An overheating diagnosis method and device for a transformer winding

Abstract

The present application discloses an overheat diagnosis method and device for a transformer winding. The method includes: obtaining temperature data monitored at a sensing monitoring point on the transformer winding, inputting it into a support vector machine model, and outputting sliced temperature data to obtain a multi-slice scan map. If the temperature value during longitudinal slice scanning of the multi-slice scan map exceeds the first ratio of the thermal threshold, it is determined whether the temperature value during axial slice scanning of the multi-slice scan map exceeds the second ratio of the thermal threshold and the electric field value exceeds the electric field threshold ratio. If so, it is determined that the transformer winding is overheated and a warning is issued; otherwise, it is determined that the transformer winding is normal. It can be seen that before evaluating the multi-physical field results, sliced temperature data is output through the support vector machine model. The support vector machine model simulates the correspondence between fault conditions and multi-slice data, and can achieve rapid diagnosis of the multi-physical field distribution inside the transformer, so as to accurately, comprehensively and overall evaluate the overheat state of the transformer winding.

Description

Technical Field

[0001] The present application relates to the field of power technologies, and more specifically, to a method and device for diagnosing overheating of a transformer winding. Background Art

[0002] With the development of digital technologies, the operation and maintenance of power transmission and transformation equipment have gradually changed from regular test and maintenance to condition monitoring. Visualization of internal characteristics of equipment combined with multi-physical field simulation is one of the key technologies for condition monitoring. During the operation of the equipment, this technology combines sensor data with multi-physical field simulation to obtain temperature field distribution, stress field distribution, electric field distribution, etc., so as to evaluate the equipment.

[0003] However, for large AC transformers / DC converter transformers, although current large model processing and simulation can obtain overall multi-field calculation results, their internal structures are complex, the number of components is huge, and the number of parts after assembly may be as high as thousands. Moreover, they are large in size. It is difficult to quickly and reliably carry out evaluation by using the conventional post-processing analysis of multi-physical field results for each component or simply intercepting cross-sectional data. At the same time, due to the large number of calculation grid nodes in the transformer, which can be up to hundreds of millions, if a deep learning-based node interpolation fitting method is directly used to carry out rapid calculation and output of the full model, it will cause large errors in data points inside the model.

[0004] Therefore, how to develop a method for real-time output of multi-physical field distribution of a transformer and capable of overall, accurate and comprehensive evaluation of the overheating state is of great significance for digital operation and maintenance of transformers and state deduction and early warning. Summary of the Invention

[0005] In view of the above problems, the present application provides a method and device for diagnosing overheating of a transformer winding to accurately, comprehensively and overall evaluate the overheating state of the transformer winding.

[0006] To achieve the above object, the following specific solutions are proposed:

[0007] A method for diagnosing overheating of a transformer winding, comprising:

[0008] Obtaining temperature data monitored by a sensing monitoring point on the transformer winding;

[0009] Inputting the temperature data into a pre-constructed support vector machine model, and outputting sliced temperature data;

[0010] Based on the sliced temperature data, constructing a multi-slice scan map;

[0011] If the temperature value during the longitudinal slice scanning of the multi-slice scan image exceeds the first preset ratio of the preset thermal threshold, then determine whether the temperature value during the axial slice scanning of the multi-slice scan image exceeds the second preset ratio of the preset thermal threshold, and whether the electric field value during the axial slice scanning exceeds the preset ratio of the preset electric field threshold;

[0012] If so, determine that the transformer winding is in an overheated state and send a warning prompt to the central system;

[0013] If not, determine that the transformer winding is in a normal state.

[0014] Optionally, after inputting the temperature data into a pre-constructed support vector machine model and outputting the slice temperature data, it further includes:

[0015] Judge whether the slice temperature data is outside the characteristic temperature range of the temperature data and outside the target temperature range, where the lower limit value of the target temperature range is the monitoring point temperature of the transformer winding in the normal fault state, and the upper limit value of the target temperature range is the monitoring point temperature of the transformer winding in the most serious fault state;

[0016] If so, prompt that no feature is recognized.

[0017] Optionally, the construction process of the support vector machine model includes:

[0018] Obtain multiple temperature-carrying capacity distribution data of the transformer winding in the normal state and multiple temperature-carrying capacity distribution data of the transformer winding in the abnormal state;

[0019] Under each temperature-carrying capacity distribution data, perform multiple slice scans on the transformer winding under the temperature-carrying capacity distribution data to obtain multiple slices, and determine the temperature field distribution and electro-thermal coupling field distribution inside the transformer winding on each slice;

[0020] For each temperature-carrying capacity distribution data, use the temperature field distribution in the temperature-carrying capacity distribution data as the initial value for electric field calculation, determine the first change relationship of the insulating oil of the transformer winding with temperature change, and determine the second change relationship of the resistivity of the insulating paper of the transformer winding with temperature change;

[0021] Under the preset attention threshold, based on the temperature field distribution and electro-thermal coupling field distribution inside the transformer winding on each slice, determine the attention area in each slice;

[0022] Based on the attention areas in each slice, determine the simulation calculation results corresponding to the positions of the sensing monitoring points of the transformer winding;

[0023] Construct a support vector machine model based on each slice, the simulation calculation results, each first variation relationship, and each second variation relationship.

[0024] Optionally, performing multiple slice scans on the transformer winding to obtain multiple slices, including:

[0025] Performing multiple axial slice scans on the transformer winding to obtain multiple axial slices;

[0026] Performing multiple longitudinal slice scans on the transformer winding to obtain multiple longitudinal slices;

[0027] Combining each axial slice and each longitudinal slice to obtain multiple slices.

[0028] Optionally, performing multiple longitudinal slice scans on the transformer winding to obtain multiple longitudinal slices, including:

[0029] Taking the middle of the primary winding coil of the transformer winding and the middle of the secondary winding coil of the transformer winding as the scanning direction, performing multiple longitudinal close-up slice scans on the transformer winding to obtain several first-type longitudinal slices;

[0030] Taking the middle of the oil gap of the primary winding of the transformer winding and the middle of the oil gap of the secondary winding of the transformer winding as the scanning direction, performing multiple longitudinal close-up slice scans on the transformer winding to obtain several second-type longitudinal slices;

[0031] Combining each first-type longitudinal slice and each second-type longitudinal slice to obtain multiple longitudinal slices.

