An oil-immersed transformer latent fault source positioning and diagnosis method based on cooling condition switching response difference

By acquiring dissolved gas data and cooling condition data from multiple locations in an oil-immersed transformer, identifying cooling condition switching events, establishing a gas background evolution model, and extracting differential response features, the problem of difficulty in locating fault characteristic gases after cooling condition switching in an oil-immersed transformer was solved, and high-precision location of latent fault source areas was achieved.

CN123066449APending Publication Date: 2026-11-10CHINA THREE GORGES UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611237612.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-14
Publication Date
2026-11-10

AI Technical Summary

Technical Problem

Existing technologies, after switching cooling conditions in oil-immersed transformers, make it difficult for fault-characteristic gases to migrate, leading to difficulties in using fixed oil sampling or single-point oil dissolved gas diagnostic methods to utilize the differences in response arrival order and response amplitude between different oil flow-related monitoring locations, thus failing to accurately locate latent fault sources.

Method used

By acquiring dissolved gas data and cooling condition data from multiple oil flow-related monitoring locations, cooling condition switching events are identified, a gas background evolution model is established, gas differential response features across operating conditions are extracted, and the gas migration response fingerprints of candidate fault source regions are matched to determine the target fault source region.

Benefits of technology

This method enables fault location by utilizing changes in oil flow path after switching cooling conditions in oil-immersed transformers. This reduces the impact of load and oil temperature changes and improves the reliability and accuracy of fault location.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123066449A_ABST
    Figure CN123066449A_ABST
Patent Text Reader

Abstract

The application discloses a kind of oil-immersed transformer latent fault source positioning diagnosis methods based on cooling condition switching response difference, it is related to power transformer condition monitoring and fault diagnosis technical field.The method obtains the dissolved gas data and cooling condition data of multiple oil flow correlation monitoring positions of oil-immersed transformer;Identify cooling condition switching event, determine the background period before switching and the response period after switching;Based on the data before switching, establish gas background evolution model, obtain background predicted concentration;Actual concentration after switching is different from background predicted concentration, and difference response feature vector is constructed;And it is matched with candidate fault source area gas migration response fingerprint, and positioning reliability is calculated, and target fault source area is determined.The application can use the gas migration difference after cooling condition switching to narrow the scope of fault source investigation, improve positioning reliability.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of power transformer condition monitoring and fault diagnosis technology, specifically to a method for locating and diagnosing latent fault sources in oil-immersed transformers based on differential response to cooling condition switching. Background Technology

[0002] Oil-immersed power transformers are critical equipment in power systems. During transformer operation, the insulating oil not only provides insulation and cooling but also serves as the medium for transmitting information about latent internal faults. When problems such as loose winding lead connections, increased local contact resistance, localized overheating of the windings, abnormal core clamping, or localized blockage of oil channels occur, the insulating oil and oil-paper insulation material near the fault source will decompose under thermal and electrical stress, producing characteristic gases such as hydrogen, methane, ethane, ethylene, acetylene, carbon monoxide, and carbon dioxide.

[0003] Dissolved gas analysis in oil is an important means of identifying latent faults in oil-immersed transformers. Existing methods typically obtain oil samples through fixed sampling points or online oil chromatography devices, and determine the presence and type of fault based on characteristic gas concentrations, gas ratios, or gas growth rates. While these methods can reflect abnormal gas states in the oil samples, the detection results mainly correspond to mixed oil samples from fixed sampling locations, making it difficult to directly reflect the specific oil flow correlation area where the fault source is located.

[0004] For oil-immersed transformers with forced oil circulation or forced oil-guided cooling, the operating status of the submersible pump, cooler group, fan group, or cooling oil circuit valve group will affect the oil flow velocity, oil flow direction, local circulation path, and temperature gradient inside the tank. After the cooling condition changes, the fault characteristic gas that was originally concentrated in a certain oil area will migrate to other areas with the changed oil flow path, causing different oil sampling locations to exhibit different response sequence, response amplitude, and response hysteresis characteristics in different detection cycles.

[0005] In actual operation, an increase in the concentration of a characteristic gas detected at a fixed oil sampling port may be due to continuous gas production from a fault source, or it may be caused by existing fault gas being transported to that sampling port via a new oil flow path after a cooling condition switch. Because existing methods for diagnosing dissolved gases in fixed oil sampling ports or single-point oil sources primarily focus on single-point concentration changes, they fail to fully utilize the differences in the arrival order and amplitude of responses between multiple oil flow-related monitoring locations after a cooling condition switch. Therefore, it is difficult to determine which oil flow-related region the abnormal gas primarily originates from.

[0006] In particular, for candidate fault source areas such as the upper part of the high-voltage winding and lead joint area, the middle area of ​​the high-voltage winding, the lower area of ​​the high-voltage winding, the core and clamping area, the low-voltage winding area, and the area associated with the cooler inlet, these areas may be close to each other in the internal space of the transformer, but their relationships with the oil flow channels and cooling oil circuits are different. If only the gas concentration or growth rate at a single oil sampling location is used, it is often only possible to determine that an anomaly exists, but it is difficult to further narrow down the scope of fault source investigation.

[0007] Therefore, the existing technology has the following technical problems: after the cooling condition of an oil-immersed transformer is switched, the fault characteristic gas migrates again with the changed oil flow path. Existing fixed oil sampling or single-point oil dissolved gas diagnosis methods are difficult to use the difference in response arrival order and response amplitude between different oil flow associated monitoring locations to determine the target oil flow associated area where the latent fault source is located. Summary of the Invention

[0008] To address the aforementioned issues, the present invention aims to provide a method for locating and diagnosing latent fault sources in oil-immersed transformers based on differential response to cooling condition switching. This method transforms the changes in oil flow path caused by cooling condition switching into diagnostic excitations. By using dissolved gas data from multiple locations in the oil, background trend stripping, differential response feature extraction, and gas migration response fingerprint matching, the latent fault source region can be located.

