heat source power is non-negative; the sparsity constraint is used to make the additional heat source preferentially concentrated in a smaller number of candidate fault regions; the spatial smoothness constraint is used to suppress unreasonable spatial jumps when multiple adjacent candidate regions are affected by the same actual fault heat source.

[0072] The candidate fault region with the highest intensity of additional heat source obtained from the inversion was identified as the main suspected inter-turn short-circuit thermal fault region:

[0073] in, Indicates the area number of the main suspected inter-turn short-circuit thermal fault; Indicates the first j The inversion of additional heat source intensity for each candidate fault region.

[0074] When multiple spatially adjacent candidate fault regions simultaneously have high additional heat source intensity, the suspected fault range can be determined based on the spatial relationship between the multiple candidate fault regions, rather than using only a single candidate region as the location result.

[0075] To evaluate whether the internal additional heat source obtained from the inversion can adequately explain the anomalous temperature residuals on the outer wall of the chamber, the temperature fitting error is further calculated:

[0076] in, This indicates the temperature fitting error; This represents a predetermined positive number set to avoid a denominator of zero; its value can be determined based on the magnitude of the temperature data.

[0077] The smaller the temperature fitting error, the higher the consistency between the outer wall temperature response calculated based on the inverted additional heat source and the actual abnormal temperature residual.

[0078] The concentration of heat sources is further evaluated based on the spatial distribution of the intensity of additional heat sources in each candidate inter-turn short-circuit fault region. When the additional heat sources are mainly concentrated in one or a few candidate regions, it indicates that the fault location has a relatively clear spatial concentration; when the additional heat sources are dispersed in many candidate regions, it indicates that the location result has a high degree of uncertainty.

[0079] The prominence of the main suspected fault area is evaluated based on the difference in the intensity of the additional heat source between the region with the largest additional heat source and other candidate regions. When the intensity of the largest additional heat source is significantly higher than that of other regions, it indicates that the main suspected fault area has good regional identification; when the inverted additional heat source intensities of multiple regions are close to each other, it indicates that there may be regional confusion in the localization results.

[0080] The thermal response distinguishability can also be evaluated based on the differences in the external wall temperature response corresponding to different candidate fault regions in the dynamic thermal response database. When the temperature responses corresponding to different candidate regions are significantly different, it indicates that the current sensor layout and dynamic response matrix can effectively distinguish the locations of different internal heat sources; when the temperature responses of multiple candidate regions are highly similar, the reliability of the corresponding location results should be reduced.

[0081] In one implementation, the temperature fitting error can be converted into a normalized fitting confidence index:

[0082] in, This means that the input value is limited to a numerical range of 0 to 1.

[0083] Further evaluation can be conducted by normalizing the temperature fitting, the spatial concentration of additional heat sources, the prominence of major suspected fault areas, and the distinguishability of dynamic thermal responses in different candidate areas. The weights used for each evaluation component are non-negative and satisfy the following conditions:

[0084] This allows us to obtain the reliability of fault location.

[0085] When the maximum additional heat source intensity exceeds the preset heat source threshold, the fault location confidence exceeds the preset confidence threshold, and the same candidate fault area is identified as the main suspected fault area in multiple consecutive time windows, the corresponding fault location result is output.

[0086] The fault location results may include the suspected fault phase, the suspected fault axial position, the suspected fault radial position, the suspected fault spatial coordinates, the inverted additional heat source intensity, the estimated internal temperature and additional temperature rise, the temperature fitting error, and the location reliability.

[0087] The location results output by this invention are used to indicate the location of internal local thermal anomalies suspected to be caused by minor inter-turn short circuits in the winding. The final fault type can be further confirmed by combining power outage maintenance or other fault diagnosis methods.

[0088] Beneficial effects (1) By establishing a dynamic thermal response database between internal candidate heat sources and the outer wall temperature of the enclosure offline, the local abnormal heat sources that are difficult to measure directly are transformed into spatial temperature responses that can be obtained by the outer wall temperature sensor, thereby realizing the inversion of internal thermal faults from external temperature measurement information.

[0089] (2) The online fault location stage of this invention mainly utilizes the temperature data of the outer wall of the enclosure, without the need for terminal voltage, terminal current, internal oil temperature, dissolved gas in the oil, vibration or acoustic signals as necessary inputs for fault location. Therefore, it is particularly suitable for early micro-turn short-circuit thermal faults with a small number of short-circuit turns and no obvious changes in electrical parameters.

[0090] (3) By establishing a healthy public temperature change model, the overall temperature change caused by load change, ambient temperature change and normal cooling state change can be removed from the outer wall temperature of the box. This can highlight the abnormal temperature residual caused by local additional heat sources and reduce mispositioning caused by changes in normal operating conditions.

[0091] (4) The spatial division of the candidate fault area, the location of the sensor on the outer wall of the box and the dynamic thermal response database are correlated with each other, and the heat source is inverted by using the spatial temperature distribution of multiple temperature measurement points on the outer wall of the box. Compared with a single temperature measurement point, more information related to the fault phase and the fault axial position can be obtained.

[0092] (5) The present invention completes the three-dimensional temperature field calculation and candidate heat source scanning in the offline stage, and mainly performs temperature compensation, matrix matching and non-negative sparse inversion in the online stage, which can reduce the amount of calculation for online fault location and is suitable for online condition monitoring of transformers.

[0093] (6) The present invention simultaneously uses temperature fitting error, spatial concentration of additional heat source, prominence of main suspected fault area and distinguishability of temperature response of candidate area to evaluate the reliability of positioning results, which can avoid taking unstable or low distinguishability inversion results directly as fault conclusions.

[0094] Other advantages, objectives, and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination, or may be learned from practice of the invention. The objectives and other advantages of the invention can be realized and obtained through the following description. Attached Figure Description

[0095] To make the objectives, technical solutions, and advantages of the present invention clearer, the preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings, wherein: Figure 1 This is a schematic diagram of the overall structure and composition of the three-phase oil-immersed transformer of the present invention; Figure 2 This is a schematic diagram showing the division of the nine candidate inter-turn short-circuit fault regions in the three-phase winding of the present invention; Figure 3 This is a schematic diagram showing the arrangement of eighteen temperature measuring points on the front and rear side walls of the fuel tank of the present invention; Figure 4 This is a flowchart of the multiphysics coupling forward simulation of the present invention; Figure 5 This is a temperature comparison diagram of the eighteen measuring points on the outer wall of the enclosure of the present invention under healthy conditions and under the condition of a short circuit in the middle of phase B with 3 turns; Figure 6This is an additional temperature rise distribution diagram of the front and rear walls of the fuel tank in the B-phase middle 3-turn short-circuit state relative to the healthy state of the present invention. Figure 7 This is a graph showing the correlation between the additional temperature rise inside the nine candidate fault areas of the present invention and the paired temperature rise of the corresponding outer wall of the enclosure. Figure 8 This is a diagram showing the location response of the outer wall of the enclosure for the nine candidate fault regions of this invention.

