Transformer winding deformation diagnosis method and system based on credibility and consistency check

Abstract

The present application relates to the technical field of power transformer winding deformation detection, and particularly relates to a transformer winding deformation diagnosis method and system based on credibility and consistency verification, taking phase or phase group as an evaluation unit, performing consistency verification on the test configuration of current test data and review test data to form an effective test data pair, calculating the test credibility of frequency response and short-circuit impedance and determining the main credibility abnormal contribution item. The data reaching the threshold value is compared with the reference data to obtain the frequency response abnormal score and impedance abnormal score, and the substandard data does not participate in fusion and generates a recovery-oriented retest. When both types of tests are credible, the cross-test result difference is calculated, the winding deformation level is fused and output when the consistency threshold is met, otherwise a verification-oriented retest is generated. The retest data is updated after the credibility is confirmed, and the diagnosis result is output. The present application realizes traceable and terminable closed-loop diagnosis through configuration verification, credibility gating and differential retest.

Description

Technical Field

[0001] This invention relates to the field of power transformer winding deformation detection technology, and in particular to a method and system for diagnosing transformer winding deformation based on reliability and consistency verification. Background Technology

[0002] After a power transformer is in operation, during transportation, or subjected to a sudden short circuit in the vicinity, the windings may be subjected to the electrodynamic force generated by the short circuit current or external mechanical forces, resulting in localized twisting, bulging, displacement, loosening, or other permanent deformation. Winding deformation affects the transformer's insulation condition, mechanical strength, and operational safety. Therefore, it is necessary to inspect the winding deformation condition at the factory, before handover, before commissioning, after transportation, after a short circuit, or during maintenance.

[0003] Frequency response analysis involves injecting a sweep frequency signal into the transformer windings and acquiring the response signal to obtain the amplitude-frequency response curves of the windings in different frequency bands. The differences between the current curve and historical curves, phase-to-phase curves, or similar curves are then used to determine whether the windings have deformed. Short-circuit impedance analysis measures parameters such as transformer short-circuit impedance, resistance, reactance, power factor, three-phase deviation, and inductance. The current test values ​​are compared with factory values, acceptance values, initial field test values, or historical test values ​​to reflect whether the winding structure or electrical parameters have changed.

[0004] Existing comprehensive diagnostic methods for winding deformation can make a comprehensive judgment by utilizing frequency response characteristics, impedance parameters, and lateral or longitudinal comparison results. However, the frequency response curve is easily affected by the test lead arrangement, grounding status, clamp contact, and high-frequency interference in the field. Short-circuit impedance testing may also be affected by changes in test voltage, test current, contact status, and test mode. Therefore, the data differences formed by the same winding under different test conditions do not necessarily correspond to the actual changes in the winding structure.

[0005] Even when the same winding is repeatedly tested, differences in test results may still arise from the test configuration itself if the phase under test, tap position, winding connection, wiring method, sweep frequency parameters, or impedance test mode are inconsistent between the two tests. Existing methods typically lack constraints on the comparability of test configurations before repeatability evaluation, and even if an overall test reliability evaluation result is obtained, it is difficult to further determine whether the reliability reduction is mainly caused by factors such as high-frequency non-repetitive disturbances, peak position repetitive drift, amplitude repetitive deviation, test voltage fluctuations, test current fluctuations, or abnormal phase stability.

[0006] On the other hand, when both frequency response and short-circuit impedance test data possess high repeatability, but the two types of tests still inconsistently judge the anomalies in winding condition, it is difficult to determine which test type, phase, or local frequency band the conflict mainly originates from using only the overall anomaly level, fixed threshold, or preset fusion rules. Therefore, it is usually necessary to manually conduct a large range of repeated tests. Thus, it is necessary to establish a closed-loop diagnostic mechanism that can separate test quality anomalies from winding condition anomalies while ensuring the comparability of current test data and verification test data, and further pinpoint the main factors causing reduced test reliability or cross-test result conflicts, thereby generating targeted retest tasks.

[0007] In frequency response repeatability evaluation, if the reliability of the test is evaluated based solely on the high-frequency local fluctuations of a single test curve, it may confuse the repeatable high-frequency response characteristics of the winding with the non-repeatable test disturbances. Therefore, it is necessary to evaluate the test stability by utilizing the high-frequency non-repeatable components between the current test and the verification test with the same configuration. Summary of the Invention

[0008] In view of this, the purpose of this invention is to propose a transformer winding deformation diagnosis method and system based on credibility and consistency verification, so as to solve the problems of lack of constraints on the comparability of current test data and verification test data, easy confusion between test credibility abnormalities and winding state abnormalities, low credibility data may be involved in comprehensive judgment, and lack of targeted retesting mechanism when two types of credibility tests conflict.

[0009] To achieve the above objectives, this invention provides a method for diagnosing transformer winding deformation based on reliability and consistency verification, comprising the following steps:

[0010] Obtain basic information and historical baseline data of the transformer under test;

[0011] Using the phase or phase group p of the winding under test as the evaluation unit, the current test data and the verification test data are obtained. The verification test data refers to the same configuration repeated test data used to evaluate the repeatability stability of the current test data. The test data includes frequency response test data and short-circuit impedance test data.

[0012] The current test data and the review test data are checked for consistency in test configuration, and only test data that meet the preset comparability conditions are formed into valid test data pairs;

[0013] Based on the effective test data pairs, calculate the frequency response test reliability CF(p) and impedance test reliability CZ(p) for each phase or phase group p, and determine the corresponding main reliability anomaly contribution items according to the contribution of each reliability component index to the reliability reduction.

[0014] CF(p) is compared with the frequency response confidence threshold τF, and CZ(p) is compared with the impedance confidence threshold τZ to determine whether the confidence condition is met. For frequency response test data that meets the corresponding confidence condition, it is compared with historical reference data or phase-to-phase reference data that meets the structural comparability condition to obtain the frequency response anomaly score AF(p). For short-circuit impedance test data that meets the corresponding confidence condition, it is compared with historical reference data or phase-to-phase reference data that meets the structural comparability condition to obtain the impedance anomaly score AZ(p). Test data that is determined not to meet the corresponding confidence threshold does not participate in the final dynamic fusion, and a restorative directional retest task is generated based on its main confidence anomaly contribution item.

[0015] When both CF(p) and CZ(p) satisfy the corresponding credibility conditions, calculate the cross-test result difference DFZ(p) based on AF(p) and AZ(p). When DFZ(p) is not greater than the consistency threshold δ, dynamically fuse the credibility of AF(p) and AZ(p) according to CF(p) and CZ(p).

[0016] When both CF(p) and CZ(p) satisfy the corresponding confidence conditions but DFZ(p) is greater than δ, the main state anomaly contribution items in AF(p) and AZ(p) are determined respectively, and a verification-oriented retest task is generated.

[0017] For retest data obtained from restorative or confirmatory retesting tasks, the credibility is reconfirmed, and AF(p) or AZ(p) is updated only after the corresponding credibility conditions are met.

[0018] Based on the reconfirmed credibility, abnormal status score, and consistency across test results, the final diagnostic result is output. The diagnostic result includes at least one of the following: winding deformation status level, abnormality prompt, or manual review instruction.

[0019] Preferably, the test configuration consistency verification includes discrete configuration field consistency verification and continuous test parameter consistency verification;

[0020] The discrete configuration fields for frequency response testing should include at least the winding under test, phase or phase group, tap position, winding connection method and wiring method, and the continuous test parameters should include at least the sweep range and frequency sampling settings.

[0021] The discrete configuration fields for short-circuit impedance testing include at least the winding under test, phase or phase group, tap position, winding connection method, wiring method, and test mode. The continuous test parameters include at least the test voltage and test current conditions.

[0022] When the discrete configuration fields are the same, and the continuous test parameters are the same or their difference does not exceed the preset allowable configuration deviation, the current test data and the verification test data are determined to meet the comparability conditions.

[0023] Preferably, the frequency response test confidence CF(p) is calculated in the following manner:

[0024] CF(p)=λ1ρrep(p)+λ2[1-Dpeak(p)]+λ3[1-Damp(p)]+λ4[1-NHF(p)];

[0025] Wherein, ρrep(p) is the repetitive correlation between the current frequency response test data and the frequency response verification test data with the same configuration, Dpeak(p) is the normalized repetitive drift of the corresponding resonance peak, Damp(p) is the normalized repetitive deviation of the corresponding peak and valley amplitude, NHF(p) is the high-frequency non-repetitiveness index between the current frequency response test data and the frequency response verification test data with the same configuration, and λ1 to λ4 are non-negative weights with a sum of 1.

[0026] The impedance test reliability CZ(p) is calculated as follows:

[0027] CZ(p)=μ1[1-DZrep(p)]+μ2[1-DU(p)]+μ3[1-DI(p)]+μ4[1-Dφ(p)];

[0028] Where DZrep(p) is the relative deviation of the repeating impedance between the current impedance test data and the verification test data of the same configuration impedance, DU(p) is the test voltage consistency deviation, DI(p) is the test current stability deviation, Dφ(p) is the phase or power factor fluctuation, and μ1 to μ4 are non-negative weights with a sum of 1.