[0009] To achieve the above-mentioned technical features, the objective of this invention is as follows: The first aspect of this invention provides a method for locating and diagnosing latent fault sources in oil-immersed transformers based on differential response to cooling condition switching, comprising the following steps: S1. Obtain dissolved gas data and cooling condition data in the oil of the oil-immersed transformer at multiple oil flow-related monitoring locations, and record the data as follows: Characteristic gases in the first The oil flow correlation monitoring location is at the first The concentration for each detection cycle is ,in, , This represents the total number of characteristic gas species. , This represents the total number of oil flow-related monitoring locations. S2. Identify cooling condition switching events based on the cooling condition data, and record the detection period corresponding to the cooling condition switching event as . And according to the detection cycle Determine the time period for switching between foreground and background and the response period after the switch ,in This represents the number of detection cycles included in the background time period. The number of detection cycles included in the response period; S3, according to the aforementioned background / pre-switching time period A gas background evolution model was established based on dissolved gas data in the oil, and the first... Characteristic gases in the first Background predicted concentration at each oil flow-related monitoring location ; S4, switch the response period The actual dissolved gas data in the oil is compared with the background predicted concentration to obtain the cross-condition gas differential response. And based on the cross-condition gas differential response Constructing differential response feature vectors ; S5. The differential response feature vector Compared with the pre-established gas migration response fingerprint of the candidate fault source region Matching is performed to obtain the location confidence of each candidate fault source region. The candidate fault source region with the highest location confidence was identified as the target fault source region. This is the sequence number of the candidate fault source region.

[0010] Preferably, in step S1, the process of determining the multiple oil flow associated monitoring locations includes the following sub-steps: S1.1. Based on the tank structure, winding arrangement, core position, main oil passage arrangement and cooling oil passage connection of the oil-immersed transformer, determine the candidate oil sampling locations that can reflect the gas migration state in different oil flow areas. S1.2 Select at least two oil flow associated monitoring locations from the candidate oil sampling locations. The oil flow associated monitoring locations include at least two of the following: the oil area monitoring location at the top of the oil tank, the oil area monitoring location at the side of the winding, the oil area monitoring location at the bottom of the oil tank, the oil circuit monitoring location at the cooler inlet, and the oil circuit monitoring location at the cooler outlet. S1.3. Collect oil samples from the oil flow correlation monitoring locations through multiple oil sampling branches, and obtain the corresponding dissolved gas data in the oil through the dissolved gas detection device in the oil. S1.4. Number the dissolved gas data in the oil at different oil flow-related monitoring locations according to the detection cycle to form a multi-location gas concentration sequence. .

[0011] Preferably, in step S2, the process of identifying the cooling condition switching event includes the following sub-steps: S2.1 Collect the operating status of at least one of the following cooling equipment: submersible pump, cooler group, fan group, and cooling oil circuit valve group; S2.2 Determine whether the operating status of the cooling equipment has changed; S2.3 When at least one of the following events is detected: submersible pump activation, submersible pump deactivation, submersible pump group switching, cooler group activation, cooler group deactivation, cooler group switching, fan group activation, fan group deactivation, cooling oil circuit valve group status change, or switching between natural oil circulation and forced oil circulation, a cooling condition switching event is determined to have occurred. S2.4 Record the detection cycle corresponding to the cooling condition switching event. and with the detection cycle Determine the background time period before switching based on the benchmark. and the response period after the switch .

[0012] Preferably, in step S3, the process of establishing the gas background evolution model includes the following sub-steps: S3.1, from switching the foreground and background time periods Internal extraction Characteristic gases in the first Historical concentration sequences of oil flow-related monitoring locations; S3.2 Obtain the transformer load corresponding to the historical concentration sequence. and top oil temperature ,in For the first Transformer load for each testing cycle For the first The top oil temperature of the transformer during each testing cycle; S3.3, Based on the aforementioned historical concentration sequence and transformer load and top oil temperature The model parameters of the gas background evolution model are obtained by fitting. S3.4 Predict the response period after the handover based on the model parameters. Background predicted concentration within The gas background evolution model is as follows: ; In the formula, For the first Characteristic gases in the first The trend coefficient of each monitoring location changing with the detection cycle; For the first Characteristic gases in the first Background concentration constant term for each monitoring location; For the first Characteristic gases in the first Load impact coefficient at each monitoring location; For the first Characteristic gases in the first The oil temperature influence coefficient at each monitoring location.

[0013] Preferably, in step S4, the extraction process of the gas differential response across operating conditions includes the following sub-steps: S4.1, Read the response period after the switch Inner Characteristic gases in the first Actual concentration at each oil flow-related monitoring location ; S4.2 Read the background predicted concentration for the corresponding detection period. ; S4.3, the actual concentration Compared with the background predicted concentration By performing a differential operation, the gas differential response across operating conditions is obtained: ; S4.4, Based on the aforementioned cross-condition gas differential response Extract the response start detection period, response amplitude, response arrival order, and response detection period difference.

[0014] Preferably, in step S4.4, the process of determining the initial detection period and the response amplitude includes the following sub-steps: S4.41, Based on the time period before and after switching background. The fluctuation range of gas concentration within the range is used to determine the first Characteristic gases in the first Response determination threshold for each oil flow-related monitoring location ; S4.42, Response period after handover Within, determine the gas differential response across operating conditions. Does the absolute value exceed the response determination threshold? ; S4.43, The response threshold will be exceeded for the first time. The detection cycle is determined as the response start detection cycle. ,in: ; S4.44, Response period after handover Within this range, the maximum absolute value of the gas differential response across operating conditions is determined and used as the response amplitude. ,in: .

[0015] Preferably, in step S4, the differential response feature vector The construction process includes the following sub-steps: S4.5. Determine the response arrival order vector based on the response start detection period corresponding to multiple oil flow associated monitoring locations. ; S4.6. Determine the response amplitude ratio vector based on the response amplitudes corresponding to multiple oil flow correlation monitoring locations. ; S4.7 Determine the response detection cycle difference vector based on the response start detection cycle difference between different oil flow associated monitoring locations. ; S4.8, The response arrival order vector Response amplitude scaling vector and response detection period difference vector Combining these elements, we obtain the difference response feature vector: .

[0016] Preferably, the gas migration response fingerprint of the candidate fault source region The establishment process includes the following sub-steps: S5.1 Based on the tank structure, winding position, core position, oil passage layout, cooler inlet position, cooler outlet position, and oil intake branch position of the oil-immersed transformer, divide multiple candidate fault source areas. S5.2 Set an equivalent gas release source in each candidate fault source region; S5.3 For at least one cooling condition switching event, simulate or calibrate the process of the fault gas released by the equivalent gas release source migrating to multiple oil flow associated monitoring locations; S5.4 Extract the reference response arrival order, reference response amplitude ratio, and reference response detection cycle difference for each candidate fault source region; S5.5, Combine the arrival order of the reference responses, the amplitude ratio of the reference responses, and the difference in the detection period of the reference responses to form the first... Gas migration response fingerprints of candidate fault source regions: ; In the formula, For the first The reference response arrival order vector of each candidate fault source region; For the first Reference response amplitude scaling vector for each candidate fault source region; For the first Reference response detection period difference vector for each candidate fault source region; The candidate fault source regions include at least two of the following: the upper high-voltage winding and lead joint region, the middle high-voltage winding region, the lower high-voltage winding region, the core and clamping parts region, the low-voltage winding region, and the cooler inlet associated region.