[0032] Optionally, performing multiple axial slice scans on the transformer winding to obtain multiple axial slices, including:

[0033] Determine the coil fault-prone points and temperature hot spots in the transformer winding according to the historical fault data of the transformer winding and the temperature field distribution;

[0034] Taking the axis of symmetry of the transformer winding as the rotation axis, taking the preset angle range of the coil fault-prone points and the temperature hot spots as the scanning direction, and performing axial slicing on the transformer winding at a step slice angle not exceeding the first preset step angle within the preset angle range to obtain several first-type axial slices;

[0035] Taking the axis of symmetry of the transformer winding as the rotation axis, taking the other range of the coil fault-prone points and the preset angle range as the scanning direction, and performing axial slicing on the transformer winding at a step slice angle not exceeding the second preset step angle within the other range of the preset angle range to obtain several second-type axial slices;

[0036] Combine each first-type axial slice and each second-type axial slice to obtain a plurality of axial slices.

[0037] Optionally, the plurality of axial slices include axial slices that cut through the sensing monitoring points of the transformer winding, and the plurality of longitudinal slices include longitudinal slices that cut through the sensing monitoring points of the transformer winding.

[0038] Optionally, determining the simulation calculation results corresponding to the positions of the sensing monitoring points of the transformer winding based on the regions of interest in each slice includes:

[0039] Determine the temperature field calculation results corresponding to the positions of the sensing monitoring points of the transformer winding according to the regions of interest in each longitudinal slice;

[0040] Determine the temperature field calculation results corresponding to the positions of the sensing monitoring points of the transformer winding, and the electric field calculation results corresponding to the positions of the sensing monitoring points of the transformer winding according to the regions of interest in each axial slice;

[0041] Combine each temperature field calculation result and each electric field calculation result to obtain the simulation calculation results corresponding to the positions of the sensing monitoring points of the transformer winding.

[0042] Optionally, constructing a support vector machine model based on each slice, the simulation calculation results, each first change relationship, and each second change relationship includes:

[0043] Establish a regression function based on each slice and the simulation calculation results, and the regression function is:

[0044] y = W·x + b

[0045] Where,

[0046] Where, y is a vector composed of the actual temperatures of each slice, m is the number of slices, x is a vector composed of the temperature field calculation results of each slice that cuts through the sensing monitoring points in the simulation calculation results, n is the number of slices that cut through the sensing monitoring points in each slice, W is a regression weight coefficient matrix, and b is a bias coefficient vector;

[0047] Construct an insensitive loss function, and the insensitive loss function is:

[0048]

[0049] Where, ε is a preset insensitive loss coefficient, and g(x) is the predicted value obtained for each value of the vector x on its first change relationship and the predicted value obtained on its second change relationship;

[0050] Based on the insensitive loss function, a penalty factor is determined. Under the penalty factor, an optimization problem of the support vector machine is determined, and the optimization objective of the optimization problem of the support vector machine is:

[0051]

[0052] s.t.(W·x i )-y i +b≤ε+ξ ij

[0053]

[0054] ξ (*) ≥0

[0055] where ξ (*) is a preset slack variable, C is the penalty factor, and l is the dimension of the preset slack variable;

[0056] Solve the dual problem corresponding to the optimization problem of the support vector machine to obtain the solution of the dual problem. The dual problem is:

[0057]

[0058]

[0059]

[0060] where the solution of the dual problem is:

[0061]

[0062] According to the solution of the dual problem, use the following formula to determine the weight coefficient of the support vector machine regression model and the bias coefficient of the support vector machine regression model:

[0063]

[0064] where is the weight coefficient of the support vector machine regression model, is the bias coefficient of the support vector machine regression model;

[0065] Under the weight coefficient of the support vector machine regression model and the bias coefficient of the support vector machine regression model, a support vector machine model is constructed. The support vector machine model is:

[0066]

[0067] Wherein, Y is the output value of the support vector machine model, X is the input value of the support vector machine model, and K(·) is the model coefficient function of the support vector machine model.

[0068] An overheat diagnosis device for a transformer winding, comprising:

[0069] A temperature data acquisition unit, configured to acquire temperature data monitored by a sensing monitoring point on the transformer winding;

[0070] A sliced temperature data output unit, configured to input the temperature data into a pre-constructed support vector machine model and output sliced temperature data;

[0071] A multi-slice scan graph construction unit, configured to construct a multi-slice scan graph based on the sliced temperature data;

[0072] A threshold judgment unit, configured to, if the temperature value during longitudinal slice scanning of the multi-slice scan graph exceeds a first preset ratio of a preset thermal threshold, determine whether the temperature value during axial slice scanning of the multi-slice scan graph exceeds a second preset ratio of the preset thermal threshold, and whether the electric field value during axial slice scanning exceeds a preset ratio of a preset electric field threshold. If so, execute the overheat state determination unit; if not, execute the normal state determination unit;

[0073] The overheat state determination unit is configured to determine that the transformer winding is in an overheat state and send a warning prompt to the central system;

[0074] The normal state determination unit is configured to determine that the transformer winding is in a normal state.

[0075] Optionally, the device further comprises:

[0076] A temperature range judgment unit, configured to, after inputting the temperature data into a pre-constructed support vector machine model and outputting sliced temperature data, determine whether the sliced temperature data is outside the characteristic temperature range of the temperature data and outside a target temperature range, where the lower limit value of the target temperature range is the monitoring point temperature of the transformer winding in a normal fault state, and the upper limit value of the target temperature range is the monitoring point temperature of the transformer winding in the most severe fault state. If so, execute the unrecognized prompt unit;

[0077] The unrecognized prompt unit is configured to prompt that no feature is recognized.

