A current transformer reliability evaluation and early warning method, device and computer program product based on digital twinning

Abstract

This invention discloses a method, device, and computer program product for reliability assessment and early warning of current transformers based on digital twins. The method includes: acquiring real-time operating data of the current transformer and inputting it into a pre-constructed multi-physics digital twin model to obtain a predicted response; the model is coupled with electromagnetic, thermal, insulation, secondary circuit, and error transmission characteristics; acquiring the measured response and calculating the consistency deviation between it and the predicted response; locating the source of anomalies and distinguishing anomaly types based on the physical fields associated with the consistency deviation; calculating reliability sub-indices corresponding to different physical fields and fusing them to generate a comprehensive health index based on the consistency deviation and anomaly location and distinction results; and performing degradation pattern recognition and graded early warning. This invention can distinguish whether anomalies originate from primary body degradation, secondary circuit anomalies, sensor drift, or changes in operating conditions, improving the accuracy and traceability of condition assessment of ultra-high voltage current transformers and providing support for condition-based maintenance decisions.

Description

Technical Field

[0001] This invention relates to the field of power equipment monitoring technology, specifically to a method, device, and computer program product for reliability assessment and early warning of current transformers based on digital twins. Background Technology

[0002] Current transformers are key primary equipment interfaces in ultra-high voltage substations, responsible for relay protection, metering, fault recording, and status awareness. Their transformation accuracy, insulation reliability, secondary circuit stability, and interpretability of operating status directly affect the safety of main equipment and the correctness of protection actions. For current transformers with voltage levels of 500kV and above, long-term operation under the multi-physics field coupling effect of high voltage, high current, strong electromagnetic interference, and complex external environment can easily lead to various degradation problems, such as reduced core saturation margin, ratio and phase difference drift, abnormal secondary load, aging of main insulation, partial discharge, abnormal temperature rise, and deterioration of grounding circuit. Moreover, these problems often exhibit characteristics of mutual coupling and gradual evolution.

[0003] Current maintenance methods primarily rely on periodic power outage tests, online monitoring of single parameters, and fixed threshold alarms. While these methods can detect some significant anomalies, they struggle to reveal the coupling degradation mechanisms between multiple physical fields such as electromagnetic, thermal, and mechanical fields. They also fail to determine whether a single monitored anomaly originates from primary performance degradation, secondary circuit malfunctions, sensor drift, or changes in operating conditions. In ultra-high voltage current transformer operation scenarios, parameters such as metering error, core flux density, winding temperature rise, partial discharge, leakage current, and secondary load exhibit dynamic coupling relationships. Relying solely on a single threshold for judgment can easily lead to delayed early detection of potential hazards, a high false alarm rate, and render the resulting maintenance strategies impractical.

[0004] Existing digital twin technologies are mostly used for 3D equipment visualization, single-physics simulation, or general condition monitoring, and cannot directly meet the actual needs of improving the reliability of ultra-high voltage current transformers. This is because the core operation and maintenance objectives of current transformers differ fundamentally from those of ordinary power transformers: current transformers not only need to determine "whether a fault has occurred," but also require precise assessment of whether metering winding errors have drifted, whether the protective core is approaching saturation, whether secondary circuit characteristics have altered current transmission performance, whether there is partial discharge or surface discharge trend in the main insulation, and whether the coupling effects of different degradation factors will jointly trigger the risk of protection maloperation. Current technologies cannot yet achieve accurate characterization and comprehensive online reliability assessment of the aforementioned multi-physics coupled degradation processes, thus hindering the scientific formulation of condition-based maintenance strategies and the improvement of risk warning capabilities for ultra-high voltage current transformers. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide a method, device and computer program product for reliability assessment and early warning of current transformers based on digital twins, so as to improve the accuracy and traceability of current transformer condition assessment and provide support for condition-based maintenance decision-making.

[0006] To address the aforementioned technical problems, this invention provides a method for reliability assessment and early warning of current transformers based on digital twins, comprising: Step S1: Obtain the real-time operating data of the current transformer; Step S2: Input the real-time operating data into a pre-constructed multiphysics digital twin model to obtain a predicted response characterizing the theoretical state of the equipment; wherein, the multiphysics digital twin model is coupled with electromagnetic, thermal, insulation, secondary loop and error transmission characteristics; Step S3: Obtain the measured response of the current transformer and calculate the consistency deviation between the measured response and the predicted response; Step S4: Based on the consistency deviation, locate the source of the anomaly and distinguish the anomaly type in the physical field associated with the multiphysics digital twin model; Step S5: Based on the consistency deviation and combined with the location and differentiation results of the anomaly source, calculate multiple reliability sub-indicators corresponding to different physical fields, and fuse them to generate a comprehensive health index. Step S6: Based on the comprehensive health index and the reliability sub-indicator, perform degradation pattern identification and graded early warning.

[0007] Preferably, step S4 specifically includes: Calculate the anomaly score for each of the multiple preset components, and determine the component with the highest anomaly score as the anomaly component location result; The anomaly score is determined based on the consistency deviation vector, the weight vectors corresponding to different deviation items, and the device structure correlation.

[0008] Preferably, in step S4, the distinction of anomaly types includes: the distinction of anomalies as at least one of primary body degradation, secondary loop anomaly, sensor drift, and change in operating conditions.