[0029] Preferably, the main confidence anomaly contribution items include main frequency response confidence anomaly contribution items and main impedance confidence anomaly contribution items, which are obtained in the following ways:

[0030] Construct the frequency response confidence anomaly contribution vector QF(p) from the components of CF(p):

[0031] QF(p)=[λ1(1-ρrep(p)), λ2Dpeak(p), λ3Damp(p), λ4NHF(p)];

[0032] The component with the largest value in QF(p) is determined as the main frequency response confidence anomaly contribution item;

[0033] Construct the impedance confidence anomaly contribution vector QZ(p) from the components of CZ(p):

[0034] QZ(p)=[μ1DZrep(p),μ2DU(p),μ3DI(p),μ4Dφ(p)];

[0035] The component with the largest value in QZ(p) is determined as the main impedance reliability anomaly contribution item.

[0036] Preferably, the restorative targeted retest is generated based on the main confidence anomaly contribution items, wherein:

[0037] When the frequency response test confidence level is lower than the frequency response confidence level threshold, and λ1[1-ρrep(p)] is the main contributor, the repetition correlation between the current test and the review test in the low frequency band, mid frequency band and high frequency band are compared respectively. The frequency band with the lowest repetition correlation is determined as the key retest frequency band, and the frequency response recovery retest task of the corresponding phase or phase group is generated.

[0038] When the frequency response test reliability is lower than the frequency response reliability threshold, and λ4NHF(p) is the main contributing factor, the interval where the high-frequency non-repeating component is located is determined as the key retest frequency band, a frequency response recovery retest task is generated, and the inspection items of grounding status, test lead layout and wiring contact status are output.

[0039] When the frequency response test confidence level is lower than the frequency response confidence level threshold, and λ2Dpeak(p) or λ3Damp(p) is the main contributor, the frequency bands with repeated drift of the main peak position, repeated deviation of peak and valley amplitude, or corresponding feature mismatch will be identified as key retest frequency bands, generating a frequency response recovery retest task, and outputting test lead position and wiring status inspection items.

[0040] When the impedance test confidence level is lower than the impedance confidence level threshold and μ1DZrep(p) is the main contributor, a short-circuit impedance recovery retest task is generated, and the wiring contact status, tap position and short-circuit impedance test mode consistency check items are output.

[0041] When the impedance test reliability is lower than the impedance reliability threshold, and μ2DU(p), μ3DI(p), or μ4Dφ(p) are the main contributors, a corresponding short-circuit impedance recovery retest task is generated, and the corresponding test voltage stability, current loop and contact status or phase sampling and wiring status inspection items are output.

[0042] Preferably, when using inter-phase reference data or same three-phase data for lateral state comparison, the structural comparability of the objects to be compared is first verified;

[0043] The fields for structural comparability verification include at least winding type, winding connection method, tap position, test side, corresponding test configuration, and winding structure.

[0044] Only when the field meets the preset structural comparability conditions, the horizontal state comparison is performed using inter-phase reference data or three-phase data of the same time; if the structural comparability conditions are not met, the vertical state comparison is performed using the historical reference data of the phase or phase group itself.

[0045] Preferably, AF(p) and AZ(p) are calculated by weighting the corresponding state anomaly features after a unified 0-1 anomaly severity mapping, where 0 corresponds to a state with no obvious anomaly and 1 corresponds to a preset high anomaly degree state, and are calculated according to:

[0046] DFZ(p) = |AF(p) - AZ(p)| calculates the difference between test results;

[0047] When CF(p)≥τF, CZ(p)≥τZ and DFZ(p)≤δ, the fusion score S(p) is calculated according to: S(p)=[CF(p)×AF(p)+CZ(p)×AZ(p)] / [CF(p)+CZ(p)].

[0048] Under the condition 0≤T1<T2<T3≤1, the output is normal when 0≤S(p)<T1, slightly distorted when T1≤S(p)<T2, obviously distorted when T2≤S(p)<T3, and severely distorted when T3≤S(p)≤1.

[0049] Preferably, when CF(p)≥τF, CZ(p)≥τZ but DFZ(p)>δ, the weighted contribution of each normalized state anomaly feature constituting AF(p) is defined as PF,i(p)=αiai(p), and the weighted contribution of each normalized state anomaly feature constituting AZ(p) is defined as PZ,j(p)=βjaj(p), where αi is the non-negative frequency response state anomaly weight and the sum of all αi is 1, βj is the non-negative impedance state anomaly weight and the sum of all βj is 1, AF(p) is equal to the sum of each PF,i(p), and AZ(p) is equal to the sum of each PZ,j(p);

[0050] The largest contributing terms in PF,i(p) and PZ,j(p) are respectively identified as the main state anomaly contributing terms of the corresponding test side, and the test type, frequency band or parameters of the verification retest are determined by combining the phase or phase group p and the main state anomaly contributing terms.

[0051] Preferably, for restorative targeted retesting, the original test data that is determined not to meet the credibility conditions is retained for process traceability and marked as not participating in the final fusion; after correcting the corresponding test conditions, the test data is reacquired, and the corresponding verification test data is acquired under the same test configuration to re-form a valid test data pair, and the corresponding test credibility is recalculated based on the re-formed valid test data pair;

[0052] For frequency response recovery retests, AF(p) is calculated using the corresponding reliable frequency response data only if the recalculated CF(p) is not lower than the frequency response reliability threshold; for short-circuit impedance recovery retests, AZ(p) is calculated using the corresponding reliable impedance data only if the recalculated CZ(p) is not lower than the impedance reliability threshold; test paths that do not meet the corresponding reliability thresholds do not calculate AF(p) or AZ(p), and recovery directional retests continue to be performed.

[0053] For confirmatory targeted retests, CF(p) and CZ(p) data that already meet the credibility conditions are not marked as untrustworthy data; before the confirmatory retest data is used to update AF(p) or AZ(p), it is first confirmed that the confirmatory retest data meets the corresponding credibility conditions.

[0054] If DFZ(p) is not greater than the consistency threshold after confirmatory retesting, then fusion is performed; if the two types of tests still meet the credibility condition after confirmatory retesting but DFZ(p) is still greater than the consistency threshold, then fusion is not performed, and at least one of the following is output: continuous cross-test result conflict, suspected local anomaly, or manual review prompt.

[0055] When the automatic retesting count for the same phase or phase group, the same test type, and the same retesting reason reaches the maximum retesting count Nmax and still cannot meet the corresponding termination condition, the automatic retesting will stop and the process will be transferred to manual review.

[0056] The present invention also provides a transformer winding deformation diagnosis system based on reliability and consistency verification, including a measurement host and a computer analysis unit that is communicatively connected to the measurement host;

[0057] The measurement host includes a frequency response test channel, a short-circuit impedance test channel, a synchronous sampling and processing unit, a test control unit, and an isolation protection unit;

[0058] The computer analysis unit includes a test configuration consistency verification module, a current and review test data management module, a frequency response test reliability evaluation module, an impedance test reliability evaluation module, a reliability anomaly contribution location module, a frequency response anomaly quantification module, an impedance anomaly quantification module, a result consistency verification module, a state anomaly contribution location module, a dynamic fusion module, a directional retest decision module, a database management module, and a report generation module.

[0059] The test configuration consistency verification module is used to verify the corresponding test configuration before the current test data and the review test data form a valid test data pair, and to verify the structural comparability of the objects to be compared before using inter-phase reference data or same three-phase data for lateral state comparison.

[0060] The frequency response test reliability evaluation module calculates NHF(p) using the high-frequency non-repetitive components between the current frequency response test data and the frequency response test data with the same configuration;

[0061] The credibility anomaly contribution localization module is used to determine the main test factors that cause a decrease in the credibility of frequency response tests or impedance tests.

[0062] The state anomaly contribution localization module is used to determine the main state anomaly contribution items when both types of tests are reliable but the cross-test results do not meet the consistency condition.

[0063] The dynamic fusion module generates a fusion result only when both types of tests meet the corresponding credibility threshold and the difference between the cross-test results is no greater than the consistency threshold.

[0064] The directional retest decision module generates retest task parameters and manual inspection items, which involve manual inspection items such as the winding to be tested, phase or phase group, tap position, winding connection method, wiring method, grounding status or test lead arrangement, and are output through the test task interface.

[0065] After manual confirmation, the software-controlled retest task parameters are sent to the test control unit, which then calls the frequency response test channel or the short-circuit impedance test channel to execute the corresponding test.

[0066] The data obtained from the retest is returned to the computer analysis unit. The restorative retest data is retested, the consistency of the configuration is checked, and the valid test data pairs are reconstructed. The corresponding test confidence is recalculated, and AF(p) or AZ(p) is updated only after the corresponding confidence condition is met. The confirmatory retest data calculates the confirmatory retest confidence according to the test type. The corresponding AF(p) or AZ(p) is updated only when the corresponding confirmatory retest confidence meets its threshold.

[0067] The report generation module is used to generate and output a diagnostic result containing the winding deformation status level or the manual review prompt based on the fusion result output by the dynamic fusion module or the manual review prompt output by the directional retest decision module. The diagnostic result includes at least one of the winding deformation status level, abnormal prompt, or manual review instruction.