[0078] Optionally, the device further comprises:

[0079] A distribution data acquisition unit, configured to acquire a plurality of temperature-carrying capacity distribution data of the transformer winding in a normal state and a plurality of temperature-carrying capacity distribution data of the transformer winding in an abnormal state;

[0080] A slice scanning unit for performing multiple slice scans on the transformer winding under each temperature-carrying capacity distribution data to obtain multiple slices;

[0081] A field distribution determination unit for determining the temperature field distribution and electrothermal coupling field distribution inside the transformer winding on each slice;

[0082] A variation relationship determination unit for, for each temperature-carrying capacity distribution data, using the temperature field distribution in the temperature-carrying capacity distribution data as the initial value for electric field calculation, determining a first variation relationship of the insulating oil of the transformer winding with temperature change, and determining a second variation relationship of the resistivity of the insulating paper of the transformer winding with temperature change;

[0083] A concerned area determination unit for determining the concerned areas in each slice based on the temperature field distribution and electrothermal coupling field distribution inside the transformer winding on each slice under a preset concern threshold;

[0084] A simulation calculation result determination unit for determining the simulation calculation results at the positions corresponding to the sensing monitoring points of the transformer winding based on the concerned areas in each slice;

[0085] A support vector machine model construction unit for constructing a support vector machine model based on each slice, the simulation calculation results, each first variation relationship, and each second variation relationship.

[0086] Optionally, the slice scanning unit includes:

[0087] An axial slice scanning unit for performing multiple axial slice scans on the transformer winding to obtain multiple axial slices;

[0088] A longitudinal slice scanning unit for performing multiple longitudinal slice scans on the transformer winding to obtain multiple longitudinal slices;

[0089] A slice combination unit for combining each axial slice and each longitudinal slice to obtain multiple slices.

[0090] Optionally, the longitudinal slice scanning unit includes:

[0091] A first longitudinal slice scanning subunit for performing multiple longitudinal close-up slice scans on the transformer winding in the scanning direction of the middle of the primary winding coil of the transformer winding and the middle of the secondary winding coil of the transformer winding to obtain several first-type longitudinal slices;

[0092] The second longitudinal slicing scanning subunit is configured to perform multiple longitudinal cutting and slicing scans on the transformer winding in a scanning direction that is the midpoint of the oil gap of the primary winding of the transformer winding and the midpoint of the oil gap of the secondary winding of the transformer winding, so as to obtain a plurality of second-type longitudinal slices;

[0093] The third longitudinal slicing scanning subunit is configured to combine each first-type longitudinal slice and each second-type longitudinal slice to obtain a plurality of longitudinal slices.

[0094] Optionally, the axial slicing scanning unit includes:

[0095] The first axial slicing scanning subunit is configured to determine the coil fault-prone points and temperature hot spots in the transformer winding according to the historical fault data of the transformer winding and the temperature field distribution;

[0096] The second axial slicing scanning subunit is configured to perform axial slicing on the transformer winding with the axis of symmetry of the transformer winding as the rotation axis, with the preset angular range of the coil fault-prone points and the temperature hot spots as the scanning direction, and with a step slicing angle not exceeding a first preset step angle within the preset angular range, so as to obtain a plurality of first-type axial slices;

[0097] The third axial slicing scanning subunit is configured to perform axial slicing on the transformer winding with the axis of symmetry of the transformer winding as the rotation axis, with the other range of the coil fault-prone points and the preset angular range as the scanning direction, and with a step slicing angle not exceeding a second preset step angle within the other range of the preset angular range, so as to obtain a plurality of second-type axial slices;

[0098] The fourth axial slicing scanning subunit is configured to combine each first-type axial slice and each second-type axial slice to obtain a plurality of axial slices.

[0099] Optionally, the plurality of axial slices include axial slices that cut through the sensing monitoring points of the transformer winding, and the plurality of longitudinal slices include longitudinal slices that cut through the sensing monitoring points of the transformer winding.

[0100] Optionally, the simulation calculation result determination unit includes:

[0101] The temperature field calculation result determination subunit is configured to determine the temperature field calculation result at the position corresponding to the sensing monitoring point of the transformer winding according to the regions of interest in each longitudinal slice;

[0102] The electric field calculation result determination subunit is configured to determine the temperature field calculation result at the position corresponding to the sensing monitoring point of the transformer winding, and the electric field calculation result at the position corresponding to the sensing monitoring point of the transformer winding according to the regions of interest in each axial slice;

[0103] A calculation result combination unit for combining the calculation results of each temperature field and the calculation results of each electric field to obtain the simulation calculation result corresponding to the position of the sensing monitoring point of the transformer winding.

[0104] Optionally, the support vector machine model construction unit includes:

[0105] A regression function establishment unit for establishing a regression function based on each slice and the simulation calculation result, and the regression function is:

[0106] y = W·x + b

[0107] Where,

[0108] Where, y is a vector composed of the actual temperatures of each slice, m is the number of slices, x is a vector composed of the temperature field calculation results of each slice that cuts through the sensing monitoring point in the simulation calculation result, n is the number of slices that cut through the sensing monitoring point in each slice, W is a regression weight coefficient matrix, and b is a bias coefficient vector;

[0109] An insensitive loss function construction unit for constructing an insensitive loss function, and the insensitive loss function is:

[0110]

[0111] Where, ε is a preset insensitive loss coefficient, g(x) is the predicted value obtained for each value of vector x on its first variation relationship and the predicted value obtained on its second variation relationship;

[0112] An optimization problem determination unit for determining a penalty factor based on the insensitive loss function, and determining a support vector machine optimization problem under the penalty factor, and the optimization objective of the support vector machine optimization problem is:

[0113]

[0114] s.t. (W·x i ) - y i + b ≤ ε + ξ ij

[0115]

[0116] ξ (*) ≥ 0

[0117] Where, ξ (*) is a preset slack variable, C is the penalty factor, and l is the dimension of the preset slack variable;

[0118] A solution unit for solving the dual problem corresponding to the support vector machine optimization problem to obtain a solution to the dual problem, where the dual problem is:

[0119]

[0120]

[0121]

[0122] where the solution to the dual problem is:

[0123]

[0124] A coefficient determination unit for determining the weight coefficient of the support vector machine regression model and the bias coefficient of the support vector machine regression model according to the solution to the dual problem by using the following formula:

[0125]

[0126] where is the weight coefficient of the support vector machine regression model, is the bias coefficient of the support vector machine regression model;

[0127] A model construction unit for constructing a support vector machine model under the weight coefficient of the support vector machine regression model and the bias coefficient of the support vector machine regression model, where the support vector machine model is:

[0128]

[0129] where Y is the output value of the support vector machine model, X is the input value of the support vector machine model, and K() is the model coefficient function of the support vector machine model.