[0009] Preferably, in step S5, the plurality of reliability sub-indicators corresponding to different physical fields include: metrological reliability index, electromagnetic reliability index, insulation reliability index, thermomechanical reliability index, and secondary circuit reliability index.

[0010] Preferably, the electromagnetic reliability index is determined by the core saturation margin. The characterization is calculated using the following formula:

[0011] in, The saturation magnetic flux density of the iron core; This is the absolute value of the maximum magnetic flux density of the iron core output by the multiphysics digital twin model.

[0012] Preferably, the metrological reliability index is determined by the metrological reliability margin. The characterization is calculated using the following formula:

[0013] in, For ratio difference drift, For phase difference drift, The total harmonic distortion rate of the secondary current. This is the allowable limit for the ratio difference. This refers to the allowable limit for phase difference. This refers to the allowable limit for harmonic distortion rate. , , These are the weighting coefficients, and .

[0014] Preferably, the insulation reliability index is expressed as an insulation reliability margin. The characterization is calculated using the following formula:

[0015] in, This refers to leakage current or end-screen current. For partial discharge characteristics; This refers to the dielectric loss factor. , and These are the permissible limits for leakage current, partial discharge, and dielectric loss factor, respectively. Insulation degradation trend factor; , , , These are the weighting coefficients, and .

[0016] Preferably, the method further includes: Step S7: Feed back the results of the graded early warning, the conclusions of the maintenance test, and the information of manual confirmation to the multiphysics digital twin model to update the model parameters and the reliability of the acquisition channel; The model parameters are updated according to the following formula:

[0017] in, The model parameters at the current moment, For the updated model parameters, This is a correction coefficient matrix determined based on data reliability and maintenance confirmation results; This is the measured response vector; To predict the response vector; The update of the acquisition channel reliability is performed according to the following formula:

[0018] in, For the first The credibility of each acquisition channel or feature at the current moment; For the updated credibility; The confidence level corresponding to the inspection and confirmation results; λ is the forgetting coefficient, 0 < λ < 1.

[0019] This invention also provides a current transformer reliability assessment and early warning device based on digital twins, comprising: One or more processors; Memory; One or more applications, wherein the one or more applications are stored in the memory and configured to be executed by the one or more processors, and the one or more applications are configured to perform the digital twin-based current transformer reliability assessment and early warning method.

[0020] The present invention also provides a computer program product, including computer instructions, which instruct a computer device to perform operations corresponding to the digital twin-based current transformer reliability assessment and early warning method.

[0021] The beneficial effects of this invention are as follows: By constructing a multi-physics digital twin model that couples electromagnetic, thermal, insulation, secondary circuit, and error transmission characteristics, this invention unifies scattered online monitoring data, fault waveform data, offline test data, and operation and maintenance records into reliability sub-indicators and comprehensive health indices for each physical field. This enables early identification and graded warning of potential risks such as metering error drift, core saturation margin decline, main insulation degradation, and secondary circuit anomalies in ultra-high voltage current transformers. It effectively solves the problems of high false alarm rate, low missed alarm rate, and information fragmentation between different disciplines under the coupling effect of traditional fixed threshold alarm methods. This invention can distinguish the source of anomalies as primary body degradation, secondary circuit anomalies, sensor drift, or changes in operating conditions, thus providing targeted and actionable on-site handling suggestions, including secondary load verification, error calibration, partial discharge retesting, grounding continuity testing, terminal block inspection, and power outage maintenance suggestions, reducing blind and excessive maintenance. Meanwhile, by feeding back early warning results, maintenance conclusions, and manual confirmation information in a closed loop to the digital twin model, the model parameters, feature thresholds, and prior probabilities of degradation modes are continuously updated, so that the assessment accuracy gradually improves as the equipment operates. This forms a knowledge accumulation of the reliability of similar current transformer equipment throughout its entire life cycle, providing support for the accuracy and traceability of the status evaluation of main equipment in ultra-high voltage substations, and also laying a technical foundation for the research, development, operation, and maintenance of environmentally friendly digital twin current transformers. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of the present 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 some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a flowchart illustrating a current transformer reliability assessment and early warning method based on digital twins, according to Embodiment 1 of the present invention.

[0024] Figure 2 This is a schematic diagram of the measured-twin consistency deviation and model calibration process in an embodiment of the present invention.

[0025] Figure 3 This is a schematic diagram illustrating the calculation process of reliability indicators and comprehensive health index, as well as the early warning classification process in this embodiment of the invention.

[0026] Figure 4 This is a schematic diagram of the early warning handling and twin closed-loop update process in an embodiment of the present invention. Detailed Implementation

[0027] The following description of the embodiments is taken with reference to the accompanying drawings, which illustrate specific embodiments in which the invention can be implemented.

[0028] Please refer to Figure 1 As shown, Embodiment 1 of the present invention provides a method for reliability assessment and early warning of current transformers based on digital twins, including: Step S1: Obtain the real-time operating data of the current transformer; Step S2: Input the real-time operating data into a pre-constructed multiphysics digital twin model to obtain a predicted response characterizing the theoretical state of the equipment; wherein, the multiphysics digital twin model is coupled with electromagnetic, thermal, insulation, secondary loop and error transmission characteristics; Step S3: Obtain the measured response of the current transformer and calculate the consistency deviation between the measured response and the predicted response; Step S4: Based on the consistency deviation, locate the source of the anomaly and distinguish the anomaly type in the physical field associated with the multiphysics digital twin model; Step S5: Based on the consistency deviation and combined with the location and differentiation results of the anomaly source, calculate multiple reliability sub-indicators corresponding to different physical fields, and fuse them to generate a comprehensive health index. Step S6: Based on the comprehensive health index and the reliability sub-indicator, perform degradation pattern identification and graded early warning.