[0096] Figure reference numerals: 1: Three-phase oil-immersed transformer, 2: Oil tank, 3: Transformer oil, 4: Iron core, 5: Phase A winding, 6: Phase B winding, 7: Phase C winding, 8: Solid insulation structure, 9: Heat dissipation structure; 10: Candidate inter-turn short-circuit fault area; 11: Front side wall of oil tank, 12: Rear side wall of oil tank. Detailed Implementation

[0097] The following specific examples illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of the present invention. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0098] The accompanying drawings are for illustrative purposes only and are schematic diagrams, not actual pictures. They should not be construed as limiting the invention. To better illustrate the embodiments of the invention, some parts in the drawings may be omitted, enlarged, or reduced, and do not represent the actual product dimensions. It is understandable to those skilled in the art that some well-known structures and their descriptions may be omitted in the drawings.

[0099] In the accompanying drawings of the embodiments of the present invention, the same or similar reference numerals correspond to the same or similar components. In the description of the present invention, it should be understood that if terms such as "upper," "lower," "left," "right," "front," and "rear" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, they are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, the terms used to describe positional relationships in the drawings are only for illustrative purposes and should not be construed as limiting the present invention. For those skilled in the art, the specific meaning of the above terms can be understood according to the specific circumstances.

[0100] This invention addresses the challenges of small additional heat loss, insignificant changes in terminal voltage and current, and difficulty in directly and continuously measuring the internal winding temperature when a small number of turns in the winding of an oil-immersed transformer experience inter-turn short circuits. By establishing a mapping relationship between the internal candidate fault region and the dynamic temperature response of the outer wall of the transformer tank, the invention transforms the difficult-to-obtain local abnormal heat sources inside the winding into a spatial temperature response obtainable from multiple temperature measurement points on the outer wall of the tank. Furthermore, it eliminates the common temperature rise component caused by changes in load, ambient temperature, and cooling status through a healthy common temperature change model, and inverts the intensity of additional heat sources in each candidate fault region based on the remaining abnormal temperature residuals. This enables the identification and spatial location of minor inter-turn short-circuit thermal faults in the winding.

[0101] like Figure 1 As shown, the three-phase oil-immersed transformer 1 includes an oil tank 2, transformer oil 3 disposed inside the oil tank 2, an iron core 4, an A-phase winding 5, a B-phase winding 6, a C-phase winding 7, and a solid insulation structure 8, and is provided with a heat dissipation structure 9 for dissipating heat to the external environment.

[0102] Phase A winding 5, Phase B winding 6, and Phase C winding 7 are located in transformer oil 3. During normal operation of the transformer, winding copper loss, core iron loss, and other normal losses form heat sources. The generated heat is dissipated outward through the windings, solid insulation structure 8, transformer oil 3, and oil tank 2, and further exchanged with the external environment through oil tank 2 and heat dissipation structure 9.

[0103] When a localized inter-turn short circuit occurs in the winding, an additional circulating current is generated in the short-circuit loop, causing additional losses and forming a localized abnormal heat source in the area where the faulty turn is located. This localized abnormal heat, after passing through the insulation structure, the transformer oil flow, and the tank wall, forms a temperature change on the outer wall of the tank with spatial diffusion and time delay characteristics. Since different internal heat sources in different phases and at different axial positions have different heat transfer paths to their respective locations on the outer wall of the tank, different internal heat sources will create different spatial temperature distributions on the outer wall of the tank, thus providing a basis for inferring the location of internal heat sources using the temperature of the outer wall.

[0104] First, a three-dimensional fluid-heat transfer coupling model of the three-phase oil-immersed transformer to be tested is established.

[0105] When building the model, geometric structural parameters, material thermal property parameters, normal loss parameters, and operating status parameters are input based on the actual structure of the target transformer. Geometric structural parameters include the geometric dimensions and spatial positions of the core, windings, solid insulation structure, oil tank, oil channels, and heat dissipation structure; material thermal property parameters include the density, specific heat capacity, and thermal conductivity of the corresponding solid materials and transformer oil; normal loss parameters include at least winding copper losses and core iron losses; and operating status parameters include at least load, ambient temperature, and cooling status.

[0106] For the solid region formed by the core, windings, solid insulation structure, and outer wall of the enclosure, the transient heat transfer equation is expressed as:

[0107] in, Indicates the density of the solid region. This indicates the specific heat capacity of the solid region. Indicates the thermal conductivity of the solid region. Indicates the temperature of the solid region. This represents the volumetric heat source density of the solid region under normal operating conditions. N Indicates the number of candidate inter-turn short-circuit fault regions. Indicates the first j Additional heat generation power in each candidate inter-turn short-circuit fault region Indicates the first j Normalized spatial heat source distribution function for candidate inter-turn short-circuit fault regions Indicates spatial location.

[0108] For the transformer oil region, its flow and heat transfer equation is expressed as:

[0109] in, Indicates the density of transformer oil. This indicates the specific heat capacity of transformer oil. Indicates the thermal conductivity of transformer oil. Indicates the transformer oil temperature. Indicates the transformer oil flow velocity. This indicates the volumetric heat source density of the transformer oil region under normal operating conditions.

[0110] In order to Direct characterization of the first j The total additional heat generation power of each candidate inter-turn short-circuit fault region is normalized to the spatial heat source distribution function to satisfy:

[0111] in, This represents the j-th candidate inter-turn short-circuit fault region. Therefore, This represents the additional volumetric heat source density applied to the heat transfer control equation.

[0112] For transformer oil flow, a corresponding flow model can be established based on the actual operating mode. Under natural oil circulation, temperature-dependent density or Boussinesq can be used to approximate the buoyancy effect caused by temperature difference; for forced oil circulation with an oil pump, corresponding flow conditions can be set according to the oil pump's operating state and oil flow velocity.

[0113] At the interfaces where the windings, core, solid insulation structure, and inner wall of the tank come into contact with the transformer oil, continuous temperature and continuous normal heat flow conditions are set. For the contact interfaces between different solid components, continuous temperature and continuous heat flow conditions are used when the contact thermal resistance is negligible; when insulation layers, assembly gaps, or contact thermal resistance need to be considered, the interface heat transfer relationship is set according to the corresponding contact thermal resistance.