[0130] With the above technical solution, the present application obtains the temperature data monitored by the sensing monitoring points on the transformer winding, inputs the temperature data into a pre-constructed support vector machine model, and outputs the sliced temperature data. Based on the sliced temperature data, a multi-slice scan map is constructed. If the temperature value during the longitudinal slice scan of the multi-slice scan map exceeds a first preset ratio of the preset thermal threshold, it is determined whether the temperature value during the axial slice scan of the multi-slice scan map exceeds a second preset ratio of the preset thermal threshold, and the electric field value during the axial slice scan exceeds a preset ratio of the preset electric field threshold. If so, it is determined that the transformer winding is in an overheated state, and a warning prompt is sent to the central system. If not, it is determined that the transformer winding is in a normal state. Thus, before analyzing and evaluating the multi-physical field results, the sliced temperature data is output through the support vector machine model. The support vector machine model simulates the correspondence between the fault conditions and the multi-slice data, and can realize the rapid diagnosis of the multi-physical field distribution inside the transformer, so as to accurately, comprehensively and integrally evaluate the overheated state of the transformer winding. Description of the Drawings

[0131] By reading the following detailed description of the preferred embodiments, various other advantages and benefits will become clear to those of ordinary skill in the art. The drawings are only for the purpose of showing the preferred embodiments and are not considered to be a limitation of the present application. Moreover, throughout the drawings, the same reference numerals are used to represent the same components. In the drawings:

[0132] Figure 1 It is a schematic flowchart for realizing the overheat diagnosis of the transformer winding provided by the embodiment of the present application;

[0133] Figure 2 It is a schematic flowchart for constructing a support vector machine model provided by the embodiment of the present application;

[0134] Figure 3 It is a flowchart for realizing the calculation of the temperature field distribution provided by the embodiment of the present application;

[0135] Figure 4 It is a flowchart for realizing the calculation of the electric field distribution provided by the embodiment of the present application;

[0136] Figure 5 It is a schematic diagram of longitudinal slice scanning and axial slice scanning provided by the embodiment of the present application;

[0137] Figure 6 It is a schematic diagram of longitudinal plus-segment slice scanning provided by the embodiment of the present application;

[0138] Figure 7 It is a schematic diagram of the device structure for realizing the overheat diagnosis of the transformer winding provided by the embodiment of the present application. Detailed implementation manners

[0139] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present application without creative efforts shall fall within the protection scope of the present application.

[0140] The solution of the present application can be implemented based on a terminal with data processing capabilities, and the terminal can be a computer, a server, the cloud, etc.

[0141] Next, in combination with Figure 1 As described above, the overheat diagnosis method for the transformer winding of the present application may include the following steps:

[0142] Step S110: Obtain the temperature data monitored by the sensing monitoring points on the transformer winding.

[0143] Specifically, sensing monitoring points may be provided on the transformer winding, and temperature sensors may be arranged at the sensing monitoring points, and the temperature data of the transformer winding may be monitored through the temperature sensors.

[0144] Step S120: Input the temperature data into a pre-constructed support vector machine model, and output sliced temperature data.

[0145] Specifically, the support vector machine model can accept the input of the monitored temperature data and output sliced temperature data. The support vector machine model simulates the correspondence between fault conditions and multi-slice data, and can realize the rapid diagnosis of the multi-physical field distribution inside the transformer.

[0146] Step S130: Construct a multi-slice scan map based on the sliced temperature data.

[0147] Step S140: Determine whether the relevant parameters of the multi-slice scan map exceed the corresponding preset thresholds. If so, execute step S150; if not, execute step S160.

[0148] Specifically, the process of determining whether the relevant parameters of the multi-slice scan map exceed the corresponding preset thresholds may be as follows:

[0149] If the temperature value during the longitudinal slice scan of the multi-slice scan map exceeds the first preset ratio of the preset thermal threshold, then determine whether the temperature value during the axial slice scan of the multi-slice scan map exceeds the second preset ratio of the preset thermal threshold, and whether the electric field value during the axial slice scan exceeds the preset ratio of the preset electric field threshold. If so, execute step S150; if not, execute step S160.

[0150] It can be understood that the state diagnosis criteria for multi-slice scan images can include a thermal threshold and an electric field threshold. Among them, the thermal threshold can be determined according to the aging of the oil-paper insulation material of the transformer winding. For example, according to the long-term aging temperature when the design service life is 40 years, the electric field threshold can be the design threshold of the transformer.

[0151] Specifically, the first preset ratio can be 80%. It can be understood that if the temperature value during the longitudinal slice scan of the multi-slice scan image exceeds 80% of the preset thermal threshold, it may indicate that the transformer winding is overheating. Then, it is necessary to further judge the temperature value and electric field value during the axial slice scan of the multi-slice scan image. If the temperature value during the longitudinal slice scan of the multi-slice scan image does not exceed 80% of the preset thermal threshold, it can be indicated that the transformer winding is in a normal state and no further judgment is required.

[0152] Furthermore, the second preset ratio can be 95%, and the preset ratio of the preset electric field threshold can also be 95%. Taking 95% is to prevent the position that has not been scanned from being the maximum value position. It can be understood that when the temperature value during the longitudinal slice scan of the multi-slice scan image exceeds the first preset ratio of the preset thermal threshold, the temperature value during the axial slice scan of the multi-slice scan image exceeds 95% of the preset thermal threshold. At the same time, the electric field value during the axial slice scan of the multi-slice scan image exceeds 95% of the preset electric field threshold, then it can be indicated that the transformer winding is in an overheating state. If the temperature value during the axial slice scan of the multi-slice scan image exceeds 95% of the preset thermal threshold, but the electric field value during the axial slice scan does not exceed 95% of the preset electric field threshold, it is not judged that the transformer winding is in an overheating state, and it can still be considered that the transformer winding is in a normal state. If the temperature value during the axial slice scan of the multi-slice scan image does not exceed 95% of the preset thermal threshold, but the electric field value during the axial slice scan exceeds 95% of the preset electric field threshold, it is also not judged that the transformer winding is in an overheating state, and it can still be considered that the transformer winding is in a normal state.

[0153] Step S150: Determine that the transformer winding is in an overheating state and send a warning prompt to the central system.

[0154] It can be understood that a warning prompt is sent to the central system to take certain monitoring means and heat dissipation measures to prevent transformer winding failures.

[0155] Step S160: Determine that the transformer winding is in a normal state.