[0017] Preferably, in step S5, the process of determining the target fault source region includes the following sub-steps: S5.6, The measured differential response eigenvector Gas migration response fingerprints of each candidate fault source region Compare; S5.7 Calculate the differences in response arrival order, response amplitude ratio, and response detection cycle between the measured response and each candidate fault source region; S5.8 Calculate the first [response] based on the differences in response arrival order, response amplitude ratio, and response detection period. Differences among candidate fault source regions ,in: ; In the formula, A function is calculated to respond to the difference in arrival order; A function to calculate the difference in response amplitude proportions; The function for calculating the difference in detection cycle time is as follows; , , These are the weighting coefficients corresponding to the order of response arrival, the proportion of response amplitude, and the difference in response detection period, respectively. S5.9, Based on the differences in the regions of each candidate fault source Calculate the reliability of the location ,in: ; In the formula, This represents the total number of candidate fault source regions. The traversal sequence number of the candidate fault source region; It is a natural exponential function; S5.10. Determine the candidate fault source region with the highest location confidence as the target fault source region. ,in: .

[0018] Preferably, after step S4, the method further includes delay compensation for the oil sample acquisition link, specifically including the following sub-steps: S6.1, Obtain the first The equivalent dead volume of the oil sampling branch corresponding to each oil flow correlation monitoring location Sampling flow rate Valve group switching stabilization time Oil intake branch flushing correction time ; S6.2, based on the equivalent dead volume Sampling flow rate Valve group switching stabilization time Oil intake branch flushing correction time Calculate the first Sampling link delay time corresponding to each oil flow-related monitoring location ,in: ; S6.3 Obtain the gas response time obtained from the detection ,in For the first Characteristic gases in the first Measured response time at each monitoring location; S6.4, the gas response time Subtract the sampling link delay time Response time after compensation ,in: ; S6.5, Based on the compensated response time Correct the difference between the response arrival order and the response detection cycle.

[0019] Another aspect of the present invention provides a latent fault source location and diagnosis system for oil-immersed transformers based on differential cooling condition switching response, comprising: a data acquisition module for acquiring dissolved gas data in oil and cooling condition data at multiple oil flow associated monitoring locations of the oil-immersed transformer; a switching event identification module for identifying cooling condition switching events based on the cooling condition data and determining the background period before switching and the response period after switching; a differential response extraction module for establishing a gas background evolution model based on the dissolved gas data in oil during the background period before switching, and calculating cross-condition gas differential response based on the gas background evolution model and the actual dissolved gas data in oil during the response period after switching; a response fingerprint storage module for storing gas migration response fingerprints of candidate fault source regions; a fault source location module for matching the differential response feature vector corresponding to the cross-condition gas differential response with the gas migration response fingerprints of the candidate fault source regions and outputting the target fault source region; and a result output module for outputting the target fault source region and the corresponding location confidence level.

[0020] Preferably, the data acquisition module includes a multi-location oil sampling unit, a dissolved gas detection unit, and a cooling condition status acquisition unit; the multi-location oil sampling unit includes multiple oil sampling branches, a switching valve group, a micro-circulation pump, a bypass flushing branch, and a return oil branch; the multiple oil sampling branches are respectively connected to at least two locations among the upper oil area of ​​the oil tank, the side oil area of ​​the winding, the lower oil area of ​​the oil tank, the inlet oil line of the cooler, and the outlet oil line of the cooler; the switching valve group is used to select different oil sampling branches to connect to the dissolved gas detection unit, the bypass flushing branch is used to flush the corresponding oil sampling branch before sampling, and the return oil branch is used to return the detected oil sample to the return oil position of the oil-immersed transformer.

[0021] The present invention has the following beneficial effects: 1. This invention transforms the changes in oil flow caused by switching cooling conditions into a basis for locating fault source areas. After switching cooling conditions, characteristic gases released or enriched in different candidate fault source areas will migrate along different oil flow paths, forming different combinations of response arrival order and response amplitude at multiple oil flow-related monitoring locations. This invention utilizes these response combinations to determine the target fault source area, thereby narrowing down the scope of investigation for latent fault sources.

[0022] 2. This invention predicts the background gas concentration after cooling switching by using a gas background evolution model, and differentiates the actual gas concentration from the predicted background concentration. This can reduce the influence of the existing natural gas generation trend, load changes and oil temperature changes on the positioning results before cooling switching, and make the cross-condition gas differential response better reflect the gas migration differences shown by the cooling condition switching.

[0023] 3. This invention utilizes gas migration response fingerprints in candidate fault source regions to transform the response arrival order, response amplitude ratio, and response detection cycle difference into calculable regional location features. Compared to judging anomalies solely based on an increase in gas concentration at a single oil intake port, this invention can further output the target fault source region and its location reliability.

[0024] 4. This invention reduces the impact of oil sampling branch length, equivalent dead volume, sampling flow rate, valve group switching stabilization time and flushing correction time on the response time by compensating for oil sampling link delay, so that the corrected response arrival order and response detection cycle difference are closer to the differences in oil flow migration inside the transformer.

[0025] 5. This invention does not require the installation of additional sensors inside the high-voltage winding of the transformer. It can be implemented based on external multi-location oil sampling branches, dissolved gas detection devices in oil, and cooling condition acquisition interfaces, thus having good engineering feasibility and operational safety. Attached Figure Description

[0026] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0027] Figure 1 This is a flowchart of a method for locating and diagnosing latent fault sources in oil-immersed transformers based on differential cooling condition switching response, provided in an embodiment of the present invention. Figure 2 This is a schematic diagram of the differential positioning mechanism for cooling condition switching response provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of the multi-location oil intake and cooling oil circuit layout provided in an embodiment of the present invention; Figure 4 This is a schematic diagram of gas background evolution stripping and cross-condition differential response extraction provided in an embodiment of the present invention; Figure 5This is a schematic diagram of the construction of gas migration response fingerprints in candidate fault source regions provided in an embodiment of the present invention; Figure 6 This is a schematic diagram of response fingerprint matching and positioning provided in an embodiment of the present invention; Figure 7 This is a schematic diagram of oil sample acquisition link delay compensation provided in an embodiment of the present invention; Figure 8 This is a structural block diagram of a latent fault source location and diagnosis system provided in an embodiment of the present invention.