[0114] The outer wall of the tank is subject to convective heat transfer boundary conditions between itself and the environment. When the influence of external radiative heat transfer on the surface temperature of the tank is not negligible, a radiative heat transfer term can be further added. For transformers equipped with fans, oil pumps, or other auxiliary cooling equipment, different equivalent heat transfer coefficients, oil flow velocities, and corresponding boundary conditions are set according to the start-up and shutdown status of different cooling equipment.

[0115] At the initial moment of the simulation, the initial temperature of each solid region and transformer oil region can be set to the stable temperature field under the corresponding operating conditions; if stable temperature field data is lacking, it can also be set according to the actual shutdown temperature or ambient temperature.

[0116] like Figure 2 As shown, in order to realize the regional inversion of the fault location inside the winding, this embodiment divides the space of phase A winding 5, phase B winding 6 and phase C winding 7 respectively.

[0117] Specifically, each phase winding is divided into an upper region, a middle region, and a lower region along the winding axis. Among them, phase A winding corresponds to candidate regions R1, R2, and R3, phase B winding corresponds to candidate regions R4, R5, and R6, and phase C winding corresponds to candidate regions R7, R8, and R9, thus forming nine candidate inter-turn short-circuit fault regions 10.

[0118] R1, R4, and R7 correspond to the upper regions of the A-phase winding, B-phase winding, and C-phase winding, respectively; R2, R5, and R8 correspond to the middle regions of the A-phase winding, B-phase winding, and C-phase winding, respectively; and R3, R6, and R9 correspond to the lower regions of the A-phase winding, B-phase winding, and C-phase winding, respectively.

[0119] Using the above division method, at least the fault phase and the axial location of the fault can be located. When it is necessary to further improve the radial location resolution, the above axial candidate region can be further divided radially along the winding into an inner winding region, a middle winding region, and an outer winding region, thereby forming a more refined candidate fault region.

[0120] After dividing the candidate fault regions, forward simulations with additional heat sources are performed for different candidate regions.

[0121] First, a healthy operating state model is established. The additional heat sources of all candidate inter-turn short-circuit fault areas are set to zero. The internal temperature field and the outer wall temperature field of the enclosure under healthy conditions are obtained by solving the model based on normal copper loss, iron loss, ambient temperature and cooling state.

[0122] Subsequently, reference additional heat power is applied to each candidate fault region in R1 to R9. In each fault state simulation, only the additional heat source of the target candidate region is changed, while other geometric structures, material parameters, normal losses, and operating boundary conditions remain consistent with the corresponding healthy state, so that the main temperature difference between the fault state and the healthy state comes from the additional heat generation of the target candidate region.

[0123] To ensure that the established dynamic thermal response model can adapt to changes in the actual operating conditions of the transformer, the aforementioned forward simulation process was repeated under multiple representative load, ambient temperature, and cooling conditions. Operating state nodes were set for different load levels, ambient temperatures, and cooling methods to obtain the dynamic temperature response corresponding to different candidate fault regions under multiple operating conditions.

[0124] like Figure 4 As shown, the forward simulation and dynamic thermal response database establishment process of the present invention includes model input, establishment of a three-dimensional fluid-heat transfer coupling model, division of computational mesh, setting of physical fields and boundary conditions, health state simulation, candidate fault state simulation, calculation of external wall temperature change, normalization of unit power, and storage of dynamic thermal response data.

[0125] For any operating state and any candidate fault area, the outer wall temperature of the enclosure corresponding to the fault state is compared with the outer wall temperature of the enclosure corresponding to the healthy state to obtain the change in outer wall temperature.

[0126] The change in temperature of the outer wall of the chamber is then divided by the corresponding reference additional heat power to obtain the dynamic thermal response per unit power.

[0127] time t The corresponding dynamic thermal response matrix per unit power is expressed as:

[0128] in, Indicates time t The corresponding dynamic thermal response matrix per unit power, Indicates the first j The temperature response vector formed at different temperature measurement locations on the outer wall of the enclosure after applying a unit additional heat power to each candidate inter-turn short-circuit fault region. N Indicates the number of candidate inter-turn short-circuit fault regions. t This indicates the response time after the application of an additional heat source.

[0129] Under discrete-time conditions, the dynamic thermal response per unit power at different response times is stored to obtain the dynamic response characteristics corresponding to different heat propagation stages. Furthermore, an index is established according to different operating states, different candidate fault regions, and different response times to form a dynamic thermal response database.

[0130] The operating status includes at least load, ambient temperature, and cooling status; the candidate fault regions include R1, R2, ..., RN; and the response time includes... This establishes a quantitative mapping relationship between the location of internal candidate heat sources, the intensity of additional heat sources, and the dynamic temperature changes of the outer wall of the enclosure.

[0131] Multiple candidate temperature measurement positions are pre-set on the outer wall of the tank. The candidate temperature measurement positions cover the spatial projection areas of the A-phase winding, B-phase winding and C-phase winding on the outer wall of the tank, and respectively cover the upper, middle and lower parts of the corresponding spatial projection areas of each phase winding. At the same time, corresponding candidate temperature measurement positions are set on the front side wall 11 and the rear side wall 12 of the tank to form a set of candidate temperature measurement positions with spatial pairing on the front and rear walls.

[0132] For any candidate temperature measurement location, based on the dynamic thermal response database, the unit power dynamic temperature response formed at that candidate location after applying a unit additional heat power to each candidate inter-turn short-circuit fault region is extracted, and the differences between the unit power dynamic temperature responses corresponding to different candidate inter-turn short-circuit fault regions are compared. When the unit power dynamic temperature response formed at a certain candidate location for different candidate inter-turn short-circuit fault regions differs significantly, it indicates that the candidate location has a high ability to distinguish different internal heat source locations; conversely, when the unit power dynamic temperature response formed at a certain candidate location for different candidate inter-turn short-circuit fault regions is similar, it indicates that the candidate location has a low ability to distinguish different internal heat source locations.

[0133] Furthermore, when combining candidate temperature measurement locations, the selected locations should simultaneously meet the requirements of three-phase winding spatial coverage, axial coverage of the upper, middle, and lower parts of each phase winding, and the requirement of distinguishable dynamic thermal response between different candidate inter-turn short-circuit fault regions. For multiple candidate temperature measurement locations that meet the above requirements, priority is given to selecting candidate locations with larger differences in temperature response between different candidate inter-turn short-circuit fault regions, thus forming the final temperature sensor placement locations.

