Distribution transformer internal defect diagnosis method and device based on virtual differential current

Abstract

The invention relates to the technical field of transformer internal defect diagnosis, in particular to a distribution transformer internal defect diagnosis method and device based on virtual differential current. Comprising the following steps: synchronously acquiring data; calculating an actually measured differential current; establishing a differential current model; according to a differential current model, calculating a deviation value between the actually measured differential current of each phase and a model prediction value, and taking the deviation value as the virtual differential current; carrying out filtering processing on the virtual differential current; and judging defects. According to the method, a differential current model associated with winding impedance and excitation branch parameters is constructed based on operation data when the transformer has no defects, so that the model is bound with physical topological characteristics of the transformer, and virtual differential current (deviation between a measured value and a model predicted value) can reflect actual state change of the winding in the transformer; and the physical relevance and reliability of defect diagnosis are enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of transformer internal defect diagnosis, and in particular to a method and device for diagnosing internal defects of a distribution transformer based on virtual differential current. Background Art

[0002] Distribution transformers are critical equipment for power conversion in distribution networks. Accurately diagnosing internal winding defects (such as turn-to-turn shorts and shorts to ground) is crucial for ensuring safe grid operation. Existing diagnostic methods based on differential current often misdiagnose or miss defects due to factors such as dynamic changes in turns ratio caused by tap position adjustment and zero-sequence current interference in star-connected configurations. There is an urgent need to develop diagnostic logic that adapts to the unique operating conditions of transformers.

[0003] In the prior art, for example, Chinese patent CN202411453249.8 discloses a longitudinal differential protection method, device, and storage medium for a microgrid line. The method collects normal and short-circuit parameters of the microgrid line, calculates the virtual short-circuit current output by the inverter, and then obtains the virtual differential current and the adjustment braking current to determine the line short-circuit fault, aiming to improve the sensitivity of the protection device under small short-circuit current; for example, Chinese patent CN202411451521.9 discloses a regionalized differential protection method, device, and storage medium for a microgrid system. For the regionalized microgrid structure, the method also calculates the virtual differential current based on the virtual short-circuit current to improve the selectivity and reliability of protection under complex structures.

[0004] While the aforementioned technical solutions involve the application of virtual differential current, they all target short-circuit protection for microgrid lines or systems and fail to consider the unique properties of distribution transformers. First, they do not incorporate physical parameters such as the winding impedance and excitation branch of the transformer's T-shaped equivalent circuit. The model is decoupled from the actual device topology, making it difficult to reflect impedance changes within the transformer's internal windings. Second, they do not adapt to the dynamic changes in turns ratio caused by transformer tap position adjustment, nor do they design targeted compensation logic for zero-sequence current on the star-connected side. Third, they can only determine whether a fault has occurred, but cannot infer the defect type (such as the difference between a turn-to-turn short circuit and a short circuit to ground) based on changes in physical model parameters. In light of this, we propose a method and device for diagnosing internal defects in distribution transformers based on virtual differential current. Summary of the Invention

[0005] The object of the present invention is to provide a method and device for diagnosing internal defects of a distribution transformer based on virtual differential current, so as to solve the problems raised in the above background technology.

[0006] To solve the above technical problems, one of the objectives of the present invention is to provide a method for diagnosing internal defects of a distribution transformer based on virtual differential current, comprising the following steps: S100, synchronous data acquisition: real-time acquisition of the three-phase line voltage and line current on the high-voltage side of the transformer, the three-phase phase voltage and line current on the low-voltage side, and the tap position information; S200, calculating the measured differential current: generating a real-time differential current of each phase based on the high-voltage side line current, the low-voltage side line current, and the turns ratio determined by the tap position, combined with the zero-sequence current on the star connection side; S300, establishing a differential current model: using the operating data of the transformer when it is defect-free as training samples, the training samples include the measured differential current of each phase in the defect-free state generated in S200, the high-voltage side line voltage and the low-voltage side phase voltage obtained in S100, based on the transformer T-shaped equivalent circuit, correlating the winding impedance distribution with the excitation branch parameters, and obtaining the differential current model of the transformer when it is defect-free through training; S400, generating a virtual differential current: based on the differential current model established in S300, calculating the deviation between the measured differential current of each phase generated in S200 and the model predicted value as the virtual differential current; and filtering the virtual differential current; S500, defect determination: when the virtual differential current of any phase exceeds a preset threshold, it is determined that the transformer has an internal defect.

[0007] As a further improvement of the present technical solution, the synchronous data collection in S100 includes the following steps: S100.1. Real-time collection of the three-phase line voltage and line current on the high-voltage side of the transformer and the three-phase phase voltage and line current on the low-voltage side, with a collection frequency of not less than 50 Hz; S100.2. Synchronously obtain transformer tap position information, and determine the nominal turns ratio of the transformer (the ratio of the nominal number of turns of the high-voltage winding to the nominal number of turns of the low-voltage winding) based on the transformer tap position information. S100.3. Associate the voltage and current data collected in S100.1 with the nominal turns ratio determined in S100.2 and store them as original input parameters for calculating the real-time differential current of each phase in S200.

[0008] As a further improvement of the present technical solution, in S200, calculating the measured differential current includes the following steps: S200.1. Extracting synchronous measurement parameters: From the synchronously collected data of S100, extract: High voltage side line current: (three-phase line current); Low voltage side line current: (three-phase line current); Turns ratio parameters: (The ratio of the nominal number of turns of the transformer high-voltage winding to the nominal number of turns of the low-voltage winding is determined by the tap position); S200.2. Calculate the high voltage side line current components: The three-phase components are derived from the three-phase line current on the high-voltage side: ;in, Represents the high-voltage side three-phase current component derived from the high-voltage side three-phase line current; S200.3. Calculate the zero-sequence current on the star connection side: If the low voltage side is star-connected, calculate the zero sequence current on the low voltage side ; S200.4. Calculate the measured differential current of each phase: The measured differential current of each phase is generated according to the following formula: ;in, Indicates the final calculated measured differential current of each phase.