[0068] The beneficial effects of this invention are:

[0069] 1. This invention uses test configuration consistency verification to make the current test data and the verification test data comparable, reducing the impact of test configuration differences on repeatability stability evaluation and ensuring the reliability of credibility evaluation results.

[0070] 2. This invention employs a tiered approach, separating test reliability evaluation from winding condition anomaly evaluation. The difference between current test data and verified test data is used for test reliability evaluation, while the difference between the reliability-confirmed test data and historical benchmark data or phase-to-phase benchmark data that meets structural comparability conditions is used for winding condition anomaly evaluation. This reduces the possibility of confusion between changes in test conditions and actual condition changes.

[0071] 3. This invention, through the location of confidence anomalies, can identify the main factors causing the decrease in test confidence and generate a restorative targeted retest task accordingly. This allows low-confidence tests to directly enter the corresponding test condition check and retest from the cause location, avoiding blindly repeating tests.

[0072] 4. This invention only performs fusion when both frequency response test and short-circuit impedance test meet the reliability condition and the cross-test results meet the consistency condition. Unreliable data is retained for traceability but does not participate in the final fusion, thereby ensuring the reliability of the comprehensive diagnostic results.

[0073] 5. When two types of tests are both reliable but the state evaluation results conflict, this invention determines the test type, phase or phase group, frequency band or parameter corresponding to the conflict by locating the state anomaly contribution, and triggers a confirmatory directional retest. The original reliable data remains reliable and is not automatically discarded as low-quality data due to the conflicting results.

[0074] 6. This invention distinguishes between local key frequency band retesting and complete confirmation frequency sweeping. Local frequency band retesting is used for targeted confirmation of the cause of anomalies. When it is necessary to update the global diagnostic results, complete confirmation frequency response data and complete verification frequency response data with the same configuration are reacquired and confidence gating is performed to ensure that the retesting results can enter the comprehensive diagnosis according to unified data rules.

[0075] 7. This invention forms a traceable and terminateable closed-loop diagnostic process through retest data association, historical benchmark management, and a maximum retest limit. When results remain unreliable or conflict across test results, the process automatically switches to manual review, avoiding infinite retesting loops. Attached Figure Description

[0076] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only for this invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0077] Figure 1 This is the overall flowchart of the present invention.

[0078] Figure 2This is a logic diagram for the credibility judgment, consistency verification, and retesting decision-making of this invention.

[0079] Figure 3 This is a schematic diagram of the measurement host panel and interface of the present invention, wherein (a) is a schematic diagram of the front panel and (b) is a schematic diagram of the back panel interface.

[0080] Figure 4 This is a schematic diagram of the frequency response test wiring for the present invention.

[0081] Figure 5 This is a schematic diagram of the short-circuit impedance test wiring of the present invention, wherein (a) is a schematic diagram of the internal low-voltage short-circuit impedance test wiring and project options, and (b) is a schematic diagram of the external input impedance test wiring.

[0082] Figure 6 This is a flowchart of the database management and report generation process of this invention. Detailed Implementation

[0083] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0084] It should be noted that, unless otherwise defined, the technical or scientific terms used in this invention should have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0085] Example 1:

[0086] like Figure 1 As shown, this embodiment provides a transformer winding deformation diagnosis method based on reliability and consistency verification. This method assumes comparability of test configurations, processes test reliability anomalies and winding state anomalies in a hierarchical manner, and sequentially performs test reliability gating, cross-test result consistency verification, anomaly contribution location, cause-driven differentiated retesting, effective data updating, and diagnostic conclusion updating.

[0087] In this manual, p represents the current phase or phase group under test. For a three-phase transformer, p can represent phase A, phase B, or phase C; for wiring configurations that require evaluation by phase group, p represents the corresponding test phase group. The frequency response test reliability CF(p), impedance test reliability CZ(p), frequency response anomaly score AF(p), impedance anomaly score AZ(p), frequency response reliability anomaly contribution vector QF(p), impedance reliability anomaly contribution vector QZ(p), cross-test result difference DFZ(p), as well as PF,i(p) and PZ,j(p) are all calculated and stored in association with the phase or phase group p.

[0088] First, obtain the basic information and historical baseline data of the transformer under test. Basic information includes the transformer name or equipment identifier, winding type, winding connection method, winding under test, tap position, rated capacity, and rated voltage. Historical baseline data includes at least one of the following: factory test data, acceptance test data, initial field test data, and verified valid historical test data. Historical baseline data is used for evaluating winding condition anomalies and should not be mixed with the verification test data used for this repeated stability evaluation.

[0089] In this embodiment, the verification test refers to a repeated test conducted under the condition of comparability of the corresponding test configuration to evaluate the repeatability stability of the current test data. Targeted retesting refers to a restorative or confirmatory targeted retesting triggered based on the main confidence anomaly contribution item or the main state anomaly contribution item after insufficient test confidence or a conflict between the evaluation results of two types of confidence test states. Verification tests are used to form data pairs for test confidence evaluation, while targeted retesting is triggered by the results of previous diagnostic tests; their triggering conditions and technical functions differ.

[0090] Using phase or phase group p as the evaluation unit, acquire current frequency response test data and frequency response verification test data, as well as current short-circuit impedance test data and short-circuit impedance verification test data. Frequency response testing can employ linear or logarithmic frequency sweep. Short-circuit impedance testing can employ internal low-voltage short-circuit impedance testing or external input impedance testing. When using external input impedance testing, the test current should not exceed 50A, the test voltage should not exceed 500V, and the voltage adjustment speed should not exceed 20V / second.

[0091] Before the current test data and the verification test data are included in the credibility calculation, a test configuration consistency check is first performed. For frequency response testing, the configuration fields such as the transformer under test, the winding under test, phase or phase group p, tap position, winding connection method, wiring method, sweep range, and frequency sampling settings are used to form the frequency response test configuration identifier GF. For short-circuit impedance testing, the configuration fields such as the transformer under test, the winding under test, the phase or phase group p, tap position, winding connection method, wiring method, internal low voltage or external input test mode, and test voltage and test current conditions are used to form the impedance test configuration identifier GZ.

[0092] Test configuration consistency verification includes discrete configuration field consistency verification and continuous test parameter consistency verification. Discrete configuration fields such as the winding under test, phase or phase group, tap position, winding connection method, wiring method, and test mode should, in principle, be consistent. Continuous test parameters such as sweep range, frequency sampling settings, test voltage, and test current conditions can be the same, or their differences can not exceed the preset allowable configuration deviation. When the above conditions are met, the current test data and the verification test data form a valid test data pair. Otherwise, the relevant data is marked as not directly comparable, and test data that meets the comparability conditions is reacquired.

[0093] The information required for the test configuration identifiers GF and GZ comes from the test parameters saved in the test software, database records, and wiring information confirmed by the operator through the test task interface. The above configuration verification and subsequent data processing are achieved using data collected by the measurement host, eliminating the need for additional field detection sensors.

[0094] This embodiment stratifies test reliability anomalies and winding state anomalies. ρrep(p), Dpeak(p), Damp(p), and NHF(p) in CF(p) are obtained by comparing the current frequency response test data with the same configuration verification frequency response test data, used to evaluate the repeatability stability of the current test process itself. NHF(p) characterizes high-frequency non-repetitive disturbances that cannot be stably repeated between two tests. Frequency band state anomalies, resonance peak migration, peak-valley amplitude changes, and local curve differences in AF(p) are obtained by comparing the reliability-confirmed test data with historical reference data or lateral reference data that meets structural comparability conditions, used to evaluate changes in winding structural state. Similarly, DZrep(p), DU(p), DI(p), and Dφ(p) in CZ(p) are used to evaluate the repeatability stability of the current and verification short-circuit impedance tests, while impedance error, three-phase impedance deviation, historical longitudinal change rate, inductance deviation, and reactance change in AZ(p) are used to evaluate changes in winding electrical state. The two types of indicators differ in their comparison objects, data sources, and technical functions, and do not double-weight the same physical differences.

[0095] For a phase or phase group p, the reliability of the frequency response test is calculated according to CF(p) = λ1ρrep(p) + λ2[1-Dpeak(p)] + λ3[1-Damp(p)] + λ4[1-NHF(p)], where λ1 to λ4 are non-negative weights and the sum of the weights is 1. ρrep(p), Dpeak(p), Damp(p), and NHF(p) are all normalized to 0 to 1. The larger CF(p) is, the higher the repeatability stability of the frequency response test for that phase or phase group.

[0096] In a preferred embodiment, the current frequency response curve and the review frequency response curve are first mapped to the same frequency sampling point, and the curve correlation coefficients ρLF, ρMF, and ρHF are calculated in the low-frequency, mid-frequency, and high-frequency bands, respectively. If the original correlation coefficient ranges from -1 to 1, the correlation coefficient of the b-th frequency band can be converted to the range of 0 to 1 using ρb,n = (ρb + 1) / 2. The frequency response repetition correlation can be calculated using ρrep = wLFρLF,n + wMFρMF,n + wHFρHF,n, where wLF, wMF, and wHF are non-negative weights and the sum of the weights is 1. Alternatively, the arithmetic mean of the correlation coefficients of the three normalized frequency bands can be used. Here, ρLF, ρMF, and ρHF are compared with the current frequency response data and the review frequency response data with the same configuration for repeatability evaluation, and are not used as anomalies in the frequency band status.