[0134] In this embodiment, according to the above screening principle, nine temperature measuring points F1 to F9 are determined on the front sidewall 11 of the oil tank, corresponding to the upper, middle, and lower parts of the A-phase winding, B-phase winding, and C-phase winding, respectively. Nine temperature measuring points B1 to B9 are determined on the rear sidewall 12 of the oil tank, corresponding to the spaces of F1 to F9. Specifically, F1, F2, and F3 correspond to the upper, middle, and lower parts of the A-phase winding, respectively; F4, F5, and F6 correspond to the upper, middle, and lower parts of the B-phase winding, respectively; and F7, F8, and F9 correspond to the upper, middle, and lower parts of the C-phase winding, respectively. B1 to B9 form front and rear wall spatial pairings with F1 to F9, resulting in a total of eighteen temperature measuring points on the outer wall of the tank.

[0135] It needs to be made particularly clear that, Figure 3 The measurement point locations shown represent a temperature sensor arrangement result obtained after screening according to the aforementioned dynamic thermal response differences. Each measurement point corresponds to the spatial projection position of the corresponding candidate fault area on the outer wall of the transformer tank. However, the determination of the measurement points is not solely based on the geometric center of the candidate fault area, but rather on the spatial coverage relationship of the candidate area and the temperature response resolution capability of the dynamic thermal response database. For oil-immersed transformers of different sizes, winding structures, tank structures, or heat dissipation structures, the final measurement point locations can be adjusted based on the corresponding dynamic thermal response database.

[0136] Among them, F1, F2 and F3 correspond to the upper, middle and lower parts of the A-phase winding, respectively; F4, F5 and F6 correspond to the upper, middle and lower parts of the B-phase winding, respectively; and F7, F8 and F9 correspond to the upper, middle and lower parts of the C-phase winding, respectively.

[0137] Correspondingly, B1 to B9 form front and rear wall spatial pairing relationships with F1 to F9 respectively, so that the upper, middle and lower parts of each phase winding have front wall measuring points and rear wall measuring points, forming a total of eighteen outer wall temperature measuring points of the box.

[0138] By setting spatially paired temperature measurement positions on opposite sides of the fuel tank, the impact of localized heat dissipation conditions on one side, localized environmental disturbances, and errors from individual sensors on fault location results can be reduced. Furthermore, utilizing the spatial temperature distribution formed by multiple measurement points preserves more spatial characteristics related to the fault phase and axial location compared to using only a single external wall temperature value.

[0139] In practical applications, the number and specific placement of temperature sensors are not limited to eighteen. They can be adjusted according to the transformer tank size, winding structure, radiator position, fault location resolution, and the distinguishability of responses in different candidate regions in the dynamic thermal response database.

[0140] During online operation, temperature data of the outer wall of the enclosure is continuously collected using various temperature sensors to form a temperature measurement vector at time t. .

[0141] During normal operation of an oil-immersed transformer, changes in load, ambient temperature, and the operating status of the cooling system can cause common or related temperature changes at multiple external wall measuring points. Therefore, the absolute temperature or single-point temperature rise at a certain location on the outer wall of the transformer cannot be directly used as the basis for locating minor inter-turn short-circuit thermal faults.

[0142] Therefore, this invention establishes a healthy common temperature change model based on the external wall temperature data of the transformer tank collected during healthy operation.

[0143] A healthy temperature sample set is formed by combining temperature measurement vectors corresponding to multiple sampling times under healthy operating conditions. The healthy average temperature vector is determined based on the healthy temperature sample set. Then, the average healthy temperature vector is subtracted from each healthy temperature sample to form a centralized healthy temperature sample.

[0144] In one implementation, the centralized healthy temperature sample can be decomposed using principal component analysis, singular value decomposition, or other methods capable of extracting common temperature change patterns from multiple measurement points. The main common temperature change basis vectors are selected based on a preset cumulative contribution rate or a preset number of basis vectors, and the selected common temperature change basis vectors form the healthy common temperature change basis matrix. .

[0145] Further configure the weight matrix The weight matrix is ​​used to reflect the measurement reliability, measurement noise level, or current data validity of different temperature measurement points on the outer wall of the enclosure. In one embodiment, the weight matrix... It can be a diagonal matrix, where the diagonal elements correspond to the weighting coefficients of each temperature measurement point.

[0146] by Represents the healthy average temperature vector, with This represents the basis matrix of public temperature changes obtained from health operation data, in which... This represents the weight matrix established based on the current interference situation, with... Indicates time t The collected temperature measurement vector of the outer wall of the enclosure. according to:

[0147] Obtain the projection coefficients of the current temperature data onto the subspace of public health temperature changes; where, Indicates time t The corresponding coefficient of variation in public health temperature; according to:

[0148] Obtain the weighted orthogonal projection matrix of the subspace of public health temperature changes, where, The weighted orthogonal projection matrix representing the subspace of public health temperature variation; Further evidence:

[0149] Obtain the weighted orthogonal projection complement matrix of the subspace of public health temperature changes, where, The weighted orthogonal projection complement matrix represents the subspace of public health temperature variation. It is an identity matrix.

[0150] according to:

[0151] Obtaining abnormal temperature residuals . Indicates time t The corresponding abnormal temperature residual; The abnormal temperature residual mainly retains local spatial anomalies in the current outer wall temperature of the enclosure that cannot be explained by the healthy common temperature change model, thereby reducing the interference of normal load changes, ambient temperature changes, and normal cooling state changes on the location of minor inter-turn short-circuit thermal faults.

[0152] To ensure that the online abnormal temperature residual and the offline established dynamic thermal response model are in the same feature space, it is also necessary to use the same weighted orthogonal projection complement matrix to perform a projection transformation on the dynamic thermal response matrix corresponding to the current operating state:

[0153] in, This represents the dynamic thermal response matrix obtained from the dynamic thermal response database based on the current operating status. This represents the equivalent dynamic thermal response matrix in the residual space of abnormal temperatures.

[0154] When determining the dynamic thermal response matrix corresponding to the current operating state, the corresponding data subset is first determined from the dynamic thermal response database based on the current cooling state of the transformer.

[0155] When the cooling state includes fan start-up and shutdown, oil pump start-up and shutdown, or different forced cooling levels, corresponding dynamic thermal response data subsets can be established separately to avoid interpolation between cooling states with large differences in physical properties.

[0156] After determining the data subset corresponding to the current cooling state, the appropriate operating state node is selected based on the current load and ambient temperature. When the current load and ambient temperature are consistent with existing nodes in the database, the corresponding dynamic thermal response matrix is ​​directly invoked; when the current load or ambient temperature is between adjacent operating state nodes, interpolation can be performed based on the dynamic thermal response matrices corresponding to adjacent operating state nodes to obtain the dynamic thermal response matrix corresponding to the current operating state.