[0009] As a further improvement of the present technical solution, in S300, the training samples are time-stamped synchronized phasor data acquired during the trial operation, including: No. Measured differential current of phase ; High voltage side line voltage: Phase voltage 、 Phase voltage 、 Phase voltage ; Low voltage side phase voltage: Phase voltage 、 Phase voltage 、 Phase voltage ; The operating data of the transformer when it is defect-free is steady-state operating data without faults for more than 72 consecutive hours during the trial operation.

[0010] As a further improvement of the present technical solution, the step of obtaining the differential current model of the transformer when it is free of defects through training in S300 includes the following steps: S300.1. Data preparation: During the trial run, the time series ( is the number of sampling points), collect: Phase: High voltage side line voltage , low voltage side phase voltage , measured differential current ; Phase: High voltage side line voltage , low voltage side phase voltage , measured differential current ; Phase: High voltage side line voltage , low voltage side phase voltage , measured differential current ; S300.2. Construct matrices and vectors: For the first ,structure: Input matrix: ;in, For the Phase high voltage side line voltage; For the Phase voltage on the low voltage side; For the Sampling moments; Output vector: ;in, For the Phase measured differential current; Parameter vector: ;in, For the The coefficients of line voltage in the phase model; For the The coefficients of phase voltage in the phase model; S300.3. Least squares method for parameter identification: ; in, for The conjugate transposed matrix of ; for The inverse matrix of S300.4. Establish the differential current model of each phase: ; in, For the The model predicts the differential current values ​​of the phases.

[0011] As a further improvement of this technical solution, the differential current model parameters established in S300 are ), which is quantitatively related to the T-shaped equivalent circuit parameters of each phase winding of the transformer, including: Differential current model prediction value Relationship with voltage: The relationship between the predicted value of the differential current model and the high-voltage side line voltage and the low-voltage side phase voltage is: ; in, Indicates the Impedance of phase high voltage winding; Indicates the The impedance of the phase low voltage winding referred to the high voltage side; Indicates the Impedance of the phase excitation branch; Indicates the turns ratio of the transformer high voltage side to the low voltage side ( , is the number of turns of the high voltage winding, is the number of turns of the low voltage winding); The differential current model parameters are related to the impedance of the transformer T-shaped equivalent circuit: 、 The associated expressions with the transformer T-shaped equivalent circuit parameters are: .

[0012] As a further improvement of the present technical solution, generating the virtual differential current in S400 includes the following steps: S410.1. Extracting the comparison amount: Get the first Measured differential current of phase , and the S300 Predicted values ​​of the phase differential current model ; S410.2. Calculate the deviation modulus: According to the formula , calculate the The virtual differential current of the phase, where: The modular operation of complex phasors is used to convert phasor differences into scalar amplitudes and quantify the degree of deviation. For the Moment of presence The virtual differential current is used to reflect the deviation between the measured value and the model predicted value.

[0013] As a further improvement of the present technical solution, the filtering process of the virtual differential current in S400 includes the following steps: S420.1, yes Using sliding window mean filtering, the formula is: ; in, Indicates the Virtual differential current after phase filtering; Indicates the sliding window size; Represents the backtracking index within the window; Indicates the Phase at the historical sampling moment The original virtual differential current; Indicates the current sampling time; S420.2, the filtered As an input parameter for internal defect determination in S500.

[0014] As a further improvement of the present technical solution, the setting of the preset threshold and the determination of internal defects in S500 include the following steps: S500.1. Determination of Preset Threshold: The preset threshold is calculated based on the defect-free training sample in S300, and the model virtual differential current phasor after each phase is filtered by the sliding window mean in S420 during the training phase , determined as follows: Extract training cycle The maximum value of the amplitude is recorded as ; No. Phase preset threshold Defined as: ;in, is the threshold coefficient, ranging from 1.2 to 1.5, used to suppress normal operation fluctuation interference; S500.2, Internal Defect Determination and Type Inference: When Amplitude of virtual differential current after phase filtering When the phase is determined to have an internal defect, the defect type is further inferred based on the correlation between the differential current model parameters in S300 and the transformer T-shaped equivalent circuit. The defect types include: Winding turn short circuit: If At the same time, it deviates significantly from the training value (reflecting the sudden change in the winding impedance distribution), and Continue to increase (corresponding to the gradual change in the number of short-circuit turns); Phase short circuit (including lead wire short circuit): If multiple phases (such as or equal phase combination) At the same time, the standards are exceeded (reflecting abnormalities in the interphase electrical circuit); The winding is short-circuited to ground: A single abnormality (sudden change in the excitation branch characteristics, corresponding to ground insulation damage), and It shows a step-like growth (characteristic of instantaneous insulation breakdown).

[0015] The second object of the present invention is to provide a distribution transformer internal defect diagnosis device based on virtual differential current, which is equipped with a distribution transformer internal defect diagnosis system based on virtual differential current. When the computer program of the distribution transformer internal defect diagnosis system based on virtual differential current is run, it is used to execute any step of the above-mentioned distribution transformer internal defect diagnosis method based on virtual differential current.

[0016] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention synchronously collects tap position information and uses it to determine the turns ratio. It then calculates the measured differential current based on the zero-sequence current on the star-connected side. This method can adapt to the operating conditions of dynamic adjustment of transformer tap positions, reduce fault misjudgments caused by turns ratio changes or zero-sequence current interference, and improve diagnostic adaptability.

[0017] 2. Based on the operating data of a defect-free transformer, this invention constructs a differential current model that associates winding impedance with excitation branch parameters. This model is tied to the physical topological characteristics of the transformer. The virtual differential current (the deviation between the measured value and the model's predicted value) can reflect the actual state changes of the transformer's internal windings, enhancing the physical relevance and reliability of defect diagnosis.