[0097] For the current frequency response curve and the same configuration verification frequency response curve, the corresponding resonant peaks are first identified and matched based on peak order, frequency proximity, and a preset matching frequency window. For the k-th matching resonant peak, dpeak,k is defined as |fcur,k-frev,k| / (|frev,k|+ε), where fcur,k is the frequency of the k-th matching resonant peak in the current test, frev,k is the frequency of the corresponding resonant peak in the verification test, and ε is a positive number used to avoid zero in the denominator. If no corresponding resonant peak is found within the preset matching frequency window, Mpeak,k is set to 1; otherwise, Mpeak,k is set to 0. A resonant peak mismatch contribution parameter cpeak,miss is set to be normalized to 0 to 1. The dpeak,k of all successfully matched peaks and the cpeak,miss × Mpeak,k of unmatched peaks are averaged or weighted averaged together, and then normalized to obtain Dpeak(p). When there is a complete mismatch, cpeak,miss preferably takes the normalized anomaly upper limit value of 1; however, this invention does not limit it to a unique fixed value. Correspondingly, for the amplitude of a matched peak or valley, the amplitude repetition deviation damp,k is defined as |Acur,k-Arev,k| / [0.5(|Acur,k|+|Arev,k|)+ε], where Acur,k and Arev,k are the amplitudes of the corresponding peak or valley in the current test and the verification test, respectively. If no corresponding peak or valley is found within the preset matching window, Mamp,k=1; otherwise, Mamp,k=0. A peak-valley feature mismatch contribution parameter camp,miss, normalized to 0 to 1, is set. The damp,k of successfully matched peaks and valleys and the camp,miss × Mamp,k of unmatched peaks and valleys are averaged or weighted, and then normalized to obtain Damp(p). During the above calculation, the frequency bands containing the unmatched resonance peaks and peak-valley features are recorded simultaneously, and the frequency band with the largest contribution is used as the candidate frequency band for restorative retesting. For complete mismatch, camp,miss can also take the normalized upper limit of 1. Dpeak(p) characterizes the repeated drift or mismatch of the resonance peak between the current test and the verification test, and Damp(p) characterizes the repeated deviation or characteristic mismatch of the amplitude between the current test and the verification test. Neither of them is equal to the amount of winding state change in AF(p) relative to the historical reference or the phase reference that meets the structural comparability conditions.

[0098] For the high-frequency band, the current frequency response curve ycur(f) and the same configuration-based refactored frequency response curve yrev(f) are first mapped to the same frequency sampling points, and the same smoothing process or local trend fitting is used to obtain the corresponding trend terms ŷcur(f) and ŷrev(f). Define rcur(f) = ycur(f) - ŷcur(f), rrev(f) = yrev(f) - ŷrev(f), and ΔrHF(f) = rcur(f) - rrev(f). The high-frequency non-repetitive raw quantity can be calculated according to NHF,raw(p) = √{(1 / NHF)Σ(q=1 to NHF)[rcur(fq) - rrev(fq)]²}, where q is the index of the high-frequency band aligned sampling point, fq is the q-th aligned frequency sampling point, and NHF is the number of high-frequency band aligned sampling points. Then, based on the normal high-frequency non-repetitive amplitude scale or the residual difference distribution of historical effective repeated tests, NHF,raw(p) is normalized to 0 to 1 to obtain NHF(p). The larger the NHF(p), the higher the degree of high-frequency non-repetitiveness between the current test and the verification test with the same configuration. The high-frequency resonant structure that appears stably and repeatedly in both tests will not increase NHF(p) simply because of the large local fluctuations of a single curve. This invention does not limit the specific smoothing algorithm.

[0099] To pinpoint specific testing factors for insufficient frequency response test reliability, the formula QF(p) = [λ1(1-ρrep(p)), λ2Dpeak(p), λ3Damp(p), λ4NHF(p)] is formed. The component with the largest contribution value in QF(p) is taken as the main frequency response reliability anomaly contribution item. When the difference between multiple contribution components is within a preset range, multiple contribution items can be selected simultaneously as the basis for restorative retesting. The fourth component of QF(p) represents the high-frequency non-repetitive contribution and does not directly indicate the high-frequency state anomaly of the winding.

[0100] For a specific phase or phase group p, the reliability of the impedance test can be calculated according to CZ(p) = μ1[1-DZrep(p)] + μ2[1-DU(p)] + μ3[1-DI(p)] + μ4[1-Dφ(p)], where μ1 to μ4 are non-negative weights and the sum of the weights is 1. All component indices are normalized to 0 to 1. The larger CZ(p) is, the higher the repeatability stability of the short-circuit impedance test for that phase or phase group.

[0101] For two non-negative test quantities xcur and xrev that need to be compared, a preferred method for calculating the repeatability difference is the symmetrical relative difference D(x) = |xcur - xrev| / [0.5(|xcur| + |xrev|) + ε]. Accordingly, DZrep can be obtained from the D(x) of the current impedance value and the verified impedance value. DU can be obtained from the D(x) of the current test voltage and the verified test voltage. DI can be obtained from the D(x) of the current test current and the verified test current. When Dφ is evaluated using phase angle, the minimum periodic angle difference between the current phase angle and the verified phase angle is used, normalized by a preset angle scale. When evaluated using power factor, the symmetrical relative difference between the current power factor and the verified power factor can be used. All the above indicators are normalized to 0 to 1.

[0102] The impedance reliability anomaly contribution vector is QZ(p) = [μ1DZrep(p), μ2DU(p), μ3DI(p), μ4Dφ(p)]. The component with the largest contribution value in QZ(p) is taken as the main impedance reliability anomaly contribution item. When multiple contribution components are close, multiple contribution items can be selected simultaneously as the basis for restorative retesting. DZrep only represents the repeatability between the current impedance test and the retested impedance test with the same configuration, while the impedance state change in AZ(p) is compared with the historical benchmark or the same three-phase data that meets the structural comparability conditions.

[0103] After calculating CF(p) and CZ(p), confidence gating is performed first. For example... Figure 2 As shown, when CF(p) is below the frequency response confidence threshold and CZ(p) is below the impedance confidence threshold, both types of tests enter the restorative directional retest. When CF(p) is below the frequency response confidence threshold and CZ(p) is not below the impedance confidence threshold, the original frequency response data is excluded from the final fusion, the impedance confidence result is only used as a provisional judgment, and the frequency response enters the restorative directional retest. When CF(p) is not below the frequency response confidence threshold and CZ(p) is below the impedance confidence threshold, the original impedance data is excluded from the final fusion, the frequency response confidence result is only used as a provisional judgment, and the impedance enters the restorative directional retest. Only when CF(p) is not below the frequency response confidence threshold and CZ(p) is not below the impedance confidence threshold, AF(p) and AZ(p), which are weighted separately by the state anomaly features under a unified 0 to 1 anomaly severity scale, are used to calculate DFZ(p). The single-type provisional anomaly judgment is not directly used as the final winding deformation level.

[0104] Before using interphase frequency response benchmarks or same-phase three-phase impedance data for lateral state comparison, a structural comparability judgment is performed on the objects to be compared. Structural comparability fields include at least winding type, winding connection method, tap position, test side, corresponding test configuration, and comparable winding structure. Lateral state comparison is performed only if the above fields meet the preset structural comparability conditions. If the structural comparability conditions are not met, lateral state anomalies are not directly generated; instead, the historical valid benchmark of the phase or phase group p itself is used for longitudinal state evaluation. Test configuration consistency judgment is used to determine whether the current test and the verification test can be compared for repeatability stability, while structural comparability judgment is used to determine whether different phases or objects can be compared laterally for winding state. The comparison objects and technical functions of the two are different.

[0105] AF(p) is calculated for frequency response test data that meets the credibility criteria. AF(p) includes at least one type of frequency band state correlation anomaly and one type of local curve morphology state anomaly. ρLF, ρMF, and ρHF can be used as implementations of the frequency band state correlation anomaly, and their comparison objects are the credible frequency response test data and historical reference data or phase-to-phase reference data that meet the structural comparability criteria. The local curve morphology state anomaly can include resonant peak shift, peak-to-valley amplitude variation, and local curve differences relative to the above references. Let ai(p) be the i-th frequency response state anomaly feature mapped to a unified 0 to 1 scale, and i be the index of the frequency response state anomaly feature. AF(p) can be calculated according to AF(p) = Σ(i=1 to m)αiai(p), where αi ≥ 0 and Σ(i=1 to m)αi = 1.

[0106] In CF(p), Dpeak(p), Damp(p), and NHF(p) are compared with the current test and the same configuration verification test, and are used for test reliability evaluation. In AF(p), the corresponding state characteristics are compared with the reliable data and historical or structurally comparable benchmarks, and are used for winding state anomaly evaluation.