[0029] The method of this invention is applicable to the status monitoring, reliability assessment and risk warning of current transformers in ultra-high voltage substations, and is especially applicable to the multi-physics field coupling degradation analysis and protection maloperation risk prediction of current transformers with voltage levels of 500kV and above under complex operating environments.

[0030] The steps of the method in this embodiment will be described in detail below.

[0031] This invention pre-constructs a multi-physics digital twin reference model of a current transformer. This model is coupled with electromagnetic, thermal, insulation, secondary circuit, and error transmission characteristics, and can synchronously reflect the interaction between the multi-physics fields of the current transformer under different operating conditions.

[0032] Specifically, as shown in Table 1, the construction of this digital twin model can be based on fundamental data such as the current transformer's design drawings, nameplate parameters, factory test reports, handover test data, and historical maintenance records. The model can be architecturally divided into a structural layer, a mechanistic layer, a data layer, and an evaluation layer. The structural layer describes the physical objects of the current transformer, including the primary conductor, core, secondary winding, main insulation, shielding structure, bushing, grounding terminal, and secondary terminals, and their spatial topological relationships. The mechanistic layer includes an electromagnetic field model, a thermal field model, an insulation state model, a secondary circuit model, and an error transmission model; these sub-models are coupled through shared state variables. The data layer stores real-time monitoring data, offline test data, and maintenance records. The evaluation layer outputs reliability sub-indicators and a comprehensive health index.

[0033] Table 1. Composition of Multiphysics Twin Model

[0034] This multiphysics digital twin model is pre-built and used during subsequent online evaluation and early warning processes. Model building can be completed during the system initialization phase and continuously updated based on subsequent maintenance feedback and operational data.

[0035] Step S1 will acquire real-time operating data of the current transformer and obtain the predicted response through a digital twin model.

[0036] Real-time operational data may include one or more of the following: primary-side electrical quantities, secondary-side electrical quantities, insulation monitoring quantities, temperature monitoring quantities, mechanical condition quantities, environmental parameters, and relevant operation and maintenance data. The acquired real-time operational data is input into a pre-built multiphysics digital twin model, and the model calculates the predicted response characterizing the theoretical state of the equipment under the current operating conditions.

[0037] As shown in Table 2 below, in one implementation, before inputting real-time operational data into the digital twin model, the data can be standardized, including unifying timestamps, units, equipment codes, and feature naming. For continuously monitored quantities, a sliding window can be used to extract statistical and trend features; for fault recording data, peak values, harmonic content, waveform distortion rate, transient slope, and energy features can be extracted; for inspection and maintenance records, they can be converted into structured tag information.

[0038] Table 2. Description of Multi-Source State Data and Standardized Characteristics

[0039] Step S3 will calculate the consistency deviation between the measured response and the predicted response.

[0040] Please combine Figure 2As shown, the measured response of the current transformer is first obtained. The measured response refers to the measured value or characteristic quantity that reflects the actual operating state of the current transformer, obtained through online monitoring devices or experimental methods, and corresponds to the predicted response output by the digital twin model in the physical dimension.

[0041] In this embodiment, the digital twin model is in The predicted vector output at time step can be represented as: (1) in, For digital twin models in The predicted vector output at time step; For input multi-source runtime data; These are model parameters; Operating conditions and boundary conditions; This represents the mapping relationship from input to predicted output in a digital twin model.

[0042] Accordingly, the measured response vector is obtained. The measured component and the predicted vector Each component corresponds one-to-one in physical meaning. For example, It can include measured ratio difference, phase difference, magnetic flux density, temperature, insulation characteristics, and secondary circuit characteristics.

[0043] Calculate the consistency deviation vector between the measured response and the predicted response. The calculation formula is as follows: (2) in, The consistency deviation vector corresponds to the ratio difference deviation, phase difference deviation, magnetic field deviation, temperature deviation, insulation deviation, and secondary circuit deviation, respectively.

[0044] in, The ratio difference deviation reflects the degree of deviation of the metering winding transmission error; Phase difference deviation reflects the degree of deviation of phase transmission error; The magnetic flux density deviation reflects the degree of deviation in the magnetic state of the iron core; Temperature deviation reflects the degree of deviation in the temperature rise of the winding or insulation; Deviations in insulation-related parameters may include, for example, deviations in leakage current or partial discharge. This refers to the deviation of secondary circuit-related parameters, such as secondary load deviation or grounding circuit impedance deviation. This consistency deviation vector... It quantifies the deviation between the actual state of the equipment and the theoretical expected state in each physical field dimension.

[0045] Step S4 locates the source of the anomaly and distinguishes the anomaly type based on the physical field associated with the consistency deviation.

[0046] Specifically, based on the calculated consistency deviation vector This allows us to identify the physical fields associated with the deviation in the multiphysics digital twin model, thereby pinpointing the source of the anomaly and distinguishing its type.