[0156] The overheating diagnosis method for the transformer winding provided in this embodiment obtains the temperature data monitored by the sensing monitoring points on the transformer winding, inputs the temperature data into a pre-constructed support vector machine model, and outputs the sliced temperature data. Based on the sliced temperature data, a multi-slice scan map is constructed. If the temperature value during the longitudinal slice scan of the multi-slice scan map exceeds the first preset ratio of the preset thermal threshold, it is determined whether the temperature value during the axial slice scan of the multi-slice scan map exceeds the second preset ratio of the preset thermal threshold, and the electric field value during the axial slice scan exceeds the preset ratio of the preset electric field threshold. If so, it is determined that the transformer winding is in an overheating state, and a warning prompt is sent to the central system. If not, it is determined that the transformer winding is in a normal state. Thus, before analyzing and evaluating the multi-physical field results, the sliced temperature data is output through the support vector machine model. The support vector machine model simulates the correspondence between the fault conditions and the multi-slice data, and can realize the rapid diagnosis of the multi-physical field distribution inside the transformer, so as to accurately, comprehensively, and integrally evaluate the overheating state of the transformer winding.

[0157] Considering that when the result output by the support vector machine model deviates too much from the actual value and the theoretical value, it may lead to inaccurate and unreliable final judgment results. Based on this, in some embodiments of the present application, after the temperature data is input into the pre-constructed support vector machine model and the sliced temperature data is output as mentioned in the foregoing embodiment, it may include:

[0158] Judge whether the sliced temperature data is outside the characteristic temperature range of the temperature data and outside the target temperature range. If so, it is prompted that no characteristic is recognized.

[0159] Among them, the lower limit value of the target temperature range can be the temperature of the monitoring point of the transformer winding in the normal fault state, and the upper limit value of the target temperature range can be the temperature of the monitoring point of the transformer winding in the most serious fault state.

[0160] Specifically, when the sliced temperature data is outside the characteristic temperature range of the temperature data, it can indicate that the sliced temperature data deviates too much from the temperature data characteristics.

[0161] It can be understood that the temperature sensors on the surface of the transformer winding can be designed in multiple places. Since there may be unexpected faults in the input data when the support vector machine model learns the input data, resulting in the output result not fully matching the input data, which is manifested as a large error in several points on the support vector machine model. When these situations occur, the value of the sliced temperature data is outside the characteristic temperature range of the temperature data and outside the target temperature range. Then, the sliced temperature data should not be used as the judgment basis for the overheating diagnosis of the transformer winding, and it can be prompted that no characteristic is recognized and manual inspection and analysis of the data are prompted.

[0162] In some embodiments of the present application, the construction process of the support vector machine model mentioned in the above embodiments is introduced, as Figure 2 shown, the construction process may include:

[0163] Step S210: Obtain a plurality of temperature-carrying capacity distribution data of the transformer winding in the normal state and a plurality of temperature-carrying capacity distribution data of the transformer winding in the abnormal state.

[0164] For example, in the normal state, a plurality of temperature-carrying capacity distribution data can be obtained, such as distribution data A1: normal state (ambient temperature 1, carrying capacity 1, no fault), distribution data A2: normal state (ambient temperature 1, carrying capacity 2, no fault), distribution data A3: normal state (ambient temperature 2, carrying capacity 1, no fault) - distribution 3, etc. In the abnormal state, a plurality of temperature-carrying capacity distribution data can be obtained, such as distribution data B1: abnormal state (ambient temperature 1, carrying capacity 1, fault 1), distribution data B2: abnormal state (ambient temperature 1, carrying capacity 2, fault 1), distribution data B3: abnormal state (ambient temperature 1, carrying capacity 1, fault 2).

[0165] Step S220: For each temperature-carrying capacity distribution data, perform multiple slice scans on the transformer winding under the temperature-carrying capacity distribution data to obtain a plurality of slices, and determine the temperature field distribution and electro-thermal coupling field distribution inside the transformer winding on each slice.

[0166] It can be understood that the plurality of slices may include a plurality of axial slices and a plurality of longitudinal slices.

[0167] Specifically, the process of performing multiple slice scans on the transformer winding to obtain a plurality of slices may include:

[0168] S2201: Perform multiple axial slice scans on the transformer winding to obtain a plurality of axial slices.

[0169] Among them, the plurality of axial slices may include axial slices that cut through the sensing monitoring points of the transformer winding.

[0170] Specifically, when performing multiple axial slice scans, if the monitoring points are not cut through, additional axial slice scans passing through the monitoring points are added, so that the finally obtained plurality of axial slices include axial slices that cut through the sensing monitoring points of the transformer winding. The schematic diagram of the axial slice scan is as Figure 5 shown.

[0171] S2202: Perform multiple longitudinal slice scans on the transformer winding to obtain a plurality of longitudinal slices.

[0172] Among them, the multiple longitudinal slices may include longitudinal slices that cut through the sensing monitoring points of the transformer winding.

[0173] Specifically, when performing multiple longitudinal slice scans, if the monitoring points are not cut through, additional longitudinal slice scans passing through the monitoring points are added, so that the finally obtained multiple longitudinal slices include longitudinal slices that cut through the sensing monitoring points of the transformer winding. The schematic diagram of the longitudinal slice scan is as Figure 5 shown.

[0174] S2203. Combine each axial slice and each longitudinal slice to obtain multiple slices.

[0175] Step S230. Under each temperature-carrying capacity distribution data, use the temperature field distribution in the temperature-carrying capacity distribution data as the initial value for electric field calculation, determine the first change relationship of the insulating oil of the transformer winding with temperature change, and determine the second change relationship of the resistivity of the insulating paper of the transformer winding with temperature change.

[0176] Specifically, based on the operating experience data of the transformer winding, the temperature field distributions of the transformer winding under normal operation and different fault degrees can be obtained.

[0177] Among them, the calculation process of the temperature field distribution can be as Figure 3 shown. First, establish a three-dimensional finite element model, perform mesh division according to different material properties, and export the mesh file. Start the iteration. In each iteration, it is necessary to calculate the heating power of the conductor and export the load configuration file, and then import the heating power into the ANSYS-CFX software for thermal-fluid coupling analysis, then the three-dimensional temperature field distribution and flow field distribution of the research object can be obtained. Judge whether the difference between the results of two adjacent iterations is less than 0.01 K, where K is the preset minimum temperature value. If so, output the temperature field distribution under the current iteration and use it as the final temperature field distribution. Otherwise, perform the next iteration.