[0107] AZ(p) is calculated for short-circuit impedance test data that meets the reliability criteria. AZ(p) includes at least one type of overall impedance state change index and one type of lateral or longitudinal state comparison index. Impedance error and reactance change can be preferred implementations of the overall state change index. When the structural comparability condition is met, the three-phase impedance deviation can be preferred implementation of the lateral state comparison index. The historical longitudinal change rate and inductance deviation relative to the historical benchmark can be preferred implementations of the longitudinal state comparison index. Let aj(p) be the j-th impedance state anomaly feature mapped to a unified 0 to 1 scale, and j be the impedance state anomaly feature index. AZ(p) can be calculated according to AZ(p) = Σ(j=1 to n)βjaj(p), where βj ≥ 0 and Σ(j=1 to n)βj = 1.

[0108] DZrep(p) in CZ(p) is used for test credibility evaluation, and the impedance error and historical longitudinal change rate in AZ(p) are used for winding state abnormality evaluation, where the two types of indicators do not repeatedly weight the same physical difference.

[0109] To make the abnormality scores formed by different test types comparable, each original state abnormality indicator is converted to a unified dimensionless abnormality severity scale before being calculated into AF(p) or AZ(p). For an indicator x where a larger value indicates a more severe abnormality, a↑(x)=clip[(x-xN) / (xA-xN), 0, 1] can be adopted, where xN is the normal reference boundary, xA is the high abnormality reference boundary, and xA>xN. For an indicator x where a smaller value indicates a more severe abnormality, a↓(x)=clip[(xN-x) / (xN-xA), 0, 1] can be adopted, where xN is the normal reference boundary, xA is the high abnormality reference boundary, and xN>xA.

[0110] clip(y, 0, 1) means: 0 is taken when y<0; y is taken when 0≤y≤1; 1 is taken when y>1. xN and xA can be determined based on applicable test standards, confirmed historical valid samples, statistical distribution of valid data of the same type of equipment, or initial engineering configuration. 0 indicates no obvious abnormality, 1 indicates that the preset high abnormality degree is reached, and the intermediate values indicate the corresponding normalized abnormality severity. After the state abnormality features adopting the severity scale with the same meaning from 0 to 1 are respectively weighted to form AF(p) and AZ(p), DFZ(p) is calculated. In the same complete diagnosis and its automatic retest closed loop, xN and xA remain unchanged.

[0111] When CF(p) is not lower than the frequency response credibility threshold and CZ(p) is not lower than the impedance credibility threshold, the cross-test result difference is calculated according to DFZ(p)=|AF(p)-AZ(p)|. Only when DFZ(p) is not greater than the consistency threshold, the fusion score is calculated according to S(p)=[CF(p)×AF(p)+CZ(p)×AZ(p)] / [CF(p)+CZ(p)]. S(p) is not generated when the test credibility is insufficient or DFZ(p) is greater than the consistency threshold. The grade thresholds satisfy 0≤T1<T2<T3≤1. When 0≤S(p)<T1, normal is output; when T1≤S(p)<T2, mild deformation is output; when T2≤S(p)<T3, obvious deformation is output; when T3≤S(p)≤1, severe deformation is output. T1, T2 and T3 are only responsible for the final grade output, and do not change the credibility gating and consistency judgment.

[0112] Both the frequency response confidence threshold and the impedance confidence threshold are set within a range greater than 0 and not greater than 1, while the consistency threshold is set within a range of 0 to 1. Since the fusion score S(p) is calculated only when CF(p) is not lower than the frequency response confidence threshold and CZ(p) is not lower than the impedance confidence threshold, CF(p) + CZ(p) > 0 when participating in fusion. In a preferred embodiment, the frequency response confidence threshold can be determined based on the lower quantile or statistical lower limit of the CF distribution of historical valid repeated frequency response test samples under similar test configuration conditions. The impedance confidence threshold can be determined based on the lower quantile or statistical lower limit of the CZ distribution of historical valid repeated impedance test samples under similar test configuration conditions. The consistency threshold can be determined based on the upper quantile or statistical upper limit of the DFZ distribution of historical samples confirmed to be in a consistent diagnostic state. When sufficient historical samples are unavailable, an initial configuration can be made based on test standards, equipment type, and engineering verification results, and adjusted before the start of subsequent diagnostic closed-loop as confirmed valid data accumulates.

[0113] When CF(p) is not lower than the frequency response confidence threshold, CZ(p) is not lower than the impedance confidence threshold, but DFZ(p) is greater than the consistency threshold, the weighted contributions of the normalized state anomaly characteristics within AF(p) and AZ(p) are calculated respectively: PF,i(p)=αiai(p), PZ,j(p)=βjaj(p). Thus, AF(p)=Σ(i=1 to m)PF,i(p), AZ(p)=Σ(j=1 to n)PZ,j(p). The largest contribution terms in PF,i(p) and PZ,j(p) are respectively used as the main state anomaly contribution terms of the two test sides. The phase or phase group is directly determined by p. The main state anomaly contribution terms are used to determine the test type, frequency band, or parameters of the verification retest.

[0114] The frequency response testing software allows setting the scan start frequency, scan end frequency, scan mode, and number of scan points. The standard sweep range for frequency response testing refers to the complete frequency scan interval currently used for diagnosis. When anomaly contributions in confidence or status are located to a specific frequency band, the corresponding scan start and end frequencies can be set according to the located frequency band to perform targeted frequency band sweep retesting on the phase or phase group p. Targeted frequency band sweep retesting is preferentially used. If necessary, an expanded range sweep retesting including the located frequency band can also be performed.

[0115] Key frequency band sweep retesting is mainly used for rapid verification of identified test quality anomalies or state anomalies. Since CF(p) and AF(p) may be composed of repeatability stability characteristics or state anomaly characteristics from multiple frequency bands, when the key frequency band retesting range only covers a part of the original standard diagnostic sweeping range, the local frequency band retesting data is only used for the confirmation of the corresponding anomaly and retesting decisions, and does not directly replace the complete frequency response data to form the final global CF(p), AF(p), DFZ(p), or S(p).

[0116] Once local frequency band retesting confirms that the corresponding test conditions have been restored, or confirms that a certain state anomaly has repeatability stability, if a final comprehensive diagnostic result needs to be re-formed, complete confirmed frequency response data is obtained again according to the standard frequency sweep range used in the current diagnosis. Corresponding complete verification frequency response data is also obtained under the same standard frequency sweep range and test configuration. These two sets of data form a complete and valid confirmed test data pair. Based on this data pair, CF(p) is recalculated. Only when CF(p) is not lower than the frequency response confidence threshold is AF(p), DFZ(p), and S(p) when the fusion condition is met used to update the complete confirmed frequency response data. Therefore, key frequency band retesting is used for cause localization and targeted confirmation, while complete confirmed frequency sweeping is used for the final global diagnostic update.

[0117] The frequency response side of the restorative directional retest takes corresponding retest actions based on the different components in QF(p) as the main contributors. The candidate frequency bands for the restorative retest are preferentially determined based on the frequency bands where each mismatch feature is recorded during the aforementioned Dpeak and Damp calculations, and are prioritized in conjunction with the contribution ranking of each component in QF(p). When λ1[1-ρrep(p)] is the main contributor, it indicates that insufficient repetition correlation is the main reason for low reliability. In this case, the repetition correlation of the current test and the verification test in the low-frequency, mid-frequency, and high-frequency bands are compared respectively, and the frequency band with the lowest correlation is determined as the key retest frequency band, and the scan start frequency and scan end frequency are set accordingly. When λ4NHF(p) is the main contributor, it indicates that high-frequency non-repetitive disturbance is the main cause. The interval where the high-frequency non-repetitive component is located is determined as the key retest frequency band, and the inspection items of grounding status, test lead layout, and wiring contact status are output. When λ2Dpeak(p) or λ3Damp(p) is the main contributor, it indicates that peak position drift or amplitude deviation is the primary cause. The frequency bands exhibiting repeated peak position drift, repeated peak-to-valley amplitude deviation, or characteristic mismatch are identified as key retesting bands. Simultaneously, the test lead position and wiring status are checked. The phase or phase group for the above frequency response tasks is determined by p associated with QF(p). During retesting, targeted frequency sweep retesting is prioritized using the identified frequency band. If necessary, an expanded range frequency sweep retest including the identified frequency band can also be performed.

[0118] The impedance side of the restorative directional retest takes corresponding retest actions based on different components of QZ(p) as the main contributing factors. When μ1DZrep(p) is the main contributing factor, it indicates that the repetitive impedance deviation is the main reason for low reliability. A short-circuit impedance restorative retest task is generated, and the wiring contact status, tap position, and short-circuit impedance test mode consistency check items are output. When μ2DU(p) is the main contributing factor, it indicates that insufficient test voltage stability is the main reason. A short-circuit impedance restorative retest task is generated, and the test voltage stability check items are output. When μ3DI(p) is the main contributing factor, it indicates that abnormal current loop or contact status is the main reason. A short-circuit impedance restorative retest task is generated, and the current loop and contact status check items are output. When μ4Dφ(p) is the main contributing factor, it indicates that abnormal phase or power factor sampling and wiring status are the main reasons. A short-circuit impedance restorative retest task is generated, and the phase or power factor sampling and wiring status check items are output. The phase or phase group of the above impedance tasks is determined by p associated with QZ(p).