[0047] In one implementation, the location of the anomaly source can be achieved based on the calculation of anomaly scores for each component. Specifically, the location result of the anomaly component can be calculated using the following formula: (3) in, The abnormal component location result indicates the specific component or set of components that was determined to be abnormal. For the first An anomaly scoring function for a component type, whose output value characterizes the probability or severity of an anomaly in that component; The weight vectors corresponding to different deviation terms are used to reflect the importance or credibility of each deviation component in the anomaly judgment. This refers to the structural relationships within a current transformer, used to describe the physical, electrical, or spatial connections between its various components. This is achieved by introducing... When calculating the abnormal score of a component, the mutual influence between related components can be considered, avoiding judgment based on a single deviation item in isolation.

[0048] The abnormal scores of each component were calculated. Then, by taking the component index corresponding to the maximum value, the location result of the abnormal component at the current moment is determined. .

[0049] The localization method represented by formula (3) can be implemented in various specific ways. For example, as an exemplary implementation, it can be based on the consistency deviation vector. The physical meaning of each component is defined, and a mapping relationship is established between them and potentially abnormal components: if the consistency deviation is mainly reflected in and On an equal basis, the abnormal scores of metering transmission or core saturation related components are considered. The corresponding increase indicates that the source of the anomaly can be located in the aforementioned components; if the deviation is mainly manifested in In terms of weight, the abnormal scores of the iron core magnetic circuit or electromagnetic related components are considered. The corresponding increase indicates that the source of the anomaly can be located in this component; if the deviation is mainly reflected in In terms of components, for example, an increase in the deviation corresponding to the end-screen current, partial discharge quantity, or leakage current will result in an abnormal score for the main insulation-related components. The corresponding increase indicates that the source of the anomaly can be located in this component; if the deviation is mainly reflected in In terms of components, the abnormal scores are related to the thermal condition of the winding or insulation. The corresponding increase indicates that the source of the anomaly can be located in this component; if the deviation is mainly reflected in In terms of components, for example, an increase in the deviation of the secondary port voltage, secondary load impedance, or grounding loop impedance will result in an abnormal score for the relevant components of the secondary circuit or grounding loop. The corresponding increase indicates that the source of the anomaly can be located in the aforementioned components. Weights of each deviation component. Settings can be configured based on equipment type, historical fault statistics, or expert experience.

[0050] Based on the above location results, the anomaly types can be further distinguished. The distinguished anomaly types include at least one or more of the following: primary body degradation, secondary loop anomaly, sensor drift, and changes in operating conditions. Through the above distinction, it can be determined whether the same monitoring anomaly originates from equipment performance degradation, external loop interference, measurement system error, or normal fluctuations in operating conditions, thereby avoiding false alarms or missed alarms caused by a single threshold judgment.

[0051] Step S5 is used to calculate the reliability sub-indicators and integrate them to generate a comprehensive health index.

[0052] Please combine Figure 3 As shown, based on the consistency deviation and combined with the location and differentiation results of the anomaly source, multiple reliability sub-indices corresponding to different physical fields are calculated. Each reliability sub-indicator is used to characterize the health or reliability of the current transformer in the corresponding physical field dimension.

[0053] In one implementation, reliability sub-indicators may include metrological reliability indicators, electromagnetic reliability indicators, insulation reliability indicators, thermomechanical reliability indicators, and secondary circuit reliability indicators. Each sub-indicator may be expressed in a normalized manner, such as using a scoring system from 0 to 100 points, with higher values ​​indicating better reliability.

[0054] As an example implementation, the aforementioned reliability sub-indicators can be obtained through corresponding margin calculations.

[0055] For electromagnetic reliability, the core saturation margin can be calculated using the following formula: (4) in, This represents the core saturation margin. The saturation magnetic flux density of the iron core; This represents the absolute value of the maximum magnetic flux density of the iron core calculated by the digital twin model under the current operating conditions. The value range is usually [0,1]. The larger the value, the farther the iron core is from the saturation state and the higher the electromagnetic reliability.

[0056] For measurement reliability, the measurement reliability margin can be calculated using the following formula: (5) in, For measurement reliability margin; For ratio difference drift, For phase difference drift, both can be derived from the consistency deviation vector. corresponding components and Obtained through time series analysis; The total harmonic distortion rate of the secondary current is used to reflect the degree of waveform distortion. This is the allowable limit for the ratio difference. This refers to the allowable limit for phase difference. The above limits are permissible limits for harmonic distortion rate, and these limits can be determined according to relevant standards or equipment technical specifications. , , Let be the weighting coefficient, satisfying This is used to adjust the contribution ratio of each sub-item to the reliability of measurement. The higher the value, the closer the error transmission performance of the metering winding is to the ideal state.

[0057] For insulation reliability, the insulation reliability margin can be calculated using the following formula: (6) in, For insulation reliability margin; This refers to the measured or characteristic value of the leakage current or the end-screen current. These are partial discharge characteristic quantities, such as apparent discharge quantity or discharge repetition rate; This refers to the dielectric loss factor. , and These are the permissible limits for leakage current, partial discharge, and dielectric loss factor, respectively. This is an insulation degradation trend factor, used to characterize the degradation rate or acceleration trend of the above insulation parameters over time. , , , Let be the weighting coefficient, satisfying It is used to comprehensively consider the effects of different insulation parameters. The higher the value, the better the main insulation condition.