[0178] Furthermore, after obtaining the temperature field distribution of the transformer winding, use the temperature field distribution as the initial value for electric field calculation, and set the change relationships of the insulating oil and the resistivity of the insulating paper of the transformer oil-paper insulation with temperature in the electric field calculation, so that the dielectric constants and conductivities in different temperature regions during the electric field calculation correspond one-to-one with the measured values at this temperature, and thus the electric field distribution under the electromagnetic-thermal-fluid multi-physical field coupling simulation calculation method is calculated.

[0179] Among them, the calculation process of the electric field distribution can be as Figure 4As shown below. First, set the initial resistivity value, calculate the temperature field distribution by the electromagnetic-thermal-fluid method, import the temperature field results into the physical environment, calculate the resistivity values of different elements, assign different material properties to them, and calculate the temperature field distribution by the electromagnetic-thermal-fluid method. Determine whether the convergence condition is met. If not, continue to import the latest temperature field results into the physical environment for calculation. If so, calculate the electric field calculation parameters of different materials under the latest temperature field distribution, assign different material properties, and finally calculate the electric field distribution under this temperature distribution.

[0180] Step S240: Based on the temperature field distribution and electro-thermal coupling field distribution inside the transformer winding on each slice under a preset attention threshold, determine the attention areas in each slice.

[0181] Among them, the preset attention threshold may include an electric field strength threshold. For example, the area where the electric field strength ≥ 1 kV / mm can be extracted, and a rectangle is made with the maximum and minimum values of the x and y coordinates within this area as the boundaries, which is the attention area of the electric field strength. The preset attention threshold may include a temperature threshold. For example, the area where the temperature ≥ 40 °C can be extracted, and a rectangle is made with the maximum and minimum values of the x and y coordinates within this area as the boundaries, which is the attention area of the temperature field. Finally, the union of the electric field attention area and the temperature field attention area is the finally delimited attention area. By determining the attention area, the data volume of machine learning and the visualization calculation volume during output can be reduced.

[0182] Step S250: Based on the attention areas in each slice, determine the simulation calculation results corresponding to the positions of the sensing monitoring points of the transformer winding.

[0183] Specifically, the process of determining the simulation calculation results corresponding to the positions of the sensing monitoring points of the transformer winding based on the attention areas in each slice may include:

[0184] S2501: According to the attention areas in each longitudinal slice, determine the temperature field calculation results corresponding to the positions of the sensing monitoring points of the transformer winding.

[0185] S2502: According to the attention areas of each axial slice, determine the temperature field calculation results corresponding to the positions of the sensing monitoring points of the transformer winding, and the electric field calculation results corresponding to the positions of the sensing monitoring points of the transformer winding.

[0186] S2503: Combine each temperature field calculation result and each electric field calculation result to obtain the simulation calculation results corresponding to the positions of the sensing monitoring points of the transformer winding.

[0187] Step S260: Based on each slice, the simulation calculation results, each first change relationship, and each second change relationship, construct a support vector machine model.

[0188] Specifically, the process of constructing a support vector machine model based on each slice, the simulation calculation results, each first variation relationship, and each second variation relationship may include:

[0189] S2601. Establish a regression function based on each slice and the simulation calculation results.

[0190] Among them, the regression function is:

[0191] y = W·x + b

[0192] Among them,

[0193] Among them, y is a vector composed of the actual temperatures of each slice, m is the number of slices, x is a vector composed of the temperature field calculation results of each slice that cuts through the sensing monitoring points in the simulation calculation results, n is the number of slices that cut through the sensing monitoring points in each slice, W is a regression weight coefficient matrix, and b is a bias coefficient vector.

[0194] S2602. Construct an insensitive loss function.

[0195] Among them, the insensitive loss function is:

[0196]

[0197] ε is a preset insensitive loss coefficient, and g(x) is the predicted value obtained for each value of the vector x on its first variation relationship and the predicted value obtained on its second variation relationship.

[0198] S2603. Determine a penalty factor based on the insensitive loss function, and under the penalty factor, determine a support vector machine optimization problem.

[0199] Among them, the optimization objective of the support vector machine optimization problem is:

[0200]

[0201] s.t. (W·x i ) - y i + b ≤ ε + ξ ij

[0202]

[0203] ξ (*) ≥ 0

[0204] Among them, ξ (*) is a preset slack variable, C is the penalty factor, and l is the dimension of the preset slack variable.

[0205] S2604. Solve the dual problem corresponding to the support vector machine optimization problem to obtain the solution of the dual problem.

[0206] Among them, the dual problem is:

[0207]

[0208]

[0209]

[0210] Among them, the solution of the dual problem is:

[0211]

[0212] S2605. According to the solution of the dual problem, use the following formula to determine the weight coefficient of the support vector machine regression model and the bias coefficient of the support vector machine regression model:

[0213]

[0214] Among them, is the weight coefficient of the support vector machine regression model, is the bias coefficient of the support vector machine regression model.

[0215] S2606. Under the weight coefficient of the support vector machine regression model and the bias coefficient of the support vector machine regression model, construct a support vector machine model, and the support vector machine model is:

[0216]

[0217] Among them, Y is the output value of the support vector machine model, X is the input value of the support vector machine model, and K() is the model coefficient function of the support vector machine model.

[0218] For the overheating diagnosis method of the transformer winding provided in this embodiment, the learning samples of the constructed support vector machine model include multiple slice scan data obtained by multi-slice scanning of different scales according to the structural characteristics of the transformer. On the basis of multi-slice scanning, multi-physical field simulation calculation data is extracted, and the corresponding relationship between the structural simulation fault conditions and the multi-slice data is established to realize the rapid diagnosis of the multi-physical field distribution inside the transformer, providing a basis for the digital transformation and operation evaluation of the transformer.

[0219] In some embodiments of the present application, the process of performing multiple longitudinal slice scans on the transformer winding mentioned in the above embodiment to obtain multiple longitudinal slices is introduced, and this process may include:

[0220] S1. Taking the middle part of the primary winding coil of the transformer winding and the middle part of the secondary winding coil of the transformer winding as the scanning direction, perform multiple longitudinal close - up slicing scans on the transformer winding to obtain a number of first - type longitudinal slices.