[0018] 3. By calculating the virtual differential current and performing filtering, the present invention can reduce the impact of instantaneous fluctuations on the diagnostic results, making defect determination more stable. At the same time, by inferring the defect type based on the amplitude of the virtual differential current and changes in model parameters, the specific type of defect can be further distinguished based on the determination of the existence of a defect, thereby improving the targeted diagnosis. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 Schematic diagram of the method steps of the present invention. DETAILED DESCRIPTION

[0020] The following will provide a clear and complete description of the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0021] like Figure 1 As shown, this embodiment provides a method for diagnosing internal defects of a distribution transformer based on virtual differential current, comprising the following steps: S100, synchronous data acquisition: real-time acquisition of the three-phase line voltage and line current on the high-voltage side of the transformer, the three-phase phase voltage and line current on the low-voltage side, and the tap position information; It is understood that the synchronous data collection in this embodiment is achieved through an acquisition system consisting of a sensor module, a synchronous clock module, a tap information acquisition module, and a data processing unit. The sensor module collects electrical signals, the synchronous clock module ensures the time consistency of each signal, the tap information acquisition module acquires gear parameters, and the data processing unit is responsible for signal conversion, correlation, and storage. These modules work together to meet the data "synchronization" and "integrity" requirements of S100.

[0022] In this step, the synchronous data collection of S100 includes the following steps: S100.1. Real-time collection of the three-phase line voltage and line current on the high-voltage side of the transformer and the three-phase phase voltage and line current on the low-voltage side, with a collection frequency of not less than 50 Hz; As a further explanation of this step, the electrical quantity acquisition in the sensor module of this embodiment specifically includes: a high-voltage side sensor for acquiring the high-voltage side three-phase line voltage and the high-voltage side three-phase line current, and a low-voltage side sensor for acquiring the low-voltage side three-phase line voltage and the low-voltage side three-phase line current; wherein, the high-voltage side sensor is installed at the high-voltage side outlet of the transformer, and the low-voltage side sensor is installed at the low-voltage side outlet of the transformer, and both use voltage and current sensors that can adapt to the fundamental frequency of the power grid.

[0023] S100.2. Synchronously obtain transformer tap position information, and determine the nominal turns ratio of the transformer (the ratio of the nominal number of turns of the high-voltage winding to the nominal number of turns of the low-voltage winding) based on the transformer tap position information. As a further explanation of this step, the synchronization time stamp implementation of the synchronization clock module of this embodiment includes the following steps: first, the synchronization clock module adopts a global synchronization clock source (such as a Beidou synchronization clock); then, a unified timestamp (accurate to milliseconds) is assigned to the sampling time of the sensor module and the tap information acquisition module; then, it is ensured that the voltage and current data of the high-voltage side and the low-voltage side, as well as the tap gear information are collected at the same timestamp; next, the sampling time difference of each module is eliminated by timestamp comparison; finally, the synchronized timestamp is embedded in all collected data to provide a time reference for subsequent data association.

[0024] Furthermore, as a further explanation of this embodiment, the gear information acquisition in the tap information acquisition module of this embodiment specifically includes: the tap information acquisition module is connected to the gear feedback device (such as a mechanical contact encoder or an electronic gear sensor) of the transformer tap changer through a communication interface (such as RS485) to receive the current gear value (such as 1 to n gears, where n is the total number of tap gears) in real time; at the same time, the module pre-stores the ratio of different gears to the nominal turns in the transformer factory parameters. In addition, the nominal turns ratio determination of the tap information acquisition module of this embodiment includes the following steps: first, receiving the real-time gear position value transmitted by the gear position feedback device; then, determining whether the gear position value is stable (e.g., no change for 50 ms); then, if it is stable, matching and calling the nominal turns ratio K corresponding to the gear position from a pre-stored correspondence; next, if the gear position value is unstable (e.g., during a switching process), temporarily using the K value of the previous stable gear; finally, updating the K value after the gear position stabilizes to ensure the accuracy of the turns ratio used for calculation.

[0025] S100.3. Associate the voltage and current data collected in S100.1 with the nominal turns ratio determined in S100.2 and store them as original input parameters for calculating the real-time differential current of each phase in S200.

[0026] As a further explanation of this step, the voltage and current data conversion in the data processing unit of this embodiment specifically includes: the analog signal (voltage, current) collected by the sensor module is converted into a digital signal by the analog-to-digital conversion module; the high-voltage side line voltage after conversion is recorded as (time domain instantaneous value, dynamically changing with time), the line current is recorded as ; The phase voltage on the low voltage side is recorded as Since the low voltage side adopts star connection, its line current is consistent with the phase current, so the low voltage side line current is synchronously recorded as During the conversion process, second-order low-pass filtering is used to remove high-frequency noise (such as switching transients and harmonic interference) and retain the 50Hz fundamental component, ensuring that the digital signal reflects the time domain characteristics of the actual electrical quantity and providing reliable input for subsequent fundamental component extraction (such as Fourier transform).

[0027] As a further explanation of this step, the data association storage of the data processing unit of this embodiment includes the following steps: First, extract the timestamp assigned by the synchronization clock module; Then, the high-side voltage, high-side current, low-side voltage, low-side current at the same time stamp, and the nominal turns ratio determined by the tap information acquisition module are calculated. to integrate; Then, a data set containing the above parameters and timestamps is formed; Next, the dataset is verified (if the timestamp deviation exceeds 5ms, it is marked as invalid); Finally, the valid data set is stored in the time series database to provide the original input parameters for S200.

[0028] S200, calculating the measured differential current: generating a real-time differential current of each phase based on the high-voltage side line current, the low-voltage side line current, and the turns ratio determined by the tap position, combined with the zero-sequence current on the star connection side; In this step, in S200, calculating the measured differential current includes the following steps: S200.1. Extracting synchronous measurement parameters: From the synchronously collected data of S100, extract: High voltage side line current: (three-phase line current); Low voltage side line current: (three-phase line current); Turns ratio parameters: (The ratio of the nominal number of turns of the transformer high-voltage winding to the nominal number of turns of the low-voltage winding is determined by the tap position); As a further explanation of this step, in this embodiment, S200.1 synchronous measurement parameter extraction specifically includes: extracting valid parameters associated with the same timestamp from the time series database stored in S100 (verified in S100.3, the timestamp deviation is ≤5ms), including: high-voltage side line current (denoted as , are all 50Hz fundamental wave phasors, reflecting Phase line current effective value and phase); low voltage side line current (denoted as , 50Hz fundamental phasor, equal to the phase current in star connection); turns ratio parameter (denoted as , that is, the nominal number of turns of the high-voltage winding Nominal number of turns of low voltage winding The ratio of , determined by S100.2 matching). The time synchronization of the above parameters can avoid phase deviation caused by sampling time difference and ensure a reliable calculation basis.