[0028] In the diagram: 100, Latent Fault Source Location and Diagnosis System; 110, Data Acquisition Module; 111, Multi-location Oil Sampling Unit; 112, Dissolved Gas Detection Unit in Oil; 113, Cooling Condition Status Acquisition Unit; 120, Switching Event Recognition Module; 130, Differential Response Extraction Module; 140, Response Fingerprint Storage Module; 150, Fault Source Location Module; 160, Result Output Module; 200, Oil-immersed Transformer; 210, Transformer Tank; 211, First Monitoring Position; 212, Second Monitoring Position; 213, Third Monitoring Position; 214, Fourth Monitoring Position; 220, Submersible Pump Group; 230, Cooler Group; 240, Candidate Fault Source Area; 250, Oil Sampling Branch; 260, Switching Valve Group; 270, Return Oil Branch. Detailed Implementation

[0029] To make the technical problems, technical solutions, and technical effects of the present invention clearer, the present invention will be further described below with reference to the accompanying drawings and embodiments. The following embodiments are for illustrative purposes only and are not intended to limit the scope of protection of the present invention. Where there is no conflict, the technical features in the following embodiments can be combined with each other.

[0030] Example 1: A method for locating latent fault source regions based on differential cooling condition switching response like Figures 1 to 7 As shown, this embodiment uses a 220kV forced oil circulation oil-immersed transformer 200 as an example to illustrate the complete method flow of the present invention. The oil-immersed transformer 200 includes a transformer tank 210, windings, core, oil channels, a submersible pump group 220, and a cooler group 230. The submersible pump group 220 and the cooler group 230 together form a forced oil circulation cooling path. Because the oil flow relationships between different internal regions and the upper part of the tank, the side of the windings, the lower part of the tank, and the cooler inlet are different, when the cooling conditions change, the fault gases generated or enriched in different candidate fault source regions will exhibit different migration response patterns.

[0031] This embodiment first executes step S1 to acquire dissolved gas data and cooling condition data in the oil of the oil-immersed transformer 200 at multiple oil flow-related monitoring locations. Specifically, in step S1.1, candidate oil sampling locations that can reflect the gas migration state in different oil flow regions are determined based on the tank structure, winding arrangement, core position, main oil passage arrangement, and cooling oil circuit connection relationship of the oil-immersed transformer 200. This determination process is not an arbitrary selection of oil sampling points, but rather an arrangement based on the differences in the path of fault gas migration with the oil flow: the upper oil area of ​​the tank can reflect the thermal rise and upper oil flow mixing state, the winding side oil area can reflect the gas migration state in the oil passages near the windings, the lower oil return area of ​​the tank can reflect the lower oil return path state, and the cooler inlet oil circuit can reflect the gas-rich oil flow state before entering the cooling circuit.

[0032] In step S1.2, four oil flow-related monitoring locations are selected from the candidate oil sampling locations: the first monitoring location 211 is set in the upper oil area of ​​the oil tank, the second monitoring location 212 is set in the winding side oil area, the third monitoring location 213 is set in the lower oil return area of ​​the oil tank, and the fourth monitoring location 214 is set in the cooler inlet oil line. These four monitoring locations are denoted as P1, P2, P3, and P4, respectively. With this arrangement, the arrival order and response amplitude differences of faulty gas along different oil flow paths can be observed after the cooling condition switch.

[0033] In step S1.3, the multi-location oil sampling unit 111 collects oil samples from points P1 to P4 through multiple sampling branches 250. Each sampling branch 250 is connected to the dissolved gas detection unit 112 in a time-sharing manner via a switching valve group 260. The dissolved gas detection unit 112 detects at least two gases among hydrogen, methane, ethane, ethylene, acetylene, carbon monoxide, and carbon dioxide. In this embodiment, the detection cycle is 10 minutes, and ethylene (… () will be explained as the main response gas.

[0034] In step S1.4, the dissolved gas data in the oil at different oil flow-related monitoring locations are numbered according to the detection cycle to form a multi-location gas concentration sequence. Characteristic gases in the first The oil flow correlation monitoring location is at the first The concentration for each detection cycle is recorded as follows: (1) in, , This represents the total number of characteristic gas species. , This represents the total number of oil flow-related monitoring locations. This is the detection cycle number; in this embodiment, .

[0035] While acquiring dissolved gas data in oil at multiple locations, the cooling condition status acquisition unit 113 collects the operating status of the submersible pump group 220, cooler group 230, fan group, and cooling oil circuit valve group. The cooling condition data includes at least the cooling equipment number, operating status, switching time, pre-switching combination status, and post-switching combination status.

[0036] Then, step S2 is executed to identify the cooling condition switching event based on the cooling condition data. Specifically, in step S2.1, the switching event identification module 120 continuously reads the operating status of the submersible pump group 220 and the cooler group 230. At a certain operating moment, the submersible pump group 220 switches from one set of operation to two sets of operation, while the cooler group 230 remains on.

[0037] In step S2.2, the event recognition module 120 determines whether the operating status of the cooling equipment has changed. When it is detected that the submersible pump group 220 has changed from "one group operation" to "two groups operation", the system determines that the operating status of the cooling equipment has changed.

[0038] In step S2.3, because the switching of the submersible pump group changes the forced oil circulation flow rate inside the oil-immersed transformer 200, the oil flow exchange state between the winding side oil passage, the upper oil area of ​​the tank, the lower return oil area of ​​the tank, and the cooler inlet oil passage changes. Therefore, the system determines this event as a cooling condition switching event. This cooling condition switching event is not simply removed as interference data, but rather used as an oil flow excitation for subsequent fault gas migration response analysis.

[0039] In step S2.4, the system records the detection cycle corresponding to the cooling condition switching event as follows: (2) in, This is the detection cycle number corresponding to the cooling condition switching event.

[0040] and with detection cycle Based on this, determine the background time period before the switch: (3) And the response period after the switchover: (4) in, The number of detection cycles included in background modeling; The number of detection cycles included in the response analysis.

[0041] In this embodiment, the detection cycle is 10 minutes, and the background modeling time is 24 hours. The response analysis duration is set to 6 hours, therefore The 24-hour background modeling duration was chosen to cover the intraday variations in transformer load and oil temperature; the 6-hour response analysis duration was chosen because the impact of fault gas migration with the oil flow on different monitoring locations after the forced oil circulation transformer switches to cooling mode can usually be observed within several to dozens of monitoring cycles.

[0042] Next, proceed to step S3, based on the switching of the foreground and background time periods. A gas background evolution model was established based on dissolved gas data in the oil. The reason for establishing this model is that before the cooling condition switch, the concentration of dissolved gases in the oil may change slowly with load, oil temperature, and operating status. If the concentration increment after the switch is directly used as the location basis, the concentration increment will simultaneously include the natural gas generation trend and the migration response caused by the cooling switch, thereby reducing the reliability of fault source area location.

[0043] In step S3.1, the system switches from the background time period before switching. Internal extraction Characteristic gases in the first Historical concentration sequences of oil flow-related monitoring locations.