[0058] For thermomechanical reliability and secondary loop reliability, corresponding margin indices can be constructed similarly based on the consistency deviation components of the corresponding physical fields and design limits, which will not be elaborated here.

[0059] In obtaining each reliability sub-indicator Then, they can be integrated to generate a comprehensive health index for current transformers. In one implementation, The calculation formula is: (7) in, For the first Each reliability sub-indicator should include at least one measurement reliability margin. Electromagnetic reliability margin Insulation reliability margin , as well as thermomechanical reliability margin and secondary circuit reliability margin; The comprehensive health index of current transformers; For the first The weighting coefficients for each reliability sub-indicator are such that the sum of all weighting coefficients is 1. (Weighting coefficients) The value can be determined based on one or more factors, including equipment type, service life, severity of failure, and on-site maintenance strategy. For example, in applications where metering windings have strict accuracy requirements, the value can be appropriately increased. The weighting can be appropriately increased for equipment with a long service life and a high risk of insulation aging. The weight.

[0060] Furthermore, to facilitate comparison with tiered early warning thresholds, a comprehensive risk index can be defined. The calculation formula is as follows: (8) in, This is a comprehensive risk index used to compare with the tiered threshold and trigger an early warning; its value range is [0,1]. The closer it is to 0, the better the overall health of the equipment and the lower the operational risk. The closer the value is to 1, the worse the overall health of the equipment and the higher the operational risk. (This is achieved by...) By comparing the results with preset grading thresholds, an alert of the corresponding level can be triggered.

[0061] In this embodiment, the exemplary relationship between reliability sub-indicators and anomaly interpretations is shown in Table 3 below: Table 3 Relationship between reliability sub-indicators and anomaly interpretation

[0062] Step S6 is used to perform degradation pattern recognition and hierarchical early warning.

[0063] Specifically, degradation pattern identification and graded early warning are performed based on the comprehensive health index and reliability sub-indicators. Degradation pattern identification refers to determining the current degradation type or failure mode of the equipment based on the low scores of each reliability sub-indicator, the distribution characteristics of consistency deviations, and the historical case database. Graded early warning refers to determining the warning level based on the numerical range of the comprehensive health index and the lowest score of the reliability sub-indicator.

[0064] In one implementation, degradation pattern recognition can be achieved using probabilistic reasoning. Specifically, it can compute the set of anomalous features observed currently. Under these conditions, the equipment is in the first... Degeneration mode posterior probability The calculation formula is as follows: (9) in, For the first Various degradation modes, such as core saturation degradation, insulation aging, poor secondary circuit contact, and sensor zero-point drift; The set of observed anomalous features may include a consistency deviation vector. The abnormality markers of each component, the low-scoring items of the reliability sub-indicators, and the time series trends of the above features, etc. For degradation mode The prior probability can be determined based on the historical failure statistics of the equipment or the operation and maintenance experience of similar equipment. In degradation mode A set of anomalous features observed when it occurs Likelihood probability; denominator Observed under all possible degradation modes The marginal probability is used for normalization. The posterior probability of each degradation mode is calculated. This allows us to determine the most likely degradation mode, i.e., to take the mode corresponding to the maximum posterior probability. As a result of degradation pattern recognition.

[0065] After completing the degradation pattern identification, the warning level can be further determined. In one implementation, the warning level... It can be determined by considering multiple factors, and its calculation function can be expressed as: (10) in, It is at the warning level; For each reliability sub-indicator The minimum value in the equation reflects the state of the device in the weakest physical field dimension. The time change rate of the comprehensive health index is used to capture the rapid deterioration trend of equipment condition; As a consequence factor, it is used to characterize the importance of equipment, the degree of impact on relay protection, and the severity of power outage consequences. This factor can be preset according to factors such as the installation location of the current transformer in the power system, the type of protection associated with it, and the importance of the load.

[0066] function The specific form can be implemented using rule-based reasoning, fuzzy logic, or decision trees. For example, the following hierarchical rule can be set: when Above the first threshold and When it is above the second threshold, Determined to be at the normal level; when When the value is between the first and third thresholds, or when a sub-indicator shows a slight decrease, Determined to be at the attention level; when Below the third threshold, or Below the fourth threshold, or Exceeding the preset degradation rate threshold, or When the preset reliability threshold is exceeded, Determined as an alarm level; when Below the fifth threshold and If the index falls below the sixth threshold, or if multiple sub-indicators show a significant decline simultaneously, or if the consequences of the fault are deemed severe based on the Consequence factor, It has been classified as a serious level.

[0067] In one implementation, a mapping relationship of "risk level - abnormal component - degradation mode - remedial measures" can be established, outputting suggestions for on-site verification, online tracking, power outage testing, secondary circuit rectification, error verification, insulation retesting, and replacement. Specific examples are shown in Table 4 below: Table 4 Examples of Early Warning Classification and Response Rules

[0068] Through the above methods, a complete closed loop is achieved, from state awareness, reliability assessment, degradation identification to handling decisions.

[0069] Please combine again Figure 4 As shown, after obtaining the early warning results, handling feedback, maintenance test conclusions, or manual confirmation results, this embodiment of the invention writes the above information back to the digital twin model to update the model parameters, feature thresholds, and prior probabilities of degradation modes, forming a closed loop.