[0119] The frequency response side of the verification retest also takes corresponding retest actions based on the type of state anomaly contribution. When the maximum contribution in PF,i(p) comes from a state correlation anomaly in a certain frequency band, the main anomaly frequency band corresponding to phase or phase group p is determined as the verification frequency sweep range. When the maximum contribution comes from the resonant peak migration relative to the historical reference, the resonant peak and its adjacent frequency window are determined as the verification retest object. When the maximum contribution comes from the change in peak-valley amplitude state or local curve difference, the corresponding frequency window is determined as the verification retest range. All of the above verification retests require setting the corresponding scan start frequency and scan end frequency according to the determined frequency band.

[0120] On the impedance side of the verification retest, when the maximum contribution in PZ,j(p) comes from the three-phase impedance deviation and the structural comparability condition is met, a short-circuit impedance verification retest is performed on phase or phase group p and the corresponding structurally comparable reference phase, and a lateral state comparison is performed again. When the maximum contribution comes from the historical longitudinal rate of change, impedance error, inductance deviation, or reactance change, a short-circuit impedance test with the same configuration is performed again on phase or phase group p, and a longitudinal state comparison is performed again with the historical effective benchmark corresponding to that phase or phase group. The above mapping is a preferred implementation method, and the verification retest is not limited to only the listed situations.

[0121] In this embodiment, targeted retesting includes restorative targeted retesting and confirmatory targeted retesting, which differ in their triggering conditions and data processing methods. Restorative targeted retesting is triggered when CF(p) or CZ(p) falls below the corresponding confidence threshold, used to eliminate abnormal test conditions and re-obtain valid test data with repeatability stability. Previously unreliable data is only retained for traceability and does not participate in the final fusion. After the corresponding test conditions are corrected, test data and verification data with the same configuration are re-obtained, valid test data pairs are re-established, and CF(p) or CZ(p) is recalculated. AF(p) or AZ(p) is calculated only after the corresponding test confidence reaches the threshold; otherwise, restorative targeted retesting continues until the confidence condition is met or the maximum number of retests is reached.

[0122] Confirmatory directional retests are triggered when both CF(p) and CZ(p) meet the credibility condition, but DFZ(p) exceeds the consistency threshold. The purpose is not to negate the credibility of the original test data, but to verify the specific source of the conflict between the two types of credible tests in evaluating the winding state. The original credible test data is retained, and the phase or phase group, test type, frequency band, or parameters are determined based on the main state anomaly contributors in each PF,i(p) and PZ,j(p), and targeted confirmatory retests are then implemented.

[0123] Before the verification retest data enters the state anomaly evaluation, the repeatability stability is reconfirmed. When the verification retest and the original credible test meet the test configuration comparability conditions, the verification retest data and the original credible test data form a valid verification test data pair. If the original credible test data is not suitable as verification data or the configuration change exceeds the preset allowable configuration deviation, further verification verification test data is obtained under the same test configuration to form a valid verification test data pair. For frequency response verification retest, CFv(p) is calculated, and AF(p) is updated only when CFv(p) is not lower than the frequency response credibility threshold. For short-circuit impedance verification retest, CZv(p) is calculated, and AZ(p) is updated only when CZv(p) is not lower than the impedance credibility threshold. When only one type of verification retest is performed, the other type of original credible test data and its state anomaly score are retained, and DFZ(p) is recalculated by the updated AF(p) or AZ(p) and the other type of original credible state anomaly score. When both types of verification retests are performed, DFZ(p) is recalculated only when CFv(p) is not lower than the frequency response confidence threshold and CZv(p) is not lower than the impedance confidence threshold. If any verification data does not meet its corresponding threshold, only that test side is converted to a restorative directional retest. When local frequency band verification requires updating the global results, complete confirmed frequency response data is obtained again according to the standard frequency sweep range used in the current diagnosis, and corresponding complete verification frequency response data is obtained under the same standard frequency sweep range and test configuration. The two form a complete and valid test data pair. CF(p) is recalculated based on this data pair. AF(p), DFZ(p), and S(p) when the fusion condition is met are updated only when CF(p) is not lower than the frequency response confidence threshold.

[0124] After the confirmatory retest is completed, if the recalculated DFZ(p) is not greater than the consistency threshold, the evaluation of the two test states meets the consistency condition and enters the credibility fusion stage. If the corresponding test still meets the credibility condition and DFZ(p) is still greater than the consistency threshold after the confirmatory retest, the continuous cross-test result conflict is recorded. Fusion is not forced by adjusting weights, lowering the consistency threshold, or other means. Based on the main state anomaly contribution items determined by PF,i(p) and PZ,j(p), the three-phase horizontal comparison that meets the structural comparability condition, and the historical vertical comparison, a suspected local anomaly or manual review prompt is output.

[0125] The maximum number of retests represents the maximum number of retests allowed to be automatically generated for the same phase or phase group, the same test type, and the same retest reason. This is a configurable parameter. When the reliability of the corresponding test is still below the threshold after the maximum number of restorative retests, automatic retesting stops and an error message indicating abnormal test conditions and requiring manual review is output. When the maximum number of confirmatory retests is reached, and both types of tests are still reliable but DFZ(p) continues to exceed the consistency threshold, automatic confirmatory retesting stops and an error message indicating continuous cross-test result conflicts and requiring manual review is output. The maximum number of retests is not limited to a fixed value.

[0126] To ensure the comparability of evaluation results across rounds within the same diagnostic loop, the following parameters remain unchanged after the start of a complete diagnostic and its automatic retest loop: consistency judgment rules, confidence weights λ and μ, state anomaly weights α and β, resonance peak mismatch contribution parameter cpeak,miss, peak-valley feature mismatch contribution parameter camp,miss, confidence threshold, consistency threshold, unified severity mapping boundary, and grade thresholds T1, T2, and T3. Parameter updates are performed after the current diagnostic loop concludes, and parameters are not adjusted in reverse based on the results of a particular retest to alter the current diagnostic conclusion, thus ensuring the objectivity and consistency of the diagnostic conclusion. Maintaining consistency in weights and mismatch contribution parameters during retesting ensures the comparability of confidence and anomaly contributions across different test rounds.

[0127] The database associates and stores test configuration identifiers GF and GZ, phase or phase group p, test data credibility status, main credibility anomaly contributors, main status anomaly contributors, retest type, retest trigger reason, original test data status, retest task parameters, verification retest credibility, DFZ before retest, DFZ after retest, number of retests, data association before and after retest, and manual review status. The database excludes previously untrusted test data in the recovery retest path from final fusion marking. Previously trusted test data in the verification retest path remains credible and is not automatically marked as untrustworthy due to a large DFZ(p). Only data whose credibility has been confirmed and whose final conclusion is valid is allowed as subsequent historical benchmark candidate data.

[0128] Example 2:

[0129] like Figure 3 As shown, this embodiment provides a transformer winding deformation diagnostic system based on reliability and consistency verification. The system includes a measurement host and a computer analysis unit communicatively connected to the measurement host. The measurement host includes a frequency response test channel, a short-circuit impedance test channel, a synchronous sampling and processing unit, a test control unit, and an isolation protection unit, such as... Figure 3As shown in Figure (a), the front panel of the measuring unit has a power switch 1; the upper part has four terminals A, B, C, and N, labeled 10, 11, 12, and 13 respectively; the lower part has two sets of test terminals: the first set consists of current terminals Ia, Ib, Ic, and In, labeled 2, 3, 4, and 5 respectively, and the second set consists of voltage terminals Ua, Ub, Uc, and Un, labeled 6, 7, 8, and 9 respectively. The front panel is used for excitation injection, response acquisition, and voltage and current signal connection in frequency response testing and short-circuit impedance testing.

[0130] like Figure 3 As shown in Figure (b), the back panel of the measurement host has a high-voltage input area and a high-voltage output area. The high-voltage input area includes three terminals, labeled 101, 102, and 103, corresponding to terminals Ic, Ib, and Ia, respectively. The high-voltage output area includes three current terminals in the upper row and three voltage terminals in the lower row. The upper row of current terminals, from left to right, are labeled Ic, Ib, and Ia, labeled 112, 109, and 107, respectively. The lower row of voltage terminals, from left to right, are labeled Uc, Ub, and Ua, labeled 111, 110, and 108, respectively. The back panel also includes a USB interface 104, a grounding terminal 105, and a power input socket 106. The back panel interface is used for test modes that require high-voltage input / output signals to be connected from the back panel, such as external input impedance testing.

[0131] like Figure 4 and Figure 5 As shown, the frequency response test channel is used to generate a sweep frequency signal and acquire the response signal of the winding under test. The short-circuit impedance test channel is used to perform internal low-voltage short-circuit impedance testing or external input impedance testing. The synchronous sampling and processing unit completes the synchronous sampling and data processing of the corresponding signals, the test control unit executes the software-controllable parameters in the test task, and the isolation protection unit is used for the isolation and protection of the test circuit.

