Fault-recording-based power distribution area electric leakage fault diagnosis method, device and storage medium

Abstract

The application discloses a transformer area electric leakage fault diagnosis method and device based on fault recording and a storage medium, and relates to the technical field of electric leakage fault diagnosis. The method comprises the following steps: based on event recording data, calculating the fundamental wave phase angle difference between the residual current and each phase current in a first analysis window, and calculating the phase angle consistency score based on the fundamental wave phase angle difference; determining whether the corresponding phase is an electric leakage phase according to the phase angle consistency score; generating a statistical quantity sequence of each electrical quantity aggregated according to a fixed time period in a second analysis window based on the instantaneous value of each electrical quantity; calculating the mean value transition amount of the residual current, the synchronization parameter of the residual current and the total amount of three-phase current, and the duration of the residual current exceeding the threshold value according to the statistical quantity sequence; and judging the electric leakage type according to the mean value transition amount, the synchronization parameter and the duration. The application overcomes the weakness of the traditional amplitude correlation method which is easily affected by sampling deviation, and the phase positioning is more stable and accurate, and the wiring error and insulation damage can be effectively distinguished.

Description

Technical Field

[0001] This invention belongs to the field of leakage fault diagnosis technology, and particularly relates to a method, device and storage medium for diagnosing leakage faults in transformer substations based on fault recording. Background Technology

[0002] Leakage faults in distribution transformer areas are a significant factor affecting power supply safety and stable operation. Failure to diagnose and address them promptly can lead to serious consequences such as equipment damage, electric shock, and unplanned power outages. Currently, detection and diagnosis methods for low-voltage transformer area leakage faults largely rely on on-site maintenance experience or simple over-limit alarms from traditional protection devices.

[0003] Traditional methods typically rely on the operation records of residual current protection devices or the over-limit threshold of the effective value of the residual current to determine whether leakage has occurred, and then attempt to distinguish the fault type based on this. However, these methods have significant shortcomings in practical applications:

[0004] First, relying solely on whether the residual current amplitude exceeds a fixed threshold is easily affected by factors such as load fluctuations and transient processes, resulting in frequent false alarms and missed alarms.

[0005] Secondly, some methods attempt to determine the leakage phase by analyzing the correlation between the residual current and the three-phase current. However, due to interference from factors such as the polarity configuration of the on-site sampling channel, the difference in phase sequence wiring, and waveform distortion, the accuracy and stability of the determination are often difficult to guarantee.

[0006] Third, most existing technologies rely on circuit breaker tripping as the starting point for analysis, or only perform local analysis on short-time waveform data of a single trigger. They lack a unified definition of the event starting point and a sufficiently long analysis window, making it difficult to accurately capture the complete transient process and steady-state characteristics of the fault. This results in limited ability to distinguish fault types (such as user grounding wiring errors and equipment insulation damage).

[0007] Fourth, considering the storage capacity and communication bandwidth limitations of the field terminal equipment, requiring long-term storage of high-sampling-rate raw waveform data for subsequent offline analysis would result in high hardware costs and maintenance burdens.

[0008] To overcome these shortcomings, some research has attempted to use phasor information or waveform characteristics at the power frequency level for leakage current phase identification and type determination. However, most of these improved methods still cannot escape dependence on trip signals or single triggering conditions, placing high demands on the consistency of data sampling, the synchronization of channels, and the robustness of algorithms under complex operating conditions. Furthermore, they often struggle to effectively integrate the detailed characteristics of transient fault states with the macroscopic behavior of continuous processes, thus limiting their feasibility and universality in practical engineering applications.

[0009] Therefore, there is an urgent need for an intelligent method for diagnosing leakage faults that is adapted to the online operating conditions of terminal equipment in distribution transformer areas. This method should be able to fully utilize the existing data acquisition and recording functions of existing protection or metering devices without requiring the addition of new sensors or modifications to on-site wiring. Through innovative data analysis and fusion strategies, it should achieve rapid and stable determination of leakage phases and accurate differentiation of fault types, thereby improving the overall efficiency, accuracy, and engineering practical value of leakage fault diagnosis in distribution transformer areas. Summary of the Invention

[0010] In view of the above-mentioned defects in the existing technology, the purpose of the present invention is to provide a method, device and storage medium for diagnosing transformer leakage faults based on fault recording, so as to solve at least one of the problems in the existing transformer leakage fault diagnosis methods, such as over-reliance on trip signals, false alarms and missed alarms due to single threshold alarms, large influence of sampling deviation on phase identification, insufficient ability to distinguish fault types, and cost and resource pressure caused by long-term storage of high-frequency data.

[0011] This invention solves