[0029] S200.2. Calculate the high voltage side line current components: The three-phase components are derived from the three-phase line current on the high-voltage side: ;in, 、 、 Represents the high-voltage side three-phase current component derived from the high-voltage side three-phase line current; As a further explanation of this step, S200.2 of this embodiment, high-voltage side line current component calculation includes the following steps: First, the calculation basis is clear: when the high-voltage side adopts the delta (Δ) connection method, there is a specific relationship between the line current and the phase current, that is, the phase current can be derived from the line current; Then, call S200.1 to extract , calculated as follows: ; Then, perform the above phasor difference calculation (including amplitude and phase calculation) to ensure that the result conforms to the current distribution law of the delta connection; Finally, the calculated three-phase current components are temporarily stored for use in the differential current synthesis of S200.4.

[0030] S200.3. Calculate the zero-sequence current on the star connection side: If the low voltage side is star-connected, calculate the zero sequence current on the low voltage side : ; As a further explanation of this step, the calculation of the zero-sequence current on the star connection side in S200.3 of this embodiment includes the following steps: First, determine the low-voltage side wiring method (based on the transformer factory parameters preset to star connection); Then, the physical meaning of zero-sequence current is clarified. In the star connection, the zero-sequence current is the zero-sequence component of the three-phase current (equal in magnitude and phase), and its value can reflect asymmetric defects such as single-phase grounding. Next, call S200.1 to extract (In star connection, line current = phase current), calculate zero sequence current according to the following formula : ; Finally, if the low voltage side is a delta connection (no zero sequence current path), Assign a value of 0 to avoid invalid compensation.

[0031] S200.4. Calculate the measured differential current of each phase: The measured differential current of each phase is generated according to the following formula: ;in, 、 、 Indicates the final calculated measured differential current of each phase.

[0032] As a further explanation of this step, the calculation of the measured differential current of each phase in S200.4 of this embodiment includes the following steps: First, based on the transformer "ampere-turn balance" principle, the formula is derived - in an ideal state, the high-voltage side ampere-turn and low voltage side ampere-turns Balance, that is , the actual differential current is the superposition of the difference between the two and the zero-sequence compensation term; Then, the measured differential current of each phase is calculated according to the following formula: ; Then, the phasor operation in the formula (such as The compensation amount for zero-sequence current converted to high-voltage side is calculated in complex form (real part + imaginary part) to ensure accuracy; Finally, the calculated measured differential current of each phase is associated with the timestamp and stored to provide input for S300 to build a health model.

[0033] For example, Taking the phase as an example, the calculation steps are as follows: First, call S200.2 to calculate 、Calculated by S200.3 , S200.1 extracted and ; Then, the zero-sequence compensation term is calculated (Phasor divided by scalar , only the amplitude is changed); Then, calculate the low voltage side reference current ; Next, press the formula , perform phasor operations (including amplitude and phase superposition); Finally, the same calculation Mutually( )and Mutually( ), the measured differential current of each phase is obtained and stored as the input parameter for subsequent S300 modeling.

[0034] Furthermore, to ensure the reliability of the measured differential current, the following processing needs to be added during the calculation process: Accuracy guarantee of phasor operation: All phasor operations (such as ) Use complex form (real part + imaginary part) for calculation to avoid errors caused by amplitude approximation; perform range check on the calculation results (for example, if the differential current amplitude does not exceed 5% of the transformer rated current, it will be marked as abnormal and trigger recalculation); Connection verification with S100: If the data at a certain moment in S100 is marked as "invalid" (such as a sensor failure), the S200 calculation at that moment is skipped to avoid invalid input affecting the result; after the calculation is completed, the measured differential current of each phase is associated with the corresponding timestamp and stored to ensure that it matches the training sample called by S300 in the time dimension.

[0035] S300, establishing a differential current model: using the operating data of the transformer when it is defect-free as training samples, the training samples include the measured differential current of each phase in the defect-free state generated in S200, the high-voltage side line voltage and the low-voltage side phase voltage obtained in S100, based on the transformer T-shaped equivalent circuit, correlating the winding impedance distribution with the excitation branch parameters, and obtaining the differential current model of the transformer when it is defect-free through training; In this step, in S300, the training samples are time-stamped synchronized phasor data acquired during the trial operation, including: No. Measured differential current of phase ; High voltage side line voltage: Phase voltage 、 Phase voltage 、 Phase voltage ; Low voltage side phase voltage: Phase voltage 、 Phase voltage 、 Phase voltage ; The operating data of the transformer when it is defect-free is steady-state operating data without faults for more than 72 consecutive hours during the trial operation.

[0036] As a further explanation of this step, in S300 of this embodiment, "the training sample is steady-state operation data without faults for more than 72 consecutive hours during the trial operation", the judgment criteria and data screening method of "steady-state operation data" are as follows: Determination of steady-state operation data: Based on the transformer's rated parameters, we set the steady-state thresholds: the high-voltage line voltage fluctuation range must not exceed ±5% of the rated value, the high-voltage line current fluctuation range must not exceed ±10% of the rated value, and the duration must be ≥ 1 minute (to avoid transient fluctuations). The time period that meets these conditions is designated as the "steady-state period," and the data extracted from this period serves as the training sample.

[0037] Data filtering operations: First, the raw data of 72 consecutive hours during the trial operation (according to the S100 acquisition frequency, such as 50Hz, corresponding to about 1.8×10 6 Sampling points) are segmented to identify all steady-state periods; Then, outliers in the steady-state period are eliminated (e.g., points where the voltage and current increase or decrease suddenly beyond the threshold); Then, time series are evenly extracted from the remaining valid data (ensuring coverage of different load levels, such as light load and full load) to form the final training sample to avoid data bias.

[0038] In this step, the step of obtaining the differential current model of the transformer without defects through training in S300 includes the following steps: S300.1. Data preparation: During the trial run, the time series ( is the number of sampling points), collect: Phase: High voltage side line voltage , low voltage side phase voltage , measured differential current ; Phase: High voltage side line voltage , low voltage side phase voltage , measured differential current ; Phase: High voltage side line voltage , low voltage side phase voltage , measured differential current ; As a further explanation of this step, the time series in this embodiment The sampling interval is consistent with S100, is the number of sampling points (72 hours steady-state data is calculated at 50Hz, The value can be 1×10 5 ~2×10 5 , ensuring sufficient data volume).