[0157] To reduce the impact of temperature noise at a single sampling moment on the inversion results, in the case of... K Within a set time window of each sampling moment, the abnormal temperature residuals corresponding to multiple sampling moments are sequentially combined to form: Accordingly, the equivalent dynamic thermal response matrices corresponding to each sampling time are combined to form:

[0158] And combine the equivalent dynamic thermal response matrices corresponding to the respective sampling times to form:

[0159] Therefore, an inversion model for the internal additional heat source is established:

[0160] in, This represents the additional heat source intensity vector for each candidate inter-turn short-circuit fault region. This represents the residual term resulting from model error, temperature measurement error, and unmodeled disturbances.

[0161] Furthermore, when the actual start time of a suspected inter-turn short-circuit fault is unknown, multiple candidate fault start times can be set before the current analysis time window.

[0162] For each candidate fault start time, based on the time difference between each sampling time and the candidate fault start time, the dynamic thermal response matrix corresponding to the response time is extracted from the dynamic thermal response database to form the candidate time window dynamic response matrix; then, additional heat source inversion is performed, and the corresponding temperature fitting error or positioning confidence is calculated.

[0163] The candidate fault start time with smaller temperature fitting error or higher location reliability is determined as the estimated fault start time, and the inversion result corresponding to the estimated fault start time is used as the current fault location result, thereby taking into account the thermal inertia and response delay that exist during the propagation of local abnormal heat source to the outer wall of the box.

[0164] Further non-negative sparse inversion was performed on the internal additional heat source.

[0165] In one implementation, the intensity of the additional heat source in each candidate inter-turn short-circuit fault region is obtained by the following formula:

[0166] in, This represents the vector of inversion results for the additional heat source intensity; This represents the weight matrix corresponding to the set time window; This represents the sparse regularization parameter; Indicates the space smoothing regularization parameter; This represents the spatial adjacency difference matrix or the graph Laplace matrix.

[0167] Among them, the non-negativity constraint is used to ensure that the obtained additional heat source intensity meets the physical condition that the actual additional heating power is non-negative; the sparsity constraint is used to make the additional heat source preferentially concentrated in a smaller number of candidate fault regions; the spatial smoothness constraint is used to suppress drastic spatial jumps between adjacent regions that do not conform to the actual heat transfer law when the same actual fault heat source affects multiple adjacent candidate regions.

[0168] according to:

[0169] Identify the main suspected inter-turn short-circuit thermal fault areas. Among them, Indicates the main suspected fault area number, Indicates the first j The inversion of additional heat source intensity corresponding to each candidate fault region.

[0170] When multiple spatially adjacent candidate fault regions simultaneously have a large additional heat source intensity, the suspected fault range can be determined based on the spatial adjacency relationship between multiple candidate regions, rather than being limited to using a single candidate region as the sole location result.

[0171] To evaluate whether the internal additional heat source obtained from the inversion can reasonably explain the abnormal temperature residuals on the actual outer wall of the chamber, the temperature fitting error is further calculated:

[0172] in, This indicates the temperature fitting error. This represents a predetermined positive number used to prevent the denominator from being zero.

[0173] The smaller the temperature fitting error, the higher the consistency between the outer wall temperature response of the box calculated based on the inverted additional heat source and the measured abnormal temperature residual.

[0174] The spatial concentration of heat sources is further evaluated by inverting the spatial distribution of additional heat source intensity in each candidate inter-turn short-circuit fault region. When the additional heat sources are mainly concentrated in one or a few candidate regions, it indicates that the fault location has good spatial concentration; when the additional heat sources are dispersed in many candidate regions, it indicates that the uncertainty of the current location result is high.

[0175] The prominence of the main suspected fault area can also be evaluated based on the difference in the intensity of the additional heat source between the region with the largest additional heat source and other candidate regions. When the maximum additional heat source is significantly higher than that of other candidate regions, it indicates that the main suspected fault area has good regional identification; when the inverted additional heat source intensities of multiple candidate regions are close to each other, it indicates that there may be regional confusion.

[0176] The thermal response distinguishability is further evaluated based on the differences in temperature responses of the outer wall of the enclosure corresponding to different candidate fault regions in the dynamic thermal response database. The more significant the difference in temperature responses corresponding to different candidate fault regions, the more effective the current sensor arrangement and dynamic response matrix are in distinguishing the locations of different internal heat sources.

[0177] In one implementation, the temperature fitting error, the spatial concentration of additional heat sources, the prominence of the main suspected fault areas, and the distinguishability of the dynamic thermal response of the candidate areas can be normalized and comprehensively evaluated according to preset weights to obtain the reliability of fault location.

[0178] When the maximum inversion additional heat source intensity exceeds the preset heat source threshold, the fault location confidence exceeds the preset confidence threshold, and the same candidate inter-turn short-circuit fault area is identified as the main suspected fault area in multiple consecutive analysis time windows, the corresponding fault location result is output.

[0179] The fault location results may include the suspected fault phase, the suspected fault axial position, the candidate inter-turn short-circuit fault region number, the inverted additional heat source intensity, the temperature fitting error, and the location reliability. When the candidate region is further subdivided radially, the suspected fault radial position or the suspected fault spatial position can also be output.

[0180] The fault location results output by this invention are mainly used to indicate the location of internal local thermal anomalies suspected to be caused by minor inter-turn short circuits in the windings. The final fault type can be further confirmed by combining power outage maintenance and other transformer condition diagnosis methods.

[0181] To verify the feasibility of the above method, this embodiment establishes a health model and an inter-turn short-circuit fault model for a three-phase oil-immersed transformer with a rated voltage of 110 kV, a rated capacity of 31.5 MVA, and a rated frequency of 50 Hz.

[0182] The three-phase winding adopts a coaxial cylindrical structure, with a low-voltage cylindrical winding, an insulating cylinder, and a high-voltage cylindrical winding arranged sequentially from the inside to the outside of the cylindrical iron core. The high-voltage winding has an equivalent number of 900 turns, and the low-voltage winding has an equivalent number of 90 turns.

[0183] The model includes solid-fluid heat transfer, transformer oil flow, non-isothermal flow, and corresponding circuit calculation interfaces. During heat-fluid calculations, heat transfer and flow-related interfaces are enabled, while during fault electrical parameter calculations, corresponding circuit interfaces can be enabled to obtain loss parameters corresponding to healthy and inter-turn short-circuit states.

[0184] The health model and the failure model use the same geometry, material parameters and computational mesh. The ambient temperature is set to 20 ℃ and the equivalent heat transfer coefficient of the outer wall of the box is set to 55 W / (m²·K).

[0185] The healthy state model applies normal copper and iron losses; the fault state model adds an additional 0.540 kW heat source generated by a short circuit of 3 turns in the middle high-voltage cylindrical winding region of phase B.