[0221] Specifically, for the longitudinal slicing scan of the transformer winding, the middle part of the primary winding coil and the middle part of the secondary winding coil can be selected (as Figure 6 shown), and according to the fault data and temperature field distribution of the actual operation of similar products, perform close - up slicing at the coil at the fault - prone point and the hot - spot temperature to obtain the first - type longitudinal slices.

[0222] S2. Taking the middle part of the oil gap of the primary winding of the transformer winding and the middle part of the oil gap of the secondary winding of the transformer winding as the scanning direction, perform multiple longitudinal close - up slicing scans on the transformer winding to obtain a number of second - type longitudinal slices.

[0223] Specifically, for the longitudinal slicing scan of the transformer winding, the middle part of the oil gap of the primary winding and the middle part of the oil gap of the secondary winding can be selected (as Figure 6 shown), and according to the fault data and temperature field distribution of the actual operation of similar products, perform close - up slicing at the coil at the fault - prone point and the hot - spot temperature to obtain the second - type longitudinal slices.

[0224] S3. Combine each first - type longitudinal slice and each second - type longitudinal slice to obtain a number of longitudinal slices.

[0225] In some embodiments of the present application, the process of performing multiple axial slicing scans on the transformer winding mentioned in the above - mentioned embodiments to obtain a number of axial slices is introduced. This process may include:

[0226] S1. Determine the fault - prone points of the coils and the temperature hot - spots in the transformer winding according to the historical fault data of the transformer winding and the temperature field distribution.

[0227] S2. Taking the symmetry axis of the transformer winding as the rotation axis, and taking the preset angular range of the fault - prone points of the coils and the temperature hot - spots as the scanning direction, perform axial slicing on the transformer winding within the preset angular range with a step - slicing angle not exceeding the first preset step angle to obtain a number of first - type axial slices.

[0228] Specifically, the preset angular range can be a 10° range, and the first preset step angle can be 0.1°. Then axial slicing can be performed at the coil at the fault - prone point and the hot - spot temperature within a 10° range with a rotation angle step not exceeding 0.1° to obtain a number of first - type axial slices.

[0229] S3. With the axis of symmetry of the transformer winding as the rotation axis, and with the coil prone to failure points and other ranges within the preset angle range as the scanning directions, perform axial slicing on the transformer winding within other ranges of the preset angle range with a step slicing angle not exceeding a second preset step angle, to obtain a number of second-type axial slices.

[0230] Specifically, the second preset step angle can be 2°. Then, axial slicing can be performed with a rotation angle step not exceeding 2° outside the 10° range at the coil prone to failure points and at the hot spot temperature points, to obtain multiple second-type axial slices.

[0231] S4. Combine each first-type axial slice and each second-type axial slice to obtain multiple axial slices.

[0232] Next, the device for realizing overheating diagnosis of a transformer winding provided in the embodiments of the present application will be described. The device for realizing overheating diagnosis of a transformer winding described below can be correspondingly referred to with the method for realizing overheating diagnosis of a transformer winding described above.

[0233] See Figure 7 , Figure 7 which is a schematic structural diagram of a device for realizing overheating diagnosis of a transformer winding disclosed in the embodiments of the present application.

[0234] As Figure 7 shown, the device may include:

[0235] A temperature data acquisition unit 11, configured to acquire temperature data monitored at the sensor monitoring points on the transformer winding;

[0236] A sliced temperature data output unit 12, configured to input the temperature data into a pre-constructed support vector machine model and output sliced temperature data;

[0237] A multi-slice scan map construction unit 13, configured to construct a multi-slice scan map based on the sliced temperature data;

[0238] A threshold judgment unit 14, configured to, if the temperature value during longitudinal slice scanning of the multi-slice scan map exceeds a first preset ratio of a preset heat threshold, determine whether the temperature value during axial slice scanning of the multi-slice scan map exceeds a second preset ratio of the preset heat threshold, and whether the electric field value during axial slice scanning exceeds a preset ratio of a preset electric field threshold. If so, execute the overheating state determination unit 15; if not, execute the normal state determination unit 16;

[0239] The overheating state determination unit 15 is configured to determine that the transformer winding is in an overheating state and send a warning prompt to the central system;

[0240] The normal state determination unit 16 is configured to determine that the transformer winding is in a normal state.

[0241] Among them, the specific implementation logics of the above-mentioned respective units can refer to the relevant introductions in the foregoing part of the overheat diagnosis method for the transformer winding, which will not be elaborated here.

[0242] Finally, it should also be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements includes not only those elements but also other elements not expressly listed, or also includes elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "comprising an..." does not exclude the existence of additional identical elements in the process, method, article or device comprising the element.

[0243] The various embodiments in this specification are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. The various embodiments can be combined as needed, and the same or similar parts can be referred to each other.

[0244] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be obvious to those skilled in the art. The general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but rather to the broadest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for diagnosing overheating of a transformer winding, characterized in that, Including: Obtaining the temperature data monitored by the sensing monitoring points on the transformer winding; Inputting the temperature data into a pre-constructed support vector machine model to output sliced temperature data; Constructing a multi-slice scan graph based on the sliced temperature data; If the temperature value during the longitudinal slice scan of the multi-slice scan graph exceeds a first preset ratio of a preset thermal threshold, then determine whether the temperature value during the axial slice scan of the multi-slice scan graph exceeds a second preset ratio of the preset thermal threshold, and whether the electric field value during the axial slice scan exceeds a preset ratio of a preset electric field threshold; If so, determine that the transformer winding is in an overheated state and send a warning prompt to the central system; If not, determine that the transformer winding is in a normal state; The construction process of the support vector machine model includes: Obtaining a plurality of temperature-carrying capacity distribution data of the transformer winding in a normal state and a plurality of temperature-carrying capacity distribution data of the transformer winding in an abnormal state; Under each temperature-carrying capacity distribution data, performing multiple slice scans on the transformer winding to obtain a plurality of slices, and determining the temperature field distribution and electrothermal coupling field distribution inside the transformer winding on each slice; For each temperature-carrying capacity distribution data, using the temperature field distribution in the temperature-carrying capacity distribution data as the initial value for electric field calculation, determining a first change relationship of the insulating oil of the transformer winding with temperature change, and determining a second change relationship of the resistivity of the insulating paper of the transformer winding with temperature change; Under a preset attention threshold, based on the temperature field distribution and electrothermal coupling field distribution inside the transformer winding on each slice, determining the attention areas in each slice; Based on the attention areas in each slice, determining the simulation calculation results corresponding to the positions of the sensing monitoring points of the transformer winding; Based on each slice, the simulation calculation results, each first change relationship, and each second change relationship, constructing a support vector machine model.