[0039] For example, taking phase A as an example, the collected data is organized as follows: High-voltage side line voltage : No. time Fundamental phasor of phase-to-phase voltage (effective value); Low voltage side phase voltage : No. time Fundamental phasor of phase-to-phase voltage (effective value); Measured differential current : No. time Measured differential current of each phase (calculated by S200.4).

[0040] right Mutually( 、 、 )and Mutually( 、 、 ) are organized according to the same logic to ensure that the data of each phase strictly corresponds to the timestamp.

[0041] S300.2. Construct matrices and vectors: For the first Phase, structure: Input matrix: ;in, For the Phase high voltage side line voltage; For the Phase voltage on the low voltage side; For the Sampling moments; Output vector: ;in, For the Phase measured differential current; Parameter vector: ;in, For the The coefficients of line voltage in the phase model; For the The coefficients of phase voltage in the phase model; As a further explanation of this step, the phasor construction of the matrix and vector construction of this embodiment includes the following steps: First, from the fundamental wave extraction results of S100, filter the voltage phasor data of the steady-state period for 72 consecutive hours: Get the Mutually( )High voltage side line voltage fundamental phasor (like Xiangwei , reflecting the amplitude and phase characteristics of the 50Hz line voltage); Get the Mutually( ) Low voltage side phase voltage fundamental phasor (like Xiangwei , the physical meaning of the star connection is consistent with the low-voltage side line current phasor); Then, from the calculation results of S200, the first Mutually( ) Measured differential current fundamental phasor: Call S200 to calculate the fundamental phasor and get the Moment ; Then, by time series ( is the number of sampling points, matching the S100 acquisition frequency), construct the input matrix With the output vector ; Next, define the parameter vector ; The completed construction 、 、 Associated Storage: Make sure the matrix dimensions match ( for , for , for ) provides input for the least squares parameter identification of S300.3.

[0042] Furthermore, in this embodiment, the voltage, current, and differential current all use fundamental phasor symbols. The core reasons are as follows: Model adaptability: The S300 differential current model describes the linear relationship under steady-state conditions (in the absence of defects, the fundamental component of the differential current and voltage is linearly related); Data consistency: The instantaneous values ​​collected by S100 have had their 50Hz fundamental components extracted (converted to phasors) through FFT. The differential current calculation of S200 is also based on fundamental phasor operations. Therefore, the input of S300 and the model output are completely consistent in form, ensuring the operability of the algorithm.

[0043] S300.3. Least squares method for parameter identification: ; in, for The conjugate transposed matrix of ; for The inverse matrix of As a further explanation of this step, the least squares parameter calculation in this embodiment includes the following steps: First, the identification goal is to determine the parameter vector by minimizing the sum of squares of the error between the model prediction value and the measured differential current. ( ),Right now ,in ; Then, the parameter formula is derived: take the derivative of the sum of squared errors and set it to 0, simplifying to ; for The conjugate transposed matrix of ; for The inverse matrix of Then, calculate the matrix : For real data, ( ); Next, calculate the inverse matrix ; If the matrix determinant is not 0 (guaranteed by sufficient sample size), the inverse matrix is ( are diagonal elements, are non-diagonal elements); Finally, solve for the parameter vector: ,Right now , and obtain the model parameters of each phase.

[0044] S300.4. Establish the differential current model of each phase: ; in, For the The model predicts the differential current values ​​of the phases.

[0045] As a further explanation of this step, the differential current model of each phase established in S300.4 in this embodiment is , reflecting the linear relationship between the differential current and voltage in the defect-free state—where Quantify the influence weight of high-voltage side line voltage on differential current, Quantify the influence weight of the low-voltage side phase voltage. The role of this model is to provide a "health benchmark" for real-time diagnosis: by inputting the real-time voltage 、 , the predicted differential current when there is no defect can be obtained, which provides a basis for the subsequent calculation of the "deviation between the measured value and the predicted value (virtual differential current)".

[0046] In this step, the differential current model parameters established in S300 are ), which is quantitatively related to the T-shaped equivalent circuit parameters of each phase winding of the transformer, including: Differential current model prediction value Relationship with voltage: The relationship between the predicted value of the differential current model and the high-voltage side line voltage and the low-voltage side phase voltage is: ; in, Indicates the Impedance of phase high voltage winding; Indicates the The impedance of the phase low voltage winding referred to the high voltage side; Indicates the Impedance of the phase excitation branch; Indicates the turns ratio of the transformer high voltage side to the low voltage side ( , is the number of turns of the high voltage winding, is the number of turns of the low voltage winding); The differential current model parameters are related to the impedance of the transformer T-shaped equivalent circuit: 、 The associated expressions with the transformer T-shaped equivalent circuit parameters are: .

[0047] As a further explanation of this step, the derivation of the association between the model parameters and the T-shaped equivalent circuit of this embodiment includes the following steps ( Phase as an example): First, the low-voltage side parameters of the transformer are converted to the high-voltage side (a common method for engineering analysis that simplifies cross-side circuit calculations). The rules are: Voltage calculation: low voltage side phase voltage Calculated to the high pressure side ; Current calculation: low voltage side current Calculated to the high pressure side ; Impedance calculation: actual impedance of low voltage winding Calculated to the high pressure side (Impedance is proportional to the square of the turns ratio); Then, based on the calculated circuit (high voltage side perspective), write The electrical equations of the phases: High voltage side line voltage: ;in, is the high voltage side current, is the high voltage winding impedance, is the excitation branch voltage; Excitation branch: ; is the excitation current, is the excitation branch impedance; Ampere-turn balance: ; High voltage side current = excitation current + calculated low voltage side current; Low-voltage side return circuit: ; Used to verify the consistency of the reduction, and the subsequent modeling directly uses the reduced parameters; Then, combined with the approximate relationship of the differential current in the absence of defects in S200 ( ), the above equations are combined to eliminate 、 、 : from and , eliminate : ; Substitution ,get: ; Solve from the low-voltage side reference circuit , substitute into the above formula: ; Solve (Excitation current, reflecting the excitation characteristics of the transformer): ; Next, combined with the measured differential current definition of S200 ( ), zero sequence current when there is no defect ,and (In the triangle connection on the high voltage side, the phase current is approximately equal to the excitation current component), so ; Will Substituting in and simplifying, we get: ; Finally, the model formula with S300 , and compare them to get: ; Similarly, the impedance of phases B and C can be obtained by replacing the subscripts (A→B→C). The derivation shows that the model parameters directly map the impedance characteristics of the transformer T-shaped equivalent circuit, ensuring that the model has both data fitting accuracy and physical rationality.