[0044] In step S3.2, the system synchronously acquires the transformer load corresponding to the historical concentration sequence. and top oil temperature .in, For the first The transformer load during each testing cycle can affect the heating level at the windings and lead joints due to load variations. For the first The top oil temperature during each detection cycle affects the gas dissolution state, diffusion state, and oil flow state; therefore, both can be used as influencing variables in the background evolution model.

[0045] In step S3.3, the system determines the concentration based on the historical concentration sequence and transformer load. and top oil temperature The model parameters of the gas background evolution model are obtained by fitting the model. In step S3.4, the response period after the switch is predicted based on the model parameters. Background predicted concentration within .

[0046] This embodiment uses the following gas background evolution model: (5) in, For the first Characteristic gases in the first The oil flow correlation monitoring location is at the first Background predicted concentration for each detection cycle; For the first Characteristic gases in the first The trend coefficient of each monitoring location changing with the detection cycle; For the first Characteristic gases in the first Background concentration constant term for each monitoring location; For the first Characteristic gases in the first Load impact coefficient at each monitoring location; For the first Characteristic gases in the first The oil temperature influence coefficient at each monitoring location.

[0047] In one alternative implementation, when the amount of data in the background period is small or the gas concentration fluctuation is small, a moving average model, an exponential smoothing model, or a piecewise linear model can also be used, as long as the predicted background concentration in the response period after the switch can be output. All of these can be used in the implementation of this invention.

[0048] Then execute step S4 to switch the response period. The actual dissolved gas data in the oil were compared with the predicted background concentration to obtain the differential gas response across operating conditions, and a differential response feature vector was constructed based on this. .

[0049] In step S4.1, the system reads the response period after the handover. Inner Characteristic gases in the first Actual concentration at each oil flow-related monitoring location .

[0050] In step S4.2, the system reads the background predicted concentration for the corresponding detection period. .

[0051] In step S4.3, the actual concentration is differentiated from the background predicted concentration to obtain the gas differential response across operating conditions: (6) in, For the first Characteristic gases in the first Differential gas response across operating conditions at each oil flow-related monitoring location.

[0052] The differential response reflects the degree of deviation from the background trend after the cooling condition switch. If a significant positive differential response appears at a certain monitoring location, it indicates that the monitoring location is affected by the migration of gas-rich oil flow after the cooling switch; if a negative differential response appears at a certain monitoring location, it indicates that the original gas-rich oil at that monitoring location may have been carried away by the new oil flow path.

[0053] In step S4.41, the system determines the background time period before and after the switch. The fluctuation range of gas concentration within the range is used to determine the first Characteristic gases in the first Response determination threshold for each oil flow-related monitoring location In this embodiment, The standard deviation of gas differential noise during the background period before switching is taken as three times to reduce the probability of misjudgment caused by normal background fluctuations.

[0054] In step S4.42, the system responds during the post-switching period. Internal judgment of gas differential response across operating conditions Does the absolute value exceed the response determination threshold? In step S4.43, the detection period when the response determination threshold is first exceeded is determined as the response start detection period: (7) in, For the first Characteristic gases in the first The initial detection cycle of the response at each oil flow-related monitoring location.

[0055] In step S4.44, the system responds during the post-switching period. The maximum absolute value of the gas differential response across operating conditions is determined and used as the response amplitude: (8) in, For the first Characteristic gases in the first The response amplitude of each oil flow-related monitoring location.

[0056] In step S4.5, the system determines the response arrival order vector based on the response start detection period corresponding to multiple oil flow associated monitoring locations. In step S4.6, a response amplitude ratio vector is determined based on the response amplitudes corresponding to multiple oil flow associated monitoring locations. In step S4.7, the response detection cycle difference vector is determined based on the response start detection cycle difference between different oil flow associated monitoring locations. In step S4.8, the response arrival order vector is... Response amplitude scaling vector and response detection period difference vector Combining these elements, we obtain the difference response feature vector: (9) in, The response arrival order vector; For the response amplitude scaling vector; The response detection period difference vector; This is the differential response eigenvector.

[0057] This embodiment uses ethylene ( ) was used as the main analytical gas. After switching from single-pump operation to dual-pump operation under cooling conditions, the differential ethylene response at each monitoring location is shown in Table 1 below.

[0058] Table 1. Ethylene differential response at each monitoring location

[0059] Based on the above data, the arrival order vector s corresponds to the arrival order P2→P1→P3→P4; the response amplitude scaling vector... a This can include P2 / P1, P3 / P1, and P4 / P1, where P2 / P1 is 7.6 / 5.9 (approximately 1.29), P3 / P1 is 2.2 / 5.9 (approximately 0.37), and P4 / P1 is 1.5 / 5.9 (approximately 0.25); Response detection period difference vector d This could include a difference of 1 detection cycle between P2 and P1, a difference of 3 detection cycles between P3 and P1, and a difference of 5 detection cycles between P4 and P1.

[0060] Mechanistically, ethylene is typically associated with thermal failures or localized overheating in the oil. When the cooling operation switches from single-pump to dual-pump operation, the oil flow on the winding side increases. If there is poor contact or localized overheating in the upper part of the high-voltage winding and the lead joint area, the ethylene-rich oil will preferentially affect the oil area P2 on the winding side, subsequently affecting the oil area P1 at the top of the tank, while the response at the bottom of the tank P3 and the cooler inlet P4 is delayed and has a smaller amplitude. This response pattern is consistent with the oil flow migration mechanism in the upper part of the high-voltage winding and the lead joint area.

[0061] Next, step S5 is executed, which compares the differential response feature vector Z with the pre-established gas migration response fingerprint Φ of the candidate fault source region. r The system performs matching to obtain the location confidence of each candidate fault source region and determines the target fault source region.

[0062] In step S5.1, the system divides the oil-immersed transformer 200 into multiple candidate fault source regions based on the tank structure, winding position, core position, oil passage arrangement, cooler inlet position, cooler outlet position, and oil intake branch position. In this embodiment, the transformer interior is divided into five candidate fault source regions: Z1 is the upper part of the high-voltage winding and lead joint area, Z2 is the middle part of the high-voltage winding area, Z3 is the lower part of the high-voltage winding area, Z4 is the core and clamping parts area, and Z5 is the cooler inlet associated area.

[0063] The above-mentioned zoning has a clear engineering basis. Poor contact and local overheating are prone to occur in the upper part of the high-voltage winding and the lead joint area; the middle area of ​​the high-voltage winding is closely related to the heat exchange of the main oil passage; the lower area of ​​the high-voltage winding is strongly related to the lower oil return path; anomalies in the core and clamping parts area are usually related to hydrogen response and local circulation; and the area related to the cooler inlet is directly affected by the switching of the oil return path and the cooling oil circuit.