[0186] For minor inter-turn short circuits in high-voltage windings, 1-5 turn short circuit conditions can be scanned, which account for approximately 0.11% to 0.56% of a 900-turn high-voltage winding. This embodiment selects a 3-turn short circuit as the main verification condition to verify the feasibility of fault location through the temperature of the outer wall of the enclosure when the number of short-circuit turns is small and the change in electrical quantity on the terminal side is small.

[0187] The additional heat loss of 0.540 kW is relatively small compared to the overall power level corresponding to the rated capacity of the transformer. Therefore, this embodiment does not use the terminal voltage or terminal current exceeding a predetermined threshold as a necessary condition for locating minor inter-turn short-circuit thermal faults.

[0188] Nine temperature measuring points are set on the front sidewall 11 and rear sidewall 12 of the oil tank, corresponding to the upper, middle and lower parts of phases A, B and C, respectively, forming a total of eighteen temperature measuring points on the outer wall of the tank. The healthy state temperature and the fault state temperature are output at the same spatial coordinates, and the additional temperature rise of the corresponding outer wall of the tank is obtained by subtracting the healthy state temperature from the fault state temperature.

[0189] The main simulation results obtained in this embodiment are shown in Table 1.

[0190] Table 1

[0191] like Figure 5As shown, after a 3-turn short circuit occurred in the high-voltage winding in the middle of phase B, all eighteen temperature measuring points on the outer wall of the enclosure showed varying degrees of non-negative additional temperature rise. Although the absolute temperature changes at each measuring point were relatively small, after benchmark compensation for the healthy state temperature, the additional temperature rise at different measuring points showed significant spatial differences.

[0192] Among them, the measuring point in the middle of phase B on the front wall of the oil tank causes the largest additional temperature rise on the outer wall of the tank, which is about 0.748956 ℃; the average additional temperature rise of the paired measuring points on the front and rear walls of phase B is about 0.748393 ℃, and their phase and axial position are consistent with the fault location in the middle of phase B in the model.

[0193] This shows that when judging solely from the absolute temperature of the outer wall of the enclosure, minor local thermal faults may be easily masked by the normal operating temperature level. However, by extracting the additional temperature rise through healthy temperature benchmark compensation, the temperature anomalies related to the local fault heat source can be made more obvious.

[0194] like Figure 6 As shown, in the distribution of additional temperature rise on the front and rear walls of the tank under the condition of a short circuit in the middle three turns of phase B relative to the healthy condition, the high temperature rise areas are mainly concentrated in the horizontal position corresponding to phase B and the middle height of the winding, and gradually decrease from the fault-corresponding area to the adjacent phases and the upper and lower axial regions.

[0195] The results indicate that although the localized additional heat source inside the winding diffuses spatially to the outside through the transformer oil and tank structure, the temperature distribution on the outer wall of the tank still retains spatial information related to the fault phase and axial location. Therefore, fault location can be achieved by measuring the spatial temperature distribution at multiple points instead of a single point.

[0196] like Figure 7 As shown, after applying corresponding additional heat sources to the nine candidate fault areas of the high-voltage winding, there is a significant positive correlation between the additional temperature rise inside each candidate area and the average additional temperature rise of the corresponding front and rear walls of the oil tank, with a Pearson correlation coefficient of approximately 0.956839.

[0197] This indicates that temperature changes within different candidate regions can be stably mapped to temperature changes on the outer wall of the enclosure, further verifying the feasibility of using the temperature response of the outer wall of the enclosure to invert the location of local heat sources inside.

[0198] Under the condition of a 3-turn short circuit in the middle of phase B, the internal temperature of the high-voltage winding in the middle of phase B increases from approximately 83.853 ℃ to approximately 93.177 ℃. According to the temperature values ​​shown above, the corresponding internal additional temperature rise is approximately 9.324 ℃. The low-voltage winding at the same location experiences a coupled additional temperature rise of approximately 5.565 ℃ due to radial heat transfer.

[0199] The above results also demonstrate that localized abnormal heat sources can propagate to surrounding areas through the winding structure, insulation materials, and transformer oil. Therefore, the actual temperature field is not strictly limited to the location directly opposite the faulty turn. Based on this, this invention uses a dynamic thermal response matrix composed of multiple candidate regions and multiple outer wall measuring points for overall inversion, rather than directly making a fault judgment based on a single outer wall measuring point directly opposite the faulty region.

[0200] like Figure 8 As shown, among the location responses of the outer wall of the enclosure corresponding to the nine candidate fault regions, the middle candidate fault region of phase B exhibits the highest location response, with an average additional temperature rise peak of approximately 0.75 ℃ for the front and rear walls. In contrast, the location responses corresponding to the upper and lower parts of phase B, as well as different candidate regions of phases A and C, are all lower than those of the middle part of phase B.

[0201] Therefore, within the two-dimensional candidate fault region space comprised of phase and axial height, the pre-set fault location in the middle of phase B can be identified based on the inversion of additional heat source intensity or location response corresponding to different regions. In the original embodiment, the location response corresponding to the middle region of phase B is the highest value among the nine candidate regions.

[0202] Depend on Figures 5-8 As can be seen from the results in Table 1, even though the additional heat loss caused by a small inter-turn short circuit is relatively small relative to the overall operating power of the transformer, the changes in terminal voltage or terminal current may still be within the normal operating fluctuation range. After removing the normal operating temperature component by the healthy common temperature change model, the abnormal temperature residual on the outer wall of the tank can still retain the spatial characteristics related to the location of the internal local heat source.

[0203] By combining the abnormal temperature residuals from multiple consecutive sampling times with the dynamic thermal response matrix corresponding to the current operating state, and further utilizing non-negative constraints, sparse constraints, and spatial smoothness constraints to perform internal additional heat source inversion, the intensity of additional heat sources corresponding to different candidate inter-turn short-circuit fault regions can be obtained.

[0204] By comprehensively judging the maximum additional heat source intensity, temperature fitting error, heat source spatial concentration, prominence of major suspected fault areas, thermal response distinguishability of different candidate areas, and consistency of multiple consecutive time windows, the risk of mislocation caused by accidental environmental disturbances, single sensor noise, and short-term abnormal temperature changes can be further reduced.

[0205] In this embodiment, the maximum abnormal temperature rise of the outer wall of the enclosure is approximately 0.749 ℃. Therefore, in practical engineering applications, the outer wall temperature sensor and its signal acquisition and compensation system preferably have an effective temperature resolution of 0.1 ℃ or higher to ensure that the residual temperature difference of the outer wall of the enclosure caused by a small inter-turn short circuit can be effectively acquired.

[0206] It should be noted that the parameters used in this embodiment, such as 110 kV rated voltage, 31.5 MVA rated capacity, 50Hz rated frequency, 900 turns of high-voltage winding, 90 turns of low-voltage winding, 3 turns of short circuit, 0.540 kW of additional heat power, 20 ℃ ambient temperature, and 55 W / (m²·K), are as follows: Figures 5-8 The simulation results shown are only used to verify the feasibility of the technical approach of the present invention and are not intended to limit the scope of protection of the present invention.