[0048] S400, generating a virtual differential current: based on the differential current model established in S300, calculating the deviation between the measured differential current of each phase generated in S200 and the model predicted value as the virtual differential current; and filtering the virtual differential current; In this step, generating the virtual differential current in S400 includes the following steps: S410.1. Extracting the comparison amount: Get the first Measured differential current of phase , and the S300 Predicted values ​​of the phase differential current model ; As a further explanation of this step, the data association and format in the comparison amount extraction of this embodiment specifically include: Extracted Phase measured differential current Calculation results from S200, and timestamp (accurate to milliseconds) binding, the sampling interval is consistent with S100 (such as 50Hz corresponds to 20ms); Phase differential current model prediction value By calling the model formula of S300 , generates; among them 、 Taken from S100 same timestamp Real-time data is collected and the time synchronization between the two is ensured (deviation ≤ 5ms).

[0049] S410.2. Calculate the deviation modulus: According to the formula , calculate the The virtual differential current of the phase, where: The modular operation of complex phasors is used to convert phasor differences into scalar amplitudes and quantify the degree of deviation. For the Moment of presence The virtual differential current is used to reflect the deviation between the measured value and the model predicted value.

[0050] As a further explanation of this step, the specific steps of calculating the deviation modulus value in this embodiment include: First, define the input parameters: Measured differential current of phase (Complex phasor, including real part and the imaginary part ) and model predictions (Complex phasor, including real part and the imaginary part ); Then, calculate the phasor difference: ;in is an imaginary unit; Then, the complex phasor is converted into a scalar magnitude through modular operation, the formula is: ; Next, compare the calculated result with the timestamp Associated storage to form an original virtual differential current sequence; Finally, the non-negativity of the modulus value is checked (because the square root of the sum of squares is always ≥ 0) to ensure data validity.

[0051] In this step, the filtering process of the virtual differential current in S400 includes the following steps: S420.1, yes Using sliding window mean filtering, the formula is: ; in, Indicates the Virtual differential current after phase filtering; Indicates the sliding window size; Represents the backtracking index within the window; Indicates the Phase at the historical sampling moment The original virtual differential current; Indicates the current sampling time; As a further explanation of this step, the sliding window setting in the filtering process of this embodiment specifically includes: Sliding window size Based on the selection of the grid fundamental wave period, for example, when the sampling frequency is 50Hz (sampling interval 20ms), The value can be 5 (corresponding to 100ms, covering 5 consecutive sampling points), which can filter out transient noise (such as short-term disturbances caused by switching operations) while retaining the continuous deviation caused by defects; in the formula is the backtracking index within the window ( =0 corresponds to the current moment , =1 corresponds to −20ms, and so on).

[0052] S420.2, the filtered As an input parameter for internal defect determination in S500.

[0053] As a further explanation of this step, the data transmission in the filtering result output of this embodiment specifically includes: Filtered virtual differential current Stored as scalar amplitude, each record contains a timestamp and 、 、 Three-phase filtering value The data is transmitted to S500 in real time through the internal interface, providing stable and continuous input parameters for defect determination, ensuring that subsequent steps can accurately identify changes in the internal state of the transformer based on the deviation signal.

[0054] S500, defect determination: when the virtual differential current of any phase exceeds a preset threshold, it is determined that the transformer has an internal defect.

[0055] In this step, the setting of the preset threshold and the internal defect determination in S500 include the following steps: S500.1. Determination of Preset Threshold: The preset threshold is calculated based on the defect-free training sample in S300, and the model virtual differential current phasor after each phase is filtered by the sliding window mean in S420 during the training phase , determined as follows: Extract training cycle The maximum value of the amplitude is recorded as ; No. Phase preset threshold Defined as: ;in, is the threshold coefficient, ranging from 1.2 to 1.5, used to suppress normal operation fluctuation interference; As a further explanation of this step, the training sample processing in the threshold calculation of this embodiment specifically includes: the training sample used to calculate the threshold is the virtual differential current after filtering in S420 of the "72 hours of continuous defect-free steady-state data" in S300 , the sample needs to be preprocessed first: remove the instantaneous outliers (such as isolated peaks caused by short-term sensor interference, the judgment standard is the point that exceeds 3 times the standard deviation of the sample mean), retain the valid data reflecting normal operation fluctuations, and ensure the accuracy of subsequent maximum amplitude extraction.

[0056] Furthermore, the threshold determination step in the threshold calculation of this embodiment includes: First, extract the first The filtered virtual differential current sequence of the phase ( is the number of training sample points); Then, calculate the maximum amplitude of the sequence , reflecting the maximum normal fluctuation in a defect-free state; Then, based on the threshold coefficient Calculate the preset threshold using the formula: ;in The value of is based on: by analyzing the defect-free operation data of more than 10 groups of similar transformers, it is found that the peak value of normal fluctuation usually does not exceed 1.2 times, and setting =1.2-1.5 can cover more than 99% of normal fluctuations while avoiding misjudgment due to slight interference; Next, the calculated Stored as the judgment threshold of each phase; Finally, the thresholds are updated regularly (e.g., every 6 months) based on new, defect-free steady-state data to accommodate normal fluctuations due to transformer aging.