[0064] In step S5.2, the system sets an equivalent gas release source in each candidate fault source region. The equivalent gas release source is used to characterize the generation location and release path of fault gas when a latent fault occurs in that region. In step S5.3, for the cooling condition switching event of "single pump operation switching to dual pump operation", the system simulates or calibrates the migration process of ethylene released from the equivalent gas release source to P1 to P4. In step S5.4, the system extracts the reference response arrival order, reference response amplitude ratio, and reference response detection period difference corresponding to each candidate fault source region.

[0065] In step S5.5, the arrival order of the reference response, the amplitude ratio of the reference response, and the detection period difference of the reference response are combined to form the first... r Gas migration response fingerprints of candidate fault source regions: (10) In the formula, where, For the first Gas migration response fingerprints of candidate fault source regions; For the first The reference response arrival order vector of each candidate fault source region; For the first Reference response amplitude scaling vector for each candidate fault source region; For the first Reference response detection period difference vector for each candidate fault source region; This is the sequence number of the candidate fault source region.

[0066] For the cooling condition switching event of "single pump operation switching to dual pump operation", this embodiment establishes the following ethylene response fingerprint table 2.

[0067] Table 2 Ethylene Response Fingerprint

[0068] In step S5.6, the system uses the measured differential response feature vector Gas migration response fingerprints of each candidate fault source region A comparison is then made. In step S5.7, the differences in response arrival order, response amplitude ratio, and response detection period between the measured response and each candidate fault source region are calculated.

[0069] In step S5.8, the system calculates the first difference based on the above-mentioned differences. Differences among candidate fault source regions: (11) in, For the measured differential response and the first The degree of difference between candidate fault source regions; A function is calculated to respond to the difference in arrival order; A function to calculate the difference in response amplitude proportions; A function to calculate the difference in response detection cycles; , , These are the weighting coefficients corresponding to the order of response arrival, the proportion of response amplitude, and the difference in response detection period, respectively.

[0070] In step S5.9, the system determines the degree of difference between each candidate fault source region. Calculate the reliability of the location: (12) in, For the first The reliability of locating candidate fault source regions; This represents the total number of candidate fault source regions. The traversal sequence number of the candidate fault source region; It is a natural exponential function.

[0071] In step S5.10, the system determines the candidate fault source region with the highest location confidence as the target fault source region: (13) in, This is the number of the target fault source area.

[0072] Based on the measured differential response data and response fingerprint table in this embodiment, the following positioning calculation results are obtained (Table 3).

[0073] Table 3. Positioning Calculation Results

[0074] The results above show that region Z1 has the highest reliability in locating the fault, significantly higher than other candidate regions. Therefore, the system outputs the target fault source region as Z1, namely the upper part of the high-voltage winding and the lead joint area.

[0075] Preferably, in order to avoid the difference in oil sampling branches affecting the order of response arrival, this embodiment also performs oil sample acquisition link delay compensation after step S4.

[0076] In step S6.1, the system obtains the first... The equivalent dead volume of the oil sampling branch corresponding to each oil flow correlation monitoring location Sampling flow rate Valve group switching stabilization time Oil intake branch flushing correction time .

[0077] In step S6.2, the system calculates the first... Sampling link delay time corresponding to each oil flow associated monitoring location: (14) in, For the first Sampling link delay time corresponding to each oil flow associated monitoring location; For the first The equivalent dead volume of the oil sampling branch at each monitoring location; For the first Sampling flow rate at each monitoring location; For the first Valve group switching stabilization time corresponding to each monitoring location; For the first The flushing correction time for the oil intake branch corresponding to each monitoring location.

[0078] In this embodiment, the parameters of the oil intake branches corresponding to P1 to P4 are shown in Table 4 below.

[0079] Table 4. Parameters of oil intake branches corresponding to P1 to P4

[0080] According to formula (14), the delay time of link P1 is 3.0 min, the delay time of link P2 is 4.2 min, the delay time of link P3 is 5.8 min, and the delay time of link P4 is 6.7 min.

[0081] In step S6.3, the system acquires the detected gas response time. ,in For the first Characteristic gases in the first The measured response time at each monitoring location. In step S6.4, the system will record the gas response time. Deducting sampling link delay time The response time after compensation: (15) in, The gas response time is the compensated value.

[0082] In step S6.5, the system calculates the compensation response time. Correcting the difference in response arrival order and response detection cycle. This process reduces spurious response lags caused by factors such as oil sampling branch length, equivalent dead volume, and valve assembly flushing status, thus improving the differential response eigenvector. It more closely approximates the actual gas migration patterns inside a transformer.

[0083] The final diagnostic results of this embodiment are as follows: The target fault source area is the upper part of the high voltage winding and the lead joint area. The location confidence is 0.86. The main responding gas is ethylene. The location is based on the fact that after the cooling condition is switched, P2 responds first, followed by P1, and P3 and P4 lag behind. The amplitude ratios of P2 / P1 and P3 / P1 conform to the gas migration response fingerprint of the Z1 region.

[0084] The technical effect of this embodiment is that: by switching cooling conditions to form an identifiable oil flow excitation, and by combining multi-location oil dissolved gas differential response, background trend stripping, gas migration response fingerprint matching and sampling link delay compensation, the response timing difference and amplitude difference that are difficult to utilize by fixed oil sampling or single-point DGA are transformed into the basis for fault source area location, thereby narrowing the investigation scope from the entire transformer interior to the upper part of the high-voltage winding and lead joint area.

[0085] Example 2: Latent Fault Source Location and Diagnosis System Based on Cooling Condition Switching Response Differentiation like Figure 8 As shown, this embodiment provides a latent fault source location and diagnosis system 100 for oil-immersed transformers based on differential cooling condition switching response. This system is used to execute the method described in Embodiment 1.

[0086] System 100 includes a data acquisition module 110, a switching event recognition module 120, a differential response extraction module 130, a response fingerprint storage module 140, a fault source location module 150, and a result output module 160.

[0087] The data acquisition module 110 is used to acquire dissolved gas data in the oil and cooling condition data of the oil-immersed transformer 200 at multiple oil flow-related monitoring locations. The data acquisition module 110 includes a multi-location oil sampling unit 111, a dissolved gas detection unit 112, and a cooling condition status acquisition unit 113.

[0088] The multi-position oil intake unit 111 includes multiple oil intake branches 250, a switching valve group 260, a micro-circulation pump, a bypass flushing branch, and a return oil branch 270. The multiple oil intake branches 250 are respectively connected to at least two of the following locations: the upper oil area of ​​the oil tank, the side oil area of ​​the winding, the lower oil area of ​​the oil tank, the inlet oil line of the cooler, and the outlet oil line of the cooler.