[0070] In one implementation, model parameter updates can be based on the deviation between the measured response and the predicted response. Specifically, the model parameters can be corrected using the following formula: (11) in, The model parameters at the current moment, For the updated model parameters, This is a correction coefficient matrix determined based on data reliability and maintenance confirmation results. Its values ​​are determined according to these criteria. When the maintenance confirmation conclusion is equipment degradation, the correction coefficients of the corresponding physical field model parameters can be increased to accelerate the model's tracking of the actual degradation state of the equipment. When the maintenance confirmation conclusion is sensor drift or abnormalities in secondary circuit accessories, the correction coefficients of the corresponding acquisition channels can be decreased to avoid introducing external interference into model parameter updates.

[0071] In one implementation, the reliability of the acquisition channel or feature can be updated using the following formula: (12) in, For the first The credibility of each acquisition channel or feature at the current moment; For the updated credibility; The confidence level corresponding to the inspection and confirmation results is assigned based on the conclusions of on-site verification or maintenance tests. For example, if it is confirmed that the channel has sensor drift, a lower confidence level is assigned; if it is confirmed that the channel is working properly, a higher confidence level is assigned. The forgetting coefficient, with a value range of (0,1), is used to control the proportion of historical reliability in the new round of updates. The larger the value, the smoother the confidence update and the lower the sensitivity to a single confirmation result; The smaller the value, the faster the reliability can respond to the latest maintenance confirmation conclusions.

[0072] In one implementation, for events identified as device degradation, the weight coefficient of the corresponding mechanism path in the digital twin model can be increased to improve the model's sensitivity to similar degradation patterns. For events identified as sensor drift or secondary circuit accessory anomalies, in addition to updating the credibility of the corresponding acquisition channel, process pruning can be triggered in subsequent evaluations. That is, for acquisition channels or feature items with credibility below a preset threshold, their participation in complex mechanism calculations can be suspended to reduce computational overhead and improve online diagnostic efficiency.

[0073] In this embodiment, the exemplary relationship between the early warning output and the twin closed-loop update is shown in Table 5 below: Table 5 Relationship between Early Warning Output and Twin Closed-Loop Update

[0074] Through the aforementioned closed-loop update mechanism, the digital twin model can continuously evolve as the equipment operates, gradually approximating the actual degradation patterns of the equipment and improving the accuracy of subsequent assessments and early warnings.

[0075] The following three typical examples illustrate the implementation effects of the present invention.

[0076] Example 1: An abnormal secondary load of a 500kV current transformer caused a decrease in metering reliability and core saturation margin.

[0077] A 500kV current transformer with a rated ratio of 4000 / 1A was monitored online during operation. The secondary load increased from approximately 5Ω to approximately 18Ω. Under short-time inrush current, the secondary current waveform exhibited significant distortion, with the ratio difference drifting from 0.08% to 0.27% and the phase difference drifting from 3 minutes to 10 minutes. No significant abnormalities were observed in synchronous partial discharge, end-screen current, or external insulation leakage current. This example illustrates how this invention identifies secondary load anomalies and electromagnetic saturation risks, rather than misdiagnosing a main insulation fault.

[0078] Table 6. Equipment and Model Parameter Definitions for Example 1

[0079] The system receives primary current, secondary current, secondary load, error estimation, fault recording, and insulation monitoring data, and generates feature tags such as {secondary load increase, secondary waveform distortion, ratio difference drift, phase difference drift, and insulation monitoring normal}.

[0080] The digital twin model calculated a maximum magnetic flux density of approximately 1.83T under the current primary current and secondary load conditions, and a core saturation margin of [missing value]. The value is approximately 0.062; the measured ratio difference and the twin prediction value deviate significantly, while the insulation deviation is close to zero. Therefore, the abnormal components are located in the secondary circuit and the electromagnetic transmission link.

[0081] System calculation and measurement reliability Approximately 0.66, electromagnetic reliability Approximately 0.062, insulation reliability Approximately 0.93, thermomechanical reliability Approximately 0.88, secondary circuit reliability Approximately 0.58, Overall Health Index Approximately 68.

[0082] According to the early warning rules, If the value is below 70 and both the secondary circuit reliability and metering reliability decrease simultaneously, the warning level is determined to be "alarm". Recommendations include: checking the system output secondary load, tightening terminal blocks, verifying the input impedance of the protection device, error checking, and reviewing short-term impact conditions.

[0083] After confirming on-site that the increased contact resistance of a certain secondary terminal has led to an increase in the equivalent load, the confirmation result is written into the digital twin model to increase the weight of the path "abnormal secondary load - error drift - decrease in saturation margin".

[0084] Table 7 Step-by-step calculation results for Example 1

[0085] Example 2: Identification of early degradation and temperature rise coupling risks of main insulation in ultra-high voltage current transformers.

[0086] During the high-load period in summer, a 500kV current transformer maintained a load factor above 0.92 for an extended period. The hot spot temperature increased by approximately 8K compared to historical operating conditions, the end-screen current increased by approximately 35% compared to the baseline value, and the partial discharge repetition rate increased from approximately 30pps to approximately 280pps. No synchronous anomalies were observed in the secondary load and metering errors. This example illustrates how the present invention identifies decreased insulation reliability and thermal field coupling degradation.