[0057] S500.2, Internal Defect Determination and Type Inference: When Amplitude of virtual differential current after phase filtering When the phase is determined to have an internal defect, the defect type is further inferred based on the correlation between the differential current model parameters in S300 and the transformer T-shaped equivalent circuit. The defect types include: Winding turn short circuit: If At the same time, it deviates significantly from the training value (reflecting the sudden change in the winding impedance distribution), and Continue to increase (corresponding to the gradual change in the number of short-circuit turns); Phase short circuit (including lead wire short circuit): If multiple phases (such as or equal phase combination) At the same time, the standards are exceeded (reflecting abnormalities in the interphase electrical circuit); The winding is short-circuited to ground: A single abnormality (sudden change in the excitation branch characteristics, corresponding to ground insulation damage), and It shows a step-like growth (characteristic of instantaneous insulation breakdown).

[0058] As a further explanation of this step, the threshold comparison logic of this embodiment specifically includes: Real-time collection Virtual differential current after phase filtering , and compare its amplitude with the preset threshold For comparison: like , it is determined that the phase has no defects; like , and the duration of this state is ≥ 2 fundamental wave cycles (such as 40ms, to avoid instantaneous interference), it is determined that there is an internal defect in this phase.

[0059] As a further explanation of this step, the defect type inference of this embodiment specifically includes the following steps: First, define three types of input data: Model parameter baseline value: the parameters obtained by calling the defect-free steady-state training sample in S300 ;in, Indicates the Reference value of the weight coefficient of the phase high-voltage side line voltage to the differential current (typical value when there is no defect); Indicates the Reference value of the weight coefficient of the phase voltage on the low-voltage side of the phase to the differential current (typical value when there is no defect); Real-time operating parameters: obtained through online rolling identification (updated every 100ms, reusing the least squares method of S300) , reflecting the voltage-differential current correlation characteristics of the transformer under the current operating state.

[0060] Virtual differential current: extract the current phase filtered amplitude output by S420 and historical sequences (nearly five sampling points, 20ms interval, covering a 100ms time window) to analyze the dynamic trend of the deviation.

[0061] The relative deviation formula is then used to quantify the degree of deviation between the real-time parameter and the baseline value: ; ; Indicates the Model parameters identified in real time (dynamically changing with operating status); 、 Indicates the baseline value of the model parameters in a defect-free state (as a "healthy reference system" and needs to be updated regularly); 、 Indicates the relative deviation between two parameters (dimensionless, directly reflects the proportion of parameter changes); Judgment threshold: set This is the "significant deviation" standard (based on transformer fault simulation tests, parameter deviations under defective conditions usually exceed this range, not empirical values) Then, combining the parameter deviation characteristics with the dynamic laws of the virtual differential current, the defect types are matched item by item: Winding inter-turn short circuit: , It shows a monotonically increasing trend at 5 consecutive sampling points (100ms); Phase short circuit (including lead wire short circuit): at least two phases When, and multi-phase 、 The changing trend is consistent; Internal short circuit to ground in winding: , In one sampling point (20ms), it suddenly increases to 2 times or more.

[0062] Next, generate structured diagnostic results: bind the matched defect type (such as "B-phase winding turn-to-turn short circuit") with the current timestamp and record 、 and the dynamic characteristics of the virtual differential current (such as increasing slope and sudden increase multiple); Finally, perform cross validation: Combined with transformer auxiliary monitoring signals (such as oil temperature and gas relay status): If the auxiliary signal synchronization is abnormal (such as the oil temperature rises by 5°C or more within 1 minute), a defect is confirmed; If the auxiliary signal is normal, it is marked as a "suspected defect" and a secondary inspection is triggered (extending the observation to 3 fundamental wave cycles, a total of 60ms, to eliminate instantaneous interference) to avoid false alarms based on a single criterion.

[0063] It should be added that, to ensure the feasibility of defect determination, this embodiment further includes: Parameter online identification: real-time parameters in S500 This is achieved by rolling the least squares method of S300.3, updating every 100ms to ensure that parameter changes can be captured in a timely manner; Threshold adaptive adjustment: If the transformer is in a special working condition (such as overload, tap change), the threshold coefficient Increased to 1.8 to avoid misjudgment; historical data backtracking: the system automatically stores the virtual differential current and parameter data of the past 30 days to facilitate post-analysis of defect types and optimization of judgment logic.

[0064] This embodiment also provides a distribution transformer internal defect diagnosis device based on virtual differential current, which is equipped with a distribution transformer internal defect diagnosis system based on virtual differential current. The computer program equipped with the distribution transformer internal defect diagnosis system based on virtual differential current is used to execute the above-mentioned distribution transformer internal defect diagnosis method based on virtual differential current when running.

[0065] Those skilled in the art will appreciate that the process of implementing all or part of the steps of the above embodiments may be accomplished by hardware, or by instructing related hardware through a program.

[0066] The above shows and describes the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions are merely preferred examples of the present invention and are not intended to limit the present invention. Various changes and improvements may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and improvements fall within the scope of the present invention. The scope of protection claimed in the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for diagnosing internal defects of distribution transformers based on virtual differential current, characterized in that: The steps include: S100, synchronous data acquisition: real-time acquisition of the three-phase line voltage and line current on the high-voltage side of the transformer, the three-phase phase voltage and line current on the low-voltage side, and the tap position information; S200, calculating the measured differential current: generating a real-time differential current of each phase based on the high-voltage side line current, the low-voltage side line current, and the turns ratio determined by the tap position, combined with the zero-sequence current on the star connection side; S300, establishing a differential current model: using the operating data of the transformer when it is defect-free as training samples, the training samples include the measured differential current of each phase in the defect-free state generated in S200, the high-voltage side line voltage and the low-voltage side phase voltage obtained in S100, based on the transformer T-shaped equivalent circuit, correlating the winding impedance distribution with the excitation branch parameters, and obtaining the differential current model of the transformer when it is defect-free through training; S400, generating a virtual differential current: based on the differential current model established in S300, calculating the deviation between the measured differential current of each phase generated in S200 and the model predicted value as the virtual differential current; and filtering the virtual differential current; S500, defect determination: when the virtual differential current of any phase exceeds a preset threshold, it is determined that the transformer has an internal defect.

2. The method for diagnosing internal defects of distribution transformers based on virtual differential current according to claim 1, characterized in that: The synchronous data collection of S100 includes the following steps: S100.

1. Real-time collection of the three-phase line voltage and line current on the high-voltage side of the transformer and the three-phase phase voltage and line current on the low-voltage side, with a collection frequency of not less than 50 Hz; S100.