[0089] The switching valve assembly 260 is used to select different oil sampling branches 250 to connect to the dissolved gas in oil detection unit 112. The bypass flushing branch is used to flush the corresponding oil sampling branch 250 before sampling to reduce the influence of residual oil sample from the previous monitoring location on the current detection result. The return oil branch 270 is used to return the detected oil sample to the return oil position of the oil-immersed transformer 200.

[0090] The dissolved gas detection unit 112 in oil is used to detect the concentration of at least two of the following gases: hydrogen, methane, ethane, ethylene, acetylene, carbon monoxide, and carbon dioxide.

[0091] The cooling condition status acquisition unit 113 is connected to the submersible pump control circuit, cooler control circuit, fan control circuit or transformer monitoring system to obtain the operating status and switching time of the submersible pump, cooler group, fan group and cooling oil circuit valve group.

[0092] The switching event identification module 120 is used to identify cooling condition switching events based on cooling condition data, and to determine the background time period before switching and the response time period after switching.

[0093] The differential response extraction module 130 is used to establish a gas background evolution model based on the dissolved gas data in the oil during the background period before the switch, and to calculate the gas differential response across operating conditions based on the gas background evolution model and the actual dissolved gas data in the oil during the response period after the switch.

[0094] The response fingerprint storage module 140 is used to store gas migration response fingerprints of candidate fault source regions. The gas migration response fingerprint includes the arrival order of reference responses, the amplitude ratio of reference responses, and the difference in reference response detection periods for each candidate fault source region under the corresponding cooling condition switching event.

[0095] The fault source localization module 150 is used to match the differential response feature vector corresponding to the gas differential response across operating conditions with the gas migration response fingerprint of the candidate fault source region, calculate the difference degree and the localization confidence, and output the target fault source region.

[0096] The result output module 160 is used to output the target fault source area and the corresponding location confidence level. In one specific output mode, the result output module 160 outputs the following: the target fault source area is the upper part of the high-voltage winding and the lead joint area; the location confidence level is 86%; the main responding gas is ethylene; the location basis is that after the cooling condition is switched, P2 responds first, followed by P1, and P3 and P4 lag behind, and the response amplitude ratio conforms to the gas migration response fingerprint of the Z1 region; the diagnostic suggestion is to shorten the oil chromatography monitoring cycle and focus on checking the upper part of the high-voltage winding and the lead joint area.

[0097] The system structure of this embodiment corresponds to the method steps of Embodiment 1. Specifically, the data acquisition module 110 corresponds to step S1, the switching event recognition module 120 corresponds to step S2, the differential response extraction module 130 corresponds to steps S3 and S4, the response fingerprint storage module 140 corresponds to steps S5.1 to S5.5, the fault source location module 150 corresponds to steps S5.6 to S5.10, and the result output module 160 corresponds to the output process of the target fault source area and the location confidence.

Claims

1. A method for locating and diagnosing latent fault sources in oil-immersed transformers based on differential response to cooling condition switching, characterized in that, Includes the following steps: S1. Obtain dissolved gas data and cooling condition data in the oil of the oil-immersed transformer at multiple oil flow-related monitoring locations, and record the data as follows: Characteristic gases in the first The oil flow correlation monitoring location is at the first The concentration for each detection cycle is ,in, , This represents the total number of characteristic gas species. , This represents the total number of oil flow-related monitoring locations. S2. Identify cooling condition switching events based on the cooling condition data, and record the detection period corresponding to the cooling condition switching event as . And according to the detection cycle Determine the time period for switching between foreground and background and the response period after the switch ,in This represents the number of detection cycles included in the background time period. The number of detection cycles included in the response period; S3, according to the aforementioned background / pre-switching time period A gas background evolution model was established based on dissolved gas data in the oil, and the first... Characteristic gases in the first Background predicted concentration at each oil flow-related monitoring location ; S4, switch the response period The actual dissolved gas data in the oil is compared with the background predicted concentration to obtain the cross-condition gas differential response. And based on the cross-condition gas differential response Constructing differential response feature vectors ; S5. The differential response feature vector Compared with the pre-established gas migration response fingerprint of the candidate fault source region Matching is performed to obtain the location confidence of each candidate fault source region. The candidate fault source region with the highest location confidence was identified as the target fault source region. This is the sequence number of the candidate fault source region.

2. The method for locating and diagnosing latent fault sources in oil-immersed transformers based on differential cooling condition switching response as described in claim 1, characterized in that: In step S1, the process of determining the multiple oil flow associated monitoring locations includes the following sub-steps: S1.

1. Based on the tank structure, winding arrangement, core position, main oil passage arrangement and cooling oil passage connection of the oil-immersed transformer, determine the candidate oil sampling locations that can reflect the gas migration state in different oil flow areas. S1.2 Select at least two oil flow associated monitoring locations from the candidate oil sampling locations. The oil flow associated monitoring locations include at least two of the following: the oil area monitoring location at the top of the oil tank, the oil area monitoring location at the side of the winding, the oil area monitoring location at the bottom of the oil tank, the oil circuit monitoring location at the cooler inlet, and the oil circuit monitoring location at the cooler outlet. S1.

3. Collect oil samples from the oil flow correlation monitoring locations through multiple oil sampling branches, and obtain the corresponding dissolved gas data in the oil through the dissolved gas detection device in the oil. S1.

4. Number the dissolved gas data in the oil at different oil flow-related monitoring locations according to the detection cycle to form a multi-location gas concentration sequence. .

3. The method for locating and diagnosing latent fault sources in oil-immersed transformers based on differential cooling condition switching response as described in claim 1, characterized in that, In step S2, the process of identifying the cooling condition switching event includes the following sub-steps: S2.1 Collect the operating status of at least one of the following cooling equipment: submersible pump, cooler group, fan group, and cooling oil circuit valve group; S2.2 Determine whether the operating status of the cooling equipment has changed; S2.3 When at least one of the following events is detected: submersible pump activation, submersible pump deactivation, submersible pump group switching, cooler group activation, cooler group deactivation, cooler group switching, fan group activation, fan group deactivation, cooling oil circuit valve group status change, or switching between natural oil circulation and forced oil circulation, a cooling condition switching event is determined to have occurred. S2.4 Record the detection cycle corresponding to the cooling condition switching event. and with the detection cycle Determine the background time period before switching based on the benchmark. and the response period after the switch .