[0087] Table 8 Monitoring Quantities and Input Characteristics in Example 2

[0088] The system exhibits a set of characteristics: high load, high hotspot temperature, increased end-screen current, increased partial discharge repetition rate, and normal metering error.

[0089] In the consistency deviation vector and Significantly increased, and , and If the size is small, the system will locate the abnormal component in the main insulation and thermal field related areas.

[0090] Insulation reliability is calculated according to formula (6). Approximately 0.61, thermomechanical reliability Approximately 0.69, measurement reliability Approximately 0.90, secondary circuit reliability Approximately 0.92, Overall Health Index Approximately 72. Due to the continuous decline in insulator performance, the warning level has been determined to be "Attention-Alarm Critical".

[0091] The system outputs suggestions to increase the frequency of partial discharge monitoring, check the grounding and insulation monitoring circuit of the end screen, arrange infrared and partial discharge retesting, and conduct power outage dielectric loss and capacitance retesting when necessary.

[0092] If subsequent power outage tests confirm an increase in dielectric loss, the system will rewrite the maintenance conclusion and correct the insulation degradation trend factor. And insulation model threshold; if the sensing channel is confirmed to be abnormal, the confidence level of the acquisition channel is updated.

[0093] Table 9 Calculation and Output Results of Example 2

[0094] Example 3: Process pruning and false alarm suppression caused by sensor drift.

[0095] A 330kV current transformer online system showed an abnormal deviation of approximately 3% in the secondary current acquisition channel under low load. However, the offline verification of the synchronization ratio difference, secondary load, voltage recording, partial discharge, temperature, and end-screen current were all consistent with the digital twin prediction values. This example illustrates how the present invention identifies acquisition channel anomalies and avoids misjudging them as CT body failures.

[0096] The system extracts a feature set of {single secondary current channel deviation, other channels normal, no corresponding physical field anomaly}.

[0097] The consistency deviation vector contains only one acquisition channel. If the deviations in measurement, insulation, electromagnetic, and thermal fields are all small, the system will locate the anomaly as drift of the acquisition accessories or sensors.

[0098] Trigger process pruning: avoids calling branches for complex electromagnetic fields and insulation degradation calculations; only calculates the reliability of the acquisition channel. And the reliability of lightweight secondary circuits.

[0099] The system outputs suggestions for acquisition module calibration, wiring verification, time synchronization verification, and sensor replacement, but does not generate suggestions for power outage maintenance.

[0100] After the acquisition module was replaced on-site, the deviation was eliminated. The system then wrote the event back as a "sensor drift sample," reducing the credibility weight of the same acquisition channel in subsequent diagnoses.

[0101] Table 10. Process trimming results of Example 3

[0102] As can be seen from the three examples above, this invention uses a multiphysics twin model as a unified benchmark, interprets the source of anomalies through the deviation between measured and twin consistency, and forms a closed loop for reliability assessment, early warning classification, handling suggestions, and model updates. Example 1 fully invokes the electromagnetic and secondary circuit branches, Example 2 invokes the insulation and thermal field coupling branches, and Example 3 triggers process trimming and identifies abnormalities in the acquisition channel, demonstrating the interpretability, quantifiability, and customizability of this invention.

[0103] Corresponding to the method for reliability assessment and early warning of current transformers based on digital twins in Embodiment 1 of the present invention, Embodiment 2 of the present invention also provides a device for reliability assessment and early warning of current transformers based on digital twins, comprising: One or more processors; Memory; One or more applications, wherein the one or more applications are stored in the memory and configured to be executed by the one or more processors, and the one or more applications are configured to perform the digital twin-based current transformer reliability assessment and early warning method.

[0104] Corresponding to the method for reliability assessment and early warning of current transformers based on digital twins in Embodiment 1 of the present invention, Embodiment 3 of the present invention also provides a computer program product, including computer instructions, which instruct computer equipment to perform the operation corresponding to the method for reliability assessment and early warning of current transformers based on digital twins.

[0105] Preferably, the processor can be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor, or the processor can be any conventional processor. The processor is the control center of the device, connecting various parts of the device through various interfaces and lines.

[0106] The memory mainly includes a program storage area and a data storage area. The program storage area can store the operating system, applications required for at least one function, etc., while the data storage area can store related data, etc. Furthermore, the memory can be a high-speed random access memory, or a non-volatile memory, such as a plug-in hard drive, a SmartMedia Card (SMC), a Secure Digital (SD) card, and a Flash Card, or other volatile solid-state storage devices.

[0107] It should be noted that the above-mentioned devices may include, but are not limited to, processors and memory, as will be understood by those skilled in the art.

[0108] For the working principle and process of the above embodiments, please refer to the description of Embodiment 1 of the present invention, which will not be repeated here.