2. Synchronously obtain transformer tap position information, and determine the nominal turns ratio of the transformer based on the transformer tap position information; S100.

3. Associate the voltage and current data collected in S100.1 with the nominal turns ratio determined in S100.2 and store them as original input parameters for calculating the real-time differential current of each phase in S200.

3. The method for diagnosing internal defects of a distribution transformer based on virtual differential current according to claim 2, characterized in that: In S200, calculating the measured differential current includes the following steps: S200.

1. Extracting synchronous measurement parameters: From the synchronously collected data of S100, extract: High voltage side line current: ; Low voltage side line current: ; Turns ratio parameters: ; S200.

2. Calculate the high voltage side line current components: The three-phase components are derived from the three-phase line current on the high-voltage side: ;in, Represents the high-voltage side three-phase current component derived from the high-voltage side three-phase line current; S200.

3. Calculate the zero-sequence current on the star connection side: If the low voltage side is star-connected, calculate the zero sequence current on the low voltage side ; S200.

4. Calculate the measured differential current of each phase: The measured differential current of each phase is generated according to the following formula: ;in, Indicates the final calculated measured differential current of each phase.

4. The method for diagnosing internal defects of a distribution transformer based on virtual differential current according to claim 3, characterized in that: In S300, the training samples are time-stamped synchronized phasor data acquired during the trial operation, including: No. Measured differential current of phase ; High voltage side line voltage: Phase voltage 、 Phase voltage 、 Phase voltage ; Low voltage side phase voltage: Phase voltage 、 Phase voltage 、 Phase voltage ; The operating data of the transformer when it is defect-free is steady-state operating data without faults for more than 72 consecutive hours during the trial operation.

5. The method for diagnosing internal defects of distribution transformers based on virtual differential current according to claim 4, characterized in that: The step of obtaining the differential current model of the transformer without defects through training in S300 includes the following steps: S300.

1. Data preparation: During the trial run, the time series ,collection: Phase: High voltage side line voltage , low voltage side phase voltage , measured differential current ; Phase: High voltage side line voltage , low voltage side phase voltage , measured differential current ; Phase: High voltage side line voltage , low voltage side phase voltage , measured differential current ; S300.

2. Construct matrices and vectors: For the first Phase, structure: Input matrix: ;in, For the Phase high voltage side line voltage; For the Phase voltage on the low voltage side; For the Sampling moments; Output vector: ;in, For the Phase measured differential current; Parameter vector: ;in, For the The coefficients of line voltage in the phase model; For the The coefficients of phase voltage in the phase model; S300.

3. Least squares method for parameter identification: ； in, for The conjugate transposed matrix of ; for The inverse matrix of S300.

4. Establish the differential current model of each phase: ； in, For the The model predicts the differential current values ​​of the phases.

6. The method for diagnosing internal defects of distribution transformers based on virtual differential current according to claim 5, characterized in that: The differential current model parameters established in S300 are ), which is quantitatively related to the T-shaped equivalent circuit parameters of each phase winding of the transformer, including: Differential current model prediction value Relationship with voltage: The relationship between the predicted value of the differential current model and the high-voltage side line voltage and the low-voltage side phase voltage is: ； in, Indicates the Impedance of phase high voltage winding; Indicates the The impedance of the phase low voltage winding referred to the high voltage side; Indicates the Impedance of the phase excitation branch; Indicates the turns ratio between the high-voltage side and the low-voltage side of the transformer; The differential current model parameters are related to the impedance of the transformer T-shaped equivalent circuit: 、 The associated expressions with the transformer T-shaped equivalent circuit parameters are: 。 7. The method for diagnosing internal defects of distribution transformers based on virtual differential current according to claim 6, characterized in that: Generating the virtual differential current in S400 includes the following steps: S410.

1. Extracting the comparison amount: Get the first Measured differential current of phase , and the S300 Predicted values ​​of the phase differential current model ; S410.

2. Calculate the deviation modulus: According to the formula , calculate the The virtual differential current of the phase, where: The modular operation of complex phasors is used to convert phasor differences into scalar amplitudes and quantify the degree of deviation. For the Moment of presence The virtual differential current is used to reflect the deviation between the measured value and the model predicted value.

8. The method for diagnosing internal defects of distribution transformers based on virtual differential current according to claim 7, characterized in that: The filtering process of the virtual differential current in S400 includes the following steps: S420.1, yes Using sliding window mean filtering, the formula is: ； in, Indicates the Virtual differential current after phase filtering; Indicates the sliding window size; Represents the backtracking index within the window; Indicates the Phase at the historical sampling moment The original virtual differential current; Indicates the current sampling time; S420.2, the filtered As an input parameter for internal defect determination in S500.

9. The method for diagnosing internal defects of distribution transformers based on virtual differential current according to claim 8, characterized in that: The setting of the preset threshold and the internal defect determination in S500 include the following steps: S500.

1. Determination of Preset Threshold: The preset threshold is calculated based on the defect-free training sample in S300, and the model virtual differential current phasor after each phase is filtered by the sliding window mean in S420 during the training phase , determined as follows: Extract training cycle The maximum value of the amplitude is recorded as ; No. Phase preset threshold Defined as: ;in, is the threshold coefficient; S500.2, Internal Defect Determination and Type Inference: When Amplitude of virtual differential current after phase filtering When the phase is determined to have an internal defect, the defect type is further inferred based on the correlation between the differential current model parameters in S300 and the transformer T-shaped equivalent circuit. The defect types include: Winding turn short circuit: If At the same time, it deviates significantly from the training value, and Continued growth Phase short circuit: If there are multiple phases Exceeding the standard at the same time; The winding is short-circuited to ground: Single abnormality, and It shows step-by-step growth.

10. A device for diagnosing internal defects of a distribution transformer based on virtual differential current, equipped with a system for diagnosing internal defects of a distribution transformer based on virtual differential current, characterized in that: The computer program of the distribution transformer internal defect diagnosis system based on virtual differential current is used to execute the steps of the distribution transformer internal defect diagnosis method based on virtual differential current according to any one of claims 1 to 9 when running.

Images