4. The method for locating and diagnosing latent fault sources in oil-immersed transformers based on differential cooling condition switching response as described in claim 1, characterized in that, In step S3, the process of establishing the gas background evolution model includes the following sub-steps: S3.1, from switching the foreground and background time periods Internal extraction Characteristic gases in the first Historical concentration sequences of oil flow-related monitoring locations; S3.2 Obtain the transformer load corresponding to the historical concentration sequence. and top oil temperature ,in For the first Transformer load for each testing cycle For the first The top oil temperature of the transformer during each testing cycle; S3.3, Based on the aforementioned historical concentration sequence and transformer load and top oil temperature The model parameters of the gas background evolution model are obtained by fitting. S3.4 Predict the response period after the handover based on the model parameters. Background predicted concentration within The gas background evolution model is as follows: ; In the formula, For the first Characteristic gases in the first The trend coefficient of each monitoring location changing with the detection cycle; For the first Characteristic gases in the first Background concentration constant term for each monitoring location; For the first Characteristic gases in the first Load impact coefficient at each monitoring location; For the first Characteristic gases in the first The oil temperature influence coefficient at each monitoring location.

5. The method for locating and diagnosing latent fault sources in oil-immersed transformers based on differential cooling condition switching response as described in claim 1, characterized in that, In step S4, the extraction process of the gas differential response across operating conditions includes the following sub-steps: S4.1, Read the response period after the switch Inner Characteristic gases in the first Actual concentration at each oil flow-related monitoring location ; S4.2 Read the background predicted concentration for the corresponding detection period. ; S4.3, the actual concentration Compared with the background predicted concentration By performing a differential operation, the gas differential response across operating conditions is obtained: ; S4.4, Based on the aforementioned cross-condition gas differential response Extract the response start detection period, response amplitude, response arrival order, and response detection period difference.

6. The method for locating and diagnosing latent fault sources in oil-immersed transformers based on differential cooling condition switching response as described in claim 5, characterized in that, In step S4.4, the process of determining the initial detection period and response amplitude includes the following sub-steps: S4.41, Based on the time period before and after switching background. The fluctuation range of gas concentration within the range is used to determine the first Characteristic gases in the first Response determination threshold for each oil flow-related monitoring location ; S4.42, Response period after handover Within, determine the gas differential response across operating conditions. Does the absolute value exceed the response determination threshold? ; S4.43, The response threshold will be exceeded for the first time. The detection cycle is determined as the response start detection cycle. ,in: ; S4.44, Response period after handover Within this range, the maximum absolute value of the gas differential response across operating conditions is determined and used as the response amplitude. ,in: .

7. The method for locating and diagnosing latent fault sources in oil-immersed transformers based on differential cooling condition switching response as described in claim 1, characterized in that, In step S4, the differential response feature vector The construction process includes the following sub-steps: S4.

5. Determine the response arrival order vector based on the response start detection period corresponding to multiple oil flow associated monitoring locations. ; S4.

6. Determine the response amplitude ratio vector based on the response amplitudes corresponding to multiple oil flow correlation monitoring locations. ; S4.7 Determine the response detection cycle difference vector based on the response start detection cycle difference between different oil flow associated monitoring locations. ; S4.8, The response arrival order vector Response amplitude scaling vector and response detection period difference vector Combining these elements, we obtain the difference response feature vector: .

8. The method for locating and diagnosing latent fault sources in oil-immersed transformers based on differential cooling condition switching response as described in claim 1, characterized in that, The gas migration response fingerprint of the candidate fault source region The establishment process includes the following sub-steps: S5.1 Based on the tank structure, winding position, core position, oil passage layout, cooler inlet position, cooler outlet position, and oil intake branch position of the oil-immersed transformer, divide multiple candidate fault source areas. S5.2 Set an equivalent gas release source in each candidate fault source region; S5.3 For at least one cooling condition switching event, simulate or calibrate the process of the fault gas released by the equivalent gas release source migrating to multiple oil flow associated monitoring locations; S5.4 Extract the reference response arrival order, reference response amplitude ratio, and reference response detection cycle difference for each candidate fault source region; S5.5, Combine the arrival order of the reference responses, the amplitude ratio of the reference responses, and the difference in the detection period of the reference responses to form the first... Gas migration response fingerprints of candidate fault source regions: ; In the formula, For the first The reference response arrival order vector of each candidate fault source region; For the first Reference response amplitude scaling vector for each candidate fault source region; For the first Reference response detection period difference vector for each candidate fault source region; The candidate fault source regions include at least two of the following: the upper high-voltage winding and lead joint region, the middle high-voltage winding region, the lower high-voltage winding region, the core and clamping parts region, the low-voltage winding region, and the cooler inlet associated region.

9. The method for locating and diagnosing latent fault sources in oil-immersed transformers based on differential cooling condition switching response as described in claim 8, characterized in that, Step S5, the process of determining the target fault source region includes the following sub-steps: S5.6, The measured differential response eigenvector Gas migration response fingerprints of each candidate fault source region Compare; S5.7 Calculate the differences in response arrival order, response amplitude ratio, and response detection cycle between the measured response and each candidate fault source region; S5.8 Calculate the first [response] based on the differences in response arrival order, response amplitude ratio, and response detection period. Differences among candidate fault source regions ,in: ; In the formula, A function is calculated to respond to the difference in arrival order; A function to calculate the difference in response amplitude proportions; A function to calculate the difference in response detection cycle time; , , These are the weighting coefficients corresponding to the order of response arrival, the proportion of response amplitude, and the difference in response detection period, respectively. S5.9, Based on the differences in the regions of each candidate fault source Calculate the reliability of the location ,in: ; In the formula, This represents the total number of candidate fault source regions. The traversal sequence number of the candidate fault source region; It is a natural exponential function; S5.

10. Determine the candidate fault source region with the highest location confidence as the target fault source region. ,in: .

10. The method for locating and diagnosing latent fault sources in oil-immersed transformers based on differential cooling condition switching response as described in claim 1, characterized in that, Following step S4, the process also includes delay compensation for the oil sample acquisition link, specifically including the following sub-steps: S6.1, Obtain the first The equivalent dead volume of the oil sampling branch corresponding to each oil flow correlation monitoring location Sampling flow rate Valve group switching stabilization time Oil intake branch flushing correction time ; S6.2, based on the equivalent dead volume Sampling flow rate Valve group switching stabilization time Oil intake branch flushing correction time Calculate the first Sampling link delay time corresponding to each oil flow-related monitoring location ,in: ; S6.3 Obtain the gas response time obtained from the detection ,in For the first Characteristic gases in the first Measured response time at each monitoring location; S6.4, the gas response time Subtract the sampling link delay time Response time after compensation ,in: ; S6.5, Based on the compensated response time Correct the difference between the response arrival order and the response detection cycle.