[0109] Compared with existing technologies, this invention has the following significant advantages: By constructing a multi-physics digital twin model that couples electromagnetic, thermal, insulation, secondary circuit, and error transmission characteristics, this invention unifies scattered online monitoring data, fault waveform data, offline test data, and operation and maintenance records into reliability sub-indicators and comprehensive health indices for each physical field. This enables early identification and graded warning of potential risks such as metering error drift, core saturation margin decline, main insulation degradation, and secondary circuit anomalies in ultra-high voltage current transformers. It effectively solves the problems of high false alarm rate, low missed alarm rate, and information fragmentation between different disciplines in traditional fixed threshold alarm methods under the coupling effect of multi-physics fields. This invention can distinguish the source of anomalies as primary body degradation, secondary circuit anomalies, sensor drift, or changes in operating conditions, thus providing targeted and actionable on-site handling suggestions, including secondary load verification, error calibration, partial discharge retesting, grounding continuity testing, terminal block inspection, and power outage maintenance suggestions, reducing blind and excessive maintenance. Meanwhile, by feeding back early warning results, maintenance conclusions, and manual confirmation information in a closed loop to the digital twin model, the model parameters, feature thresholds, and prior probabilities of degradation modes are continuously updated, so that the assessment accuracy gradually improves as the equipment operates. This forms a knowledge accumulation of the reliability of similar current transformer equipment throughout its entire life cycle, providing support for the accuracy and traceability of the status evaluation of main equipment in ultra-high voltage substations, and also laying a technical foundation for the research, development, operation, and maintenance of environmentally friendly digital twin current transformers.

[0110] The above description is merely a preferred embodiment of the present invention and should not be construed as limiting the scope of the invention. Therefore, any equivalent variations made in accordance with the claims of the present invention are still within the scope of the present invention.

Claims

1. A method for reliability assessment and early warning of current transformers based on digital twins, characterized in that, include: Step S1: Obtain the real-time operating data of the current transformer; Step S2: Input the real-time operating data into a pre-constructed multiphysics digital twin model to obtain a predicted response characterizing the theoretical state of the equipment; wherein, the multiphysics digital twin model is coupled with electromagnetic, thermal, insulation, secondary loop and error transmission characteristics; Step S3: Obtain the measured response of the current transformer and calculate the consistency deviation between the measured response and the predicted response; Step S4: Based on the consistency deviation, locate the source of the anomaly and distinguish the anomaly type in the physical field associated with the multiphysics digital twin model; Step S5: Based on the consistency deviation and combined with the location and differentiation results of the anomaly source, calculate multiple reliability sub-indicators corresponding to different physical fields, and fuse them to generate a comprehensive health index. Step S6: Based on the comprehensive health index and the reliability sub-indicator, perform degradation pattern identification and graded early warning.

2. The method according to claim 1, characterized in that, Step S4 specifically includes: Calculate the anomaly score for each of the multiple preset components, and determine the component with the highest anomaly score as the anomaly component location result; The anomaly score is determined based on the consistency deviation vector, the weight vectors corresponding to different deviation items, and the device structure correlation.

3. The method according to claim 1, characterized in that, In step S4, the distinction of anomaly types includes: the distinction of anomalies as at least one of primary body degradation, secondary loop anomaly, sensor drift, and change in operating conditions.

4. The method according to claim 1, characterized in that, In step S5, the plurality of reliability sub-indicators corresponding to different physical fields include: metrological reliability index, electromagnetic reliability index, insulation reliability index, thermomechanical reliability index, and secondary circuit reliability index.

5. The method according to claim 4, characterized in that, The electromagnetic reliability index is obtained through the core saturation margin. The characterization is calculated using the following formula: in, The saturation magnetic flux density of the iron core; This is the absolute value of the maximum magnetic flux density of the iron core output by the multiphysics digital twin model.

6. The method according to claim 4, characterized in that, The metrological reliability index is expressed through the metrological reliability margin. The characterization is calculated using the following formula: in, For ratio difference drift, For phase difference drift, The total harmonic distortion rate of the secondary current. This is the allowable limit for the ratio difference. This refers to the allowable limit for phase difference. This refers to the permissible limit for harmonic distortion rate. , , These are the weighting coefficients, and .

7. The method according to claim 4, characterized in that, The insulation reliability index is obtained through the insulation reliability margin. The characterization is calculated using the following formula: in, This refers to leakage current or end-screen current. For partial discharge characteristics; This refers to the dielectric loss factor. , and These are the permissible limits for leakage current, partial discharge, and dielectric loss factor, respectively. Insulation degradation trend factor; , , , These are the weighting coefficients, and .

8. The method according to any one of claims 1-7, characterized in that, Also includes: Step S7: Feed back the results of the graded early warning, the conclusions of the maintenance test, and the information of manual confirmation to the multiphysics digital twin model to update the model parameters and the reliability of the acquisition channel; The model parameters are updated according to the following formula: in, The model parameters at the current moment, For the updated model parameters, This is a correction coefficient matrix determined based on data reliability and maintenance confirmation results; This is the measured response vector; To predict the response vector; The update of the acquisition channel reliability is performed according to the following formula: in, For the first The credibility of each acquisition channel or feature at the current moment; For the updated credibility; The confidence level corresponding to the inspection and confirmation results; λ is the forgetting coefficient, 0 < λ < 1.

9. A current transformer reliability assessment and early warning device based on digital twin, characterized in that, include: One or more processors; Memory; One or more applications, wherein the one or more applications are stored in the memory and configured to be executed by the one or more processors, the one or more applications being configured to perform the current transformer reliability assessment and early warning method based on digital twins as described in any one of claims 1 to 8.

10. A computer program product, characterized in that, It includes computer instructions that instruct computer equipment to perform operations corresponding to the current transformer reliability assessment and early warning method based on digital twins as described in any one of claims 1 to 8.

Images