[0132] Figure 4 The frequency response test wiring shown is used to connect the sweep frequency excitation to the winding under test and obtain the response signal. Figure 5 The wiring shown in Figure (a) is used for internal low-voltage short-circuit impedance testing. Figure 5 The wiring shown in Figure (b) is used for external input impedance testing. Each wiring method and item selection should be performed according to the corresponding test type.

[0133] The computer analysis unit includes a test configuration consistency verification module, a current and review test data management module, a frequency response test reliability evaluation module, an impedance test reliability evaluation module, a reliability anomaly contribution location module, a frequency response anomaly quantification module, an impedance anomaly quantification module, a result consistency verification module, a state anomaly contribution location module, a dynamic fusion module, a directional retest decision module, a database management module, and a report generation module. These modules perform corresponding processing based on the data collected by the measurement host.

[0134] Physical test conditions, such as the winding under test, phase or phase group, tap position, winding connection method, wiring method, grounding status, test lead arrangement, and contact status, are output for inspection or adjustment by the test task interface and confirmed or adjusted by the operator. The scan start frequency, scan end frequency, scan mode, number of scan points, and software-controllable short-circuit impedance test parameters are sent to the test control unit after the physical test conditions are confirmed. The test control unit then calls the corresponding test channel to execute the test and collect data.

[0135] The test configuration consistency verification module verifies discrete configuration fields and continuous test parameters before the current test data and the review test data form a valid test data pair. The current and review test data management modules only receive data that meets preset comparability conditions. Before using inter-phase reference data or same-phase three-phase data for transverse winding anomaly evaluation, the winding type, winding connection method, tap position, test side, corresponding test configuration, and winding structure of the object to be compared are also verified. If the structural comparability conditions are not met, the historical valid reference of the phase or phase group itself is used first.

[0136] The computer analysis unit performs test reliability evaluation, winding state anomaly evaluation, result consistency judgment, anomaly contribution location, and directional retest decision-making according to the above methods. The directional retest decision module generates retest tasks based on major reliability anomaly contributions or major state anomaly contributions, determines the phase or phase group, test type, and key frequency bands or test parameters, and generates corresponding manual inspection items. Restorative directional retest data reconstructs valid test data pairs according to the above rules, and after reliability gating, enters the winding state anomaly evaluation. Confirmatory directional retest data reconfirms CFv or CZv according to the corresponding test type, and updates the corresponding state anomaly results after meeting the corresponding reliability conditions.

[0137] New data obtained from the retest is returned to the computer analysis unit and associated with the retest task, the original test data, and the corresponding test configuration. If the original test data is unsuitable as verification data or the configuration changes exceed the allowable range, the operator confirms the required physical test conditions, and the test control unit then executes the same configuration verification test. Original unreliable data in restorative directional retests is retained for traceability but does not participate in the final fusion. Original reliable data in confirmatory directional retests remains reliable. Automatic retesting stops and manual verification is initiated when the maximum number of retests is reached.

[0138] When the frequency response directional retest only covers a local key frequency band and requires updating the global diagnostic results, obtain complete confirmed frequency response data and complete verification frequency response data with the same configuration according to the above-mentioned complete confirmation frequency sweep rules, re-form a complete and valid confirmation test data pair, and complete the subsequent evaluation. The local key frequency band data is only used for targeted confirmation of abnormal factors and does not directly replace the complete frequency response data. If the cross-test results still do not meet the consistency conditions after the confirmatory retest, record the persistent conflict and handle it according to the above-mentioned closed-loop termination rules.

[0139] like Figure 6 As shown, the database management module stores current test, review test, recovery-oriented retest, verification-oriented retest, and complete confirmation test data according to the aforementioned data status and association rules. It also records the test's trustworthiness status, anomaly contribution, retest type, results before and after the retest, and the final diagnostic conclusion. Originally untrusted data is retained for traceability but does not participate in the final fusion; trusted conflict data maintains a trusted status; and confirmed valid data serves as candidates for updating historical benchmarks.

[0140] The report generation module generates test data, curves, credibility and result consistency information, retesting process, and final diagnostic conclusions based on database records, and supports report previewing, saving, and exporting.

[0141] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed in this application can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0142] In the embodiments provided in this application, it should be understood that the disclosed devices / terminal equipment and methods can be implemented in other ways. For example, the device / terminal equipment embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling or direct coupling or communication connection may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0143] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0144] If the integrated module / unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electrical carrier signals and telecommunication signals.

[0145] The implementation of all or part of the processes in the methods of the above embodiments can also be accomplished by a computer program product. When the computer program product is run on a terminal device, the terminal device can implement the steps in the various method embodiments described above.

[0146] The embodiments described above are only used to illustrate the technical solutions of this application, and are not intended to limit it. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A method for diagnosing transformer winding deformation based on reliability and consistency verification, characterized in that, Includes the following steps: Obtain basic information and historical baseline data of the transformer under test; Using the phase or phase group p of the winding under test as the evaluation unit, the current test data and the verification test data are obtained. The verification test data refers to the same configuration repeated test data used to evaluate the repeatability stability of the current test data. The test data includes frequency response test data and short-circuit impedance test data. The current test data and the review test data are checked for consistency in test configuration, and only test data that meet the preset comparability conditions are formed into valid test data pairs; Based on the effective test data pairs, calculate the frequency response test reliability CF(p) and impedance test reliability CZ(p) for each phase or phase group p, and determine the corresponding main reliability anomaly contribution items according to the contribution of each reliability component index to the reliability reduction. CF(p) is compared with the frequency response confidence threshold τF, and CZ(p) is compared with the impedance confidence threshold τZ to determine whether the confidence condition is met. For frequency response test data that meets the corresponding confidence condition, it is compared with historical reference data or phase-to-phase reference data that meets the structural comparability condition to obtain the frequency response anomaly score AF(p). For short-circuit impedance test data that meets the corresponding confidence condition, it is compared with historical reference data or phase-to-phase reference data that meets the structural comparability condition to obtain the impedance anomaly score AZ(p). Test data that is determined not to meet the corresponding confidence threshold does not participate in the final dynamic fusion, and a restorative directional retest task is generated based on its main confidence anomaly contribution item. When both CF(p) and CZ(p) satisfy the corresponding credibility conditions, calculate the cross-test result difference DFZ(p) based on AF(p) and AZ(p). When DFZ(p) is not greater than the consistency threshold δ, dynamically fuse the credibility of AF(p) and AZ(p) according to CF(p) and CZ(p). When both CF(p) and CZ(p) satisfy the corresponding confidence conditions but DFZ(p) is greater than δ, the main state anomaly contribution items in AF(p) and AZ(p) are determined respectively, and a verification-oriented retest task is generated. For retest data obtained from restorative or confirmatory retesting tasks, the credibility is reconfirmed, and AF(p) or AZ(p) is updated only after the corresponding credibility conditions are met. Based on the reconfirmed credibility, abnormal status score, and consistency across test results, the final diagnostic result is output. The diagnostic result includes at least one of the following: winding deformation status level, abnormality prompt, or manual review instruction.

2. The transformer winding deformation diagnosis method based on reliability and consistency verification according to claim 1, characterized in that, The test configuration consistency verification includes discrete configuration field consistency verification and continuous test parameter consistency verification. The discrete configuration fields for frequency response testing should include at least the winding under test, phase or phase group, tap position, winding connection method and wiring method, and the continuous test parameters should include at least the sweep range and frequency sampling settings. The discrete configuration fields for short-circuit impedance testing include at least the winding under test, phase or phase group, tap position, winding connection method, wiring method, and test mode. The continuous test parameters include at least the test voltage and test current conditions. When the discrete configuration fields are the same, and the continuous test parameters are the same or their difference does not exceed the preset allowable configuration deviation, the current test data and the verification test data are determined to meet the comparability conditions.

3. The transformer winding deformation diagnosis method based on reliability and consistency verification according to claim 1, characterized in that, The frequency response test confidence level CF(p) is calculated as follows: CF(p)=λ1ρrep(p)+λ2[1-Dpeak(p)]+λ3[1-Damp(p)]+λ4[1-NHF(p)]; Wherein, ρrep(p) is the repetitive correlation between the current frequency response test data and the frequency response verification test data with the same configuration, Dpeak(p) is the normalized repetitive drift of the corresponding resonance peak, Damp(p) is the normalized repetitive deviation of the corresponding peak and valley amplitude, NHF(p) is the high-frequency non-repetitiveness index between the current frequency response test data and the frequency response verification test data with the same configuration, and λ1 to λ4 are non-negative weights with a sum of 1. The impedance test reliability CZ(p) is calculated as follows: CZ(p)=μ1[1-DZrep(p)]+μ2[1-DU(p)]+μ3[1-DI(p)]+μ4[1-Dφ(p)]; Where DZrep(p) is the relative deviation of the repeating impedance between the current impedance test data and the verification test data of the same configuration impedance, DU(p) is the test voltage consistency deviation, DI(p) is the test current stability deviation, Dφ(p) is the phase or power factor fluctuation, and μ1 to μ4 are non-negative weights with a sum of 1.