[0207] In practical engineering applications, the three-dimensional fluid-heat transfer coupling model should be calibrated based on the actual geometric structure, winding structure, insulation material, oil duct structure, copper loss, iron loss, heat dissipation structure, load range, ambient temperature range and cooling method of the target oil-immersed transformer. The corresponding healthy common temperature change model and fault judgment threshold should be established using the actual external wall temperature data of the tank collected during the healthy operation of the target transformer.

[0208] For oil-immersed transformers with different capacities, voltage levels, winding forms, and cooling methods, the number of candidate fault regions, the number of operating status nodes in the dynamic thermal response database, the number of response time nodes, and the number of temperature sensors on the outer wall of the tank can be changed accordingly. As long as the technical approach of positive response modeling of internal candidate heat sources, extraction of abnormal temperature residuals on the outer wall of the tank, and inversion of internal additional heat sources is still adopted, the method of this invention can be implemented.

[0209] Through the above implementation methods, the present invention can transform the local thermal anomalies of small inter-turn short circuits inside the winding, which are difficult to measure directly, into dynamic temperature spatial distributions corresponding to multiple temperature measurement locations on the outer wall of the transformer tank. In the offline stage, it completes the forward simulation of candidate heat sources and the construction of a dynamic thermal response database. In the online stage, it completes the acquisition of temperature on the outer wall of the transformer tank, elimination of healthy common temperature components, matching of current operating status, and inversion of internal additional heat sources. Thus, it can realize the online identification and spatial location of early small inter-turn short circuit thermal faults in oil-immersed transformers without the need for a large number of internal temperature sensors.

[0210] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for locating inter-turn short-circuit thermal faults in oil-immersed transformer windings based on the temperature distribution characteristics of the outer wall of the transformer housing, characterized in that, Includes the following steps: S1. Establish a three-dimensional fluid-heat coupling model of the three-phase oil-immersed transformer to be tested. The three-dimensional fluid-heat coupling model includes the core, A-phase winding, B-phase winding, C-phase winding, solid insulation structure, transformer oil, outer wall of the tank, and heat dissipation structure. S2, the A-phase winding, B-phase winding and C-phase winding are divided into spatial regions to form multiple candidate inter-turn short-circuit fault regions. Reference additional heat sources are applied to each candidate inter-turn short-circuit fault region under multiple operating states. The forward temperature field simulation is performed through the three-dimensional fluid-heat transfer coupling model to obtain the internal temperature field and outer wall temperature response of each candidate inter-turn short-circuit fault region. S3. Based on the temperature change of the outer wall of the box before and after applying the reference additional heat source and the corresponding reference additional heat power, obtain the unit power dynamic thermal response of each candidate inter-turn short-circuit fault area, and store it according to the operating status, candidate inter-turn short-circuit fault area and response time to establish a dynamic thermal response database that characterizes the mapping relationship between candidate inter-turn short-circuit fault area, additional heat source intensity and outer wall temperature response of the box. S4. Multiple candidate temperature sensor positions are set on the outer wall of the transformer tank. The temperature sensor placement positions are determined based on the temperature response differences of the outer wall of the transformer tank corresponding to different candidate inter-turn short-circuit fault areas in the dynamic thermal response database. The temperature sensors placed at the temperature sensor placement positions are used to continuously collect the outer wall temperature data of the transformer tank during the transformer operation process. S5. Based on the temperature data of the outer wall of the transformer tank collected under the healthy operating condition, a healthy common temperature change model is established. The temperature deviation of the current outer wall temperature of the transformer tank relative to the healthy average temperature is projected onto the weighted orthogonal complement space of the healthy common temperature change subspace to obtain the abnormal temperature residual. The dynamic thermal response matrix under the current operating condition is projected and transformed using the same weighted orthogonal projection complement matrix as the abnormal temperature residual to obtain the equivalent dynamic thermal response matrix under the abnormal temperature residual space. S6. Combine the abnormal temperature residuals of multiple sampling times within a set time window, and combine the equivalent dynamic thermal response matrix at the corresponding time to establish an internal additional heat source inversion model with the additional heat source intensity of each candidate inter-turn short-circuit fault region as the unknown quantity. S7. Apply non-negative constraints, sparsity constraints, and spatial smoothness constraints to the internal additional heat source inversion model, solve for the additional heat source intensity corresponding to each candidate inter-turn short-circuit fault region, determine the candidate inter-turn short-circuit fault region that meets the fault judgment conditions and has the largest additional heat source intensity as the main suspected inter-turn short-circuit thermal fault region, and output the fault location result based on the inversion result.

2. The method for locating inter-turn short-circuit thermal faults in oil-immersed transformer windings based on the temperature distribution characteristics of the outer wall of the transformer casing, as described in claim 1, is characterized in that... In S2, the A-phase winding, B-phase winding, and C-phase winding are respectively divided into an upper region, a middle region, and a lower region along the winding axis to form candidate inter-turn short-circuit fault regions corresponding to different axial positions of the three-phase windings. When it is necessary to improve the radial positioning resolution, at least one of the candidate inter-turn short-circuit fault regions is further divided into an inner winding region, a middle winding region, and an outer winding region along the winding radial direction.

3. The method for locating inter-turn short-circuit thermal faults in oil-immersed transformer windings based on the temperature distribution characteristics of the outer wall of the transformer casing, as described in claim 1, is characterized in that... In S1, for the solid region consisting of the core, windings, solid insulation structure, and outer wall of the enclosure, the transient heat transfer equation is: For the transformer oil region, the flow and heat transfer equation is: in, , and These represent the density, specific heat capacity, and thermal conductivity of the solid region, respectively. , and These are the density, specific heat capacity, and thermal conductivity of transformer oil, respectively. Temperature of the solid region The temperature of the transformer oil. The transformer oil flow velocity, and These are the volumetric heat source densities of the solid region and the transformer oil region under normal operating conditions, respectively. N The number of candidate inter-turn short-circuit fault regions. For the first j Additional heat generation power in each candidate inter-turn short-circuit fault region For the first j The normalized spatial heat source distribution function of each candidate inter-turn short-circuit fault region satisfies: in, Indicates spatial location, For the first j The three-dimensional fluid-heat coupling model further sets the initial temperature conditions, the temperature continuity and normal heat flow continuity conditions between the solid region and the transformer oil, the contact heat transfer conditions between solid components, and the heat transfer boundary conditions between the outer wall of the tank and the environment.