2. The method according to claim 1, wherein After inputting the temperature data into the pre-constructed support vector machine model and outputting the sliced temperature data, it further includes: Judging whether the sliced temperature data is outside the characteristic temperature range of the temperature data and outside the target temperature range, the lower limit value of the target temperature range is the monitoring point temperature of the transformer winding in a normal fault state, and the upper limit value of the target temperature range is the monitoring point temperature of the transformer winding in the most serious fault state; If so, prompt that no feature is recognized.

3. The method according to claim 1, characterized in that Performing multiple slice scans on the transformer winding to obtain a plurality of slices, including: Performing multiple axial slice scans on the transformer winding to obtain a plurality of axial slices; Performing multiple longitudinal slice scans on the transformer winding to obtain a plurality of longitudinal slices; Combining each axial slice and each longitudinal slice to obtain a plurality of slices.

4. The method according to claim 3, wherein Performing multiple longitudinal slice scans on the transformer winding to obtain a plurality of longitudinal slices, including: Taking the middle part of the primary winding coil of the transformer winding and the middle part of the secondary winding coil of the transformer winding as the scanning direction, perform multiple longitudinal cutting and slicing scans on the transformer winding to obtain a number of first-type longitudinal slices; Taking the middle part of the oil gap of the primary winding of the transformer winding and the middle part of the oil gap of the secondary winding of the transformer winding as the scanning direction, perform multiple longitudinal cutting and slicing scans on the transformer winding to obtain a number of second-type longitudinal slices; Combine each first-type longitudinal slice and each second-type longitudinal slice to obtain a number of longitudinal slices.

5. The method according to claim 3, wherein Perform multiple axial slicing scans on the transformer winding to obtain a number of axial slices, including: According to the historical fault data of the transformer winding and the temperature field distribution, determine the coil fault-prone points and temperature hot spots in the transformer winding; Taking the symmetry axis of the transformer winding as the rotation axis, and taking the preset angular range of the coil fault-prone points and the temperature hot spots as the scanning direction, and using no more than the first preset step angle as the step slicing angle within the preset angular range, perform axial slicing on the transformer winding to obtain a number of first-type axial slices; Taking the symmetry axis of the transformer winding as the rotation axis, and taking the other range of the preset angular range of the coil fault-prone points as the scanning direction, and using no more than the second preset step angle as the step slicing angle within the other range of the preset angular range, perform axial slicing on the transformer winding to obtain a number of second-type axial slices; Combine each first-type axial slice and each second-type axial slice to obtain a number of axial slices.

6. The method according to any one of claims 3-5, characterized in that, The number of axial slices includes axial slices that cut through the sensing monitoring points of the transformer winding, and the number of longitudinal slices includes longitudinal slices that cut through the sensing monitoring points of the transformer winding.

7. The method according to claim 3, characterized in that, Determining the simulation calculation results of the positions corresponding to the sensing monitoring points of the transformer winding based on the regions of interest in each slice includes: According to the regions of interest in each longitudinal slice, determine the temperature field calculation results of the positions corresponding to the sensing monitoring points of the transformer winding; According to the regions of interest in each axial slice, determine the temperature field calculation results of the positions corresponding to the sensing monitoring points of the transformer winding, and the electric field calculation results of the positions corresponding to the sensing monitoring points of the transformer winding; Combine each temperature field calculation result and each electric field calculation result to obtain the simulation calculation results of the positions corresponding to the sensing monitoring points of the transformer winding.

8. An overheat diagnosis device for a transformer winding, characterized in that Include: A temperature data acquisition unit for acquiring the temperature data monitored by the sensing monitoring points on the transformer winding; A sliced temperature data output unit for inputting the temperature data into a pre-constructed support vector machine model and outputting the sliced temperature data; A multi-slice scan graph construction unit for constructing a multi-slice scan graph based on the sliced temperature data; A threshold judgment unit, which is configured to, if the temperature value during the longitudinal slice scanning of the multi-slice scan map exceeds a first preset ratio of a preset thermal threshold, determine whether the temperature value during the axial slice scanning of the multi-slice scan map exceeds a second preset ratio of the preset thermal threshold, and whether the electric field value during the axial slice scanning exceeds a preset ratio of a preset electric field threshold. If so, execute the overheat state determination unit; if not, execute the normal state determination unit; The overheat state determination unit is configured to determine that the transformer winding is in an overheat state and send a warning prompt to the central system; The normal state determination unit is configured to determine that the transformer winding is in a normal state; A distribution data acquisition unit, which is configured to acquire a plurality of temperature-carrying capacity distribution data of the transformer winding in a normal state and a plurality of temperature-carrying capacity distribution data of the transformer winding in an abnormal state; A slice scanning unit, which is configured to perform multiple slice scans on the transformer winding under each temperature-carrying capacity distribution data to obtain a plurality of slices; A field distribution determination unit, which is configured to determine the temperature field distribution and the electro-thermal coupling field distribution inside the transformer winding on each slice; A variation relationship determination unit, which is configured to, for each temperature-carrying capacity distribution data, use the temperature field distribution in the temperature-carrying capacity distribution data as the initial value for electric field calculation, determine a first variation relationship of the insulating oil of the transformer winding with temperature, and determine a second variation relationship of the resistivity of the insulating paper of the transformer winding with temperature; A concerned area determination unit, which is configured to determine the concerned areas in each slice based on the temperature field distribution and the electro-thermal coupling field distribution inside the transformer winding on each slice under a preset concern threshold; A simulation calculation result determination unit, which is configured to determine the simulation calculation results at the positions corresponding to the sensing monitoring points of the transformer winding based on the concerned areas in each slice; A support vector machine model construction unit, which is configured to construct a support vector machine model based on each slice, the simulation calculation results, each first variation relationship, and each second variation relationship.

Images