4. The transformer winding deformation diagnosis method based on reliability and consistency verification according to claim 3, characterized in that, The main confidence anomaly contribution items include the main frequency response confidence anomaly contribution items and the main impedance confidence anomaly contribution items, which are obtained in the following ways: Construct the frequency response confidence anomaly contribution vector QF(p) from the components of CF(p): QF(p)=[λ1(1-ρrep(p)), λ2Dpeak(p), λ3Damp(p), λ4NHF(p)]; The component with the largest value in QF(p) is determined as the main frequency response confidence anomaly contribution item; Construct the impedance confidence anomaly contribution vector QZ(p) from the components of CZ(p): QZ(p)=[μ1DZrep(p),μ2DU(p),μ3DI(p),μ4Dφ(p)]; The component with the largest value in QZ(p) is determined as the main impedance reliability anomaly contribution item.

5. The transformer winding deformation diagnosis method based on reliability and consistency verification according to claim 4, characterized in that, The restorative targeted retest is generated based on the main confidence anomaly contribution items, wherein: When the frequency response test confidence level is lower than the frequency response confidence level threshold τF, and λ1[1-ρrep(p)] is the main contributor, the repetition correlation between the current test and the review test in the low frequency band, mid frequency band and high frequency band are compared respectively. The frequency band with the lowest repetition correlation is determined as the key retest frequency band, and the frequency response recovery retest task of the corresponding phase or phase group is generated. When the frequency response test reliability is lower than the frequency response reliability threshold τF, and λ4NHF(p) is the main contributing factor, the interval where the high-frequency non-repeating component is located is determined as the key retest frequency band, a frequency response recovery retest task is generated, and the inspection items of grounding status, test lead arrangement and wiring contact status are output. When the frequency response test confidence level is lower than the frequency response confidence level threshold τF, and λ2Dpeak(p) or λ3Damp(p) is the main contributor, the frequency bands with repeated drift of the main peak position, repeated deviation of peak and valley amplitude, or corresponding feature mismatch will be identified as key retest frequency bands, generating a frequency response recovery retest task, and outputting test lead position and wiring status inspection items. When the impedance test reliability is lower than the impedance reliability threshold τZ, and μ1DZrep(p) is the main contributor, a short-circuit impedance recovery retest task is generated, and the wiring contact status, tap position and short-circuit impedance test mode consistency check items are output. When the impedance test reliability is lower than the impedance reliability threshold τZ, and μ2DU(p), μ3DI(p), or μ4Dφ(p) are the main contributors, a corresponding short-circuit impedance recovery retest task is generated, and the corresponding test voltage stability, current loop and contact status or phase sampling and wiring status inspection items are output.

6. The transformer winding deformation diagnosis method based on reliability and consistency verification according to claim 1, characterized in that, When using interphase reference data or same three-phase data for lateral state comparison, first verify the structural comparability of the objects to be compared. The fields for structural comparability verification include at least winding type, winding connection method, tap position, test side, corresponding test configuration, and winding structure. Only when the field meets the preset structural comparability conditions, the horizontal state comparison is performed using inter-phase reference data or three-phase data of the same time; if the structural comparability conditions are not met, the vertical state comparison is performed using the historical reference data of the phase or phase group itself.

7. The transformer winding deformation diagnosis method based on reliability and consistency verification according to claim 1, characterized in that, The AF(p) and AZ(p) are calculated by weighting the corresponding state anomaly features after a unified 0-1 anomaly severity mapping, where 0 corresponds to a state with no obvious anomaly and 1 corresponds to a preset high anomaly severity state, and are calculated according to: DFZ(p) = |AF(p) - AZ(p)| calculates the difference between test results; When CF(p)≥τF, CZ(p)≥τZ and DFZ(p)≤δ, the fusion score S(p) is calculated according to: S(p)=[CF(p)×AF(p)+CZ(p)×AZ(p)] / [CF(p)+CZ(p)]. Under the condition 0≤T1<T2<T3≤1, the output is normal when 0≤S(p)<T1, slightly distorted when T1≤S(p)<T2, obviously distorted when T2≤S(p)<T3, and severely distorted when T3≤S(p)≤1.

8. The transformer winding deformation diagnosis method based on reliability and consistency verification according to claim 7, characterized in that, When CF(p)≥τF, CZ(p)≥τZ but DFZ(p)>δ, the weighted contribution of each normalized state anomaly feature constituting AF(p) is defined as PF,i(p)=αiai(p), and the weighted contribution of each normalized state anomaly feature constituting AZ(p) is defined as PZ,j(p)=βjaj(p), where αi is the non-negative frequency response state anomaly weight and the sum of all αi is 1, βj is the non-negative impedance state anomaly weight and the sum of all βj is 1, AF(p) is equal to the sum of each PF,i(p), and AZ(p) is equal to the sum of each PZ,j(p); The largest contributing terms in PF,i(p) and PZ,j(p) are respectively identified as the main state anomaly contributing terms of the corresponding test side, and the test type, frequency band or parameters of the verification retest are determined by combining the phase or phase group p and the main state anomaly contributing terms.

9. The transformer winding deformation diagnosis method based on reliability and consistency verification according to claim 1, characterized in that, For restorative targeted retesting, the original test data that was determined not to meet the credibility conditions is retained for process traceability and marked as not to participate in the final fusion. After correcting the corresponding test conditions, the test data is re-acquired, and the corresponding verification test data is obtained under the same test configuration to re-form a valid test data pair. The corresponding test credibility is recalculated based on the re-formed valid test data pair. For frequency response recovery retests, AF(p) is calculated using the corresponding reliable frequency response data only if the recalculated CF(p) is not lower than the frequency response reliability threshold τF; for short-circuit impedance recovery retests, AZ(p) is calculated using the corresponding reliable impedance data only if the recalculated CZ(p) is not lower than the impedance reliability threshold τZ; test paths that do not meet the corresponding reliability thresholds do not calculate AF(p) or AZ(p), and recovery directional retests continue to be performed. For confirmatory targeted retests, CF(p) and CZ(p) data that already meet the credibility criteria are not marked as untrustworthy data; Before using confirmatory retest data to update AF(p) or AZ(p), first confirm that the confirmatory retest data meets the corresponding confidence conditions; If DFZ(p) is not greater than the consistency threshold after confirmatory retesting, then fusion is performed; if the two types of tests still meet the credibility condition after confirmatory retesting but DFZ(p) is still greater than the consistency threshold, then fusion is not performed, and at least one of the following is output: continuous cross-test result conflict, suspected local anomaly, or manual review prompt. When the automatic retesting count for the same phase or phase group, the same test type, and the same retesting reason reaches the maximum retesting count Nmax and still cannot meet the corresponding termination condition, the automatic retesting will stop and the process will be transferred to manual review.

10. A transformer winding deformation diagnosis system based on reliability and consistency verification, characterized in that, Includes a measurement host and a computer analysis unit that communicates with the measurement host; The measurement host includes a frequency response test channel, a short-circuit impedance test channel, a synchronous sampling and processing unit, a test control unit, and an isolation protection unit; The computer analysis unit includes a test configuration consistency verification module, a current and review test data management module, a frequency response test reliability evaluation module, an impedance test reliability evaluation module, a reliability anomaly contribution location module, a frequency response anomaly quantification module, an impedance anomaly quantification module, a result consistency verification module, a state anomaly contribution location module, a dynamic fusion module, a directional retest decision module, a database management module, and a report generation module. The test configuration consistency verification module is used to verify the corresponding test configuration before the current test data and the review test data form a valid test data pair, and to verify the structural comparability of the objects to be compared before using inter-phase reference data or same three-phase data for lateral state comparison. The frequency response test reliability evaluation module calculates NHF(p) using the high-frequency non-repetitive components between the current frequency response test data and the frequency response test data with the same configuration; The credibility anomaly contribution localization module is used to determine the main test factors that cause a decrease in the credibility of frequency response tests or impedance tests. The state anomaly contribution localization module is used to determine the main state anomaly contribution items when both types of tests are reliable but the cross-test results do not meet the consistency condition. The dynamic fusion module generates a fusion result only when both types of tests meet the corresponding credibility threshold and the difference between the cross-test results is no greater than the consistency threshold. The directional retest decision module generates retest task parameters and manual inspection items, which involve manual inspection items such as the winding to be tested, phase or phase group, tap position, winding connection method, wiring method, grounding status or test lead arrangement, and are output through the test task interface. After manual confirmation, the software-controlled retest task parameters are sent to the test control unit, which then calls the frequency response test channel or the short-circuit impedance test channel to execute the corresponding test. The data obtained from the retest is returned to the computer analysis unit. The restorative retest data is retested, the consistency of the configuration is checked, and the valid test data pairs are reconstructed. The corresponding test confidence is recalculated, and AF(p) or AZ(p) is updated only after the corresponding confidence condition is met. The confirmatory retest data calculates the confirmatory retest confidence according to the test type. The corresponding AF(p) or AZ(p) is updated only when the corresponding confirmatory retest confidence meets its threshold. The report generation module is used to generate and output a diagnostic result containing the winding deformation status level or the manual review prompt based on the fusion result output by the dynamic fusion module or the manual review prompt output by the directional retest decision module. The diagnostic result includes at least one of the winding deformation status level, abnormal prompt, or manual review instruction.

Images