4. The method for locating inter-turn short-circuit thermal faults in oil-immersed transformer windings based on the temperature distribution characteristics of the outer wall of the transformer casing, as described in claim 1, is characterized in that... In S3, under multiple different loads, ambient temperatures, and cooling conditions, the additional heat source for each candidate inter-turn short-circuit fault region is first set to zero, and the temperature response of the outer wall of the enclosure under healthy conditions is calculated. Then, a reference additional heat power is applied to each candidate inter-turn short-circuit fault region. The change in the outer wall temperature of the enclosure before and after applying the reference additional heat source is divided by the reference additional heat power to obtain the unit power dynamic thermal response of the corresponding candidate inter-turn short-circuit fault region, and then according to: Constructing Time t The corresponding dynamic thermal response matrix per unit power, where, For the first j The unit power temperature response vector of the outer wall of the enclosure corresponding to each candidate inter-turn short-circuit fault region is stored, and the unit power dynamic thermal response corresponding to different operating states, different candidate inter-turn short-circuit fault regions and different response times is stored to form the dynamic thermal response database.

5. The method for locating inter-turn short-circuit thermal faults in oil-immersed transformer windings based on the temperature distribution characteristics of the outer wall of the transformer casing, as described in claim 2, is characterized in that... In S4, multiple candidate temperature sensor positions are set on the outer wall of the tank. These candidate temperature sensor positions cover the spatial projection areas of phase A, phase B, and phase C windings on the outer wall of the tank, and respectively cover the upper, middle, and lower parts of the corresponding spatial projection area of ​​each phase winding. At the same time, corresponding paired candidate temperature measurement positions are set on the opposite outer walls of the tank. Based on the dynamic thermal response database, the unit power dynamic temperature response of different candidate inter-turn short-circuit fault areas at each candidate temperature sensor position is obtained. The thermal response resolution of each candidate temperature sensor position is determined based on the difference between the unit power dynamic temperature responses of different candidate inter-turn short-circuit fault areas, and the temperature sensor arrangement position is determined from the candidate temperature sensor positions.

6. The method for locating inter-turn short-circuit thermal faults in oil-immersed transformer windings based on the temperature distribution characteristics of the outer wall of the transformer casing, as described in claim 1, is characterized in that... In S5, with Represents the healthy average temperature vector, with This represents the basis matrix of public temperature changes obtained from health operation data, in which... This represents the weight matrix established based on the current interference situation, with... Indicates time t The collected temperature measurement vector of the outer wall of the enclosure is based on: Obtain the current public health temperature change coefficient, where, Indicates time t The corresponding coefficient of variation in public health temperature; according to: Obtain the weighted orthogonal projection matrix of the subspace of public health temperature changes, where, The weighted orthogonal projection matrix representing the subspace of public health temperature variation; according to: Obtain the weighted orthogonal projection complement matrix of the subspace of public health temperature changes; where, The weighted orthogonal projection complement matrix represents the subspace of public health temperature variation. It is the identity matrix; according to: Obtain the abnormal temperature residual. Indicates time t The corresponding abnormal temperature residual; And according to: The equivalent dynamic thermal response matrix under the abnormal temperature residual space is obtained, where, The dynamic thermal response matrix is ​​obtained from the dynamic thermal response database based on the current operating status. This represents the equivalent dynamic thermal response matrix in the residual space of abnormal temperatures.

7. The method for locating inter-turn short-circuit thermal faults in oil-immersed transformer windings based on the temperature distribution characteristics of the outer wall of the transformer casing, as described in claim 6, is characterized in that... In step S6, a subset of operating state data is determined from the dynamic thermal response database based on the current cooling state of the transformer. Then, the dynamic thermal response matrix of the corresponding operating state node is selected based on the current load and ambient temperature, or interpolation is performed between adjacent operating state nodes based on the current load and ambient temperature to obtain the dynamic thermal response matrix corresponding to the current operating state. The dynamic thermal response matrix is ​​then transformed using the weighted orthogonal projection complement matrix to obtain an equivalent dynamic thermal response matrix. Finally, the abnormal temperature residuals from multiple sampling moments are combined within a set time window to form the time window abnormal temperature residual. The equivalent dynamic thermal response matrices at corresponding times are combined to form the equivalent dynamic response matrix of the time window. ,Establish: in, This represents the additional heat source intensity vector for each candidate inter-turn short-circuit fault region. This is the residual term composed of model error and measurement error.

8. The method for locating inter-turn short-circuit thermal faults in oil-immersed transformer windings based on the temperature distribution characteristics of the outer wall of the transformer casing, as described in claim 7, is characterized in that... In S7, the intensity of the additional heat source in each candidate inter-turn short-circuit fault region is non-negatively sparsely inverted according to the following formula: in, To add the heat source intensity inversion result vector, To set the weight matrix corresponding to the time window, For sparse regularization parameters, For spatial smoothing regularization parameters, It is a spatial adjacency difference matrix or a graph Laplace matrix.

9. The method for locating inter-turn short-circuit thermal faults in oil-immersed transformer windings based on the temperature distribution characteristics of the outer wall of the transformer casing, as described in claim 8, is characterized in that... according to: The main suspected inter-turn short-circuit thermal fault areas were identified, among which, This is the area number for the main suspected inter-turn short-circuit thermal fault area. For the first j The additional heat source intensity of each candidate inter-turn short-circuit fault region is inverted; when multiple spatially adjacent candidate inter-turn short-circuit fault regions simultaneously have additional heat source intensities higher than a preset value, the range of suspected inter-turn short-circuit thermal faults is determined based on the spatial adjacency relationship between the multiple candidate inter-turn short-circuit fault regions.

10. The method for locating inter-turn short-circuit thermal faults in oil-immersed transformer windings based on the temperature distribution characteristics of the outer wall of the transformer casing, as described in claim 9, is characterized in that... The temperature fitting error is calculated using the following formula: in, This is the temperature fitting error. A predetermined positive number set to prevent the denominator from being zero; The reliability of the location is determined based on the temperature fitting error, the spatial concentration of additional heat sources in each candidate inter-turn short-circuit fault area, the prominence of the main suspected inter-turn short-circuit thermal fault area relative to other candidate inter-turn short-circuit fault areas, and the distinguishability of the temperature response of the outer wall of the enclosure corresponding to different candidate inter-turn short-circuit fault areas. When the maximum inversion additional heat source intensity exceeds the preset heat source threshold, the location confidence exceeds the preset confidence threshold, and the same candidate inter-turn short-circuit fault area is identified as the main suspected inter-turn short-circuit thermal fault area in multiple consecutive time windows, the suspected fault phase, suspected fault axial position, candidate inter-turn short-circuit fault area number, inversion additional heat source intensity, temperature fitting error, and location confidence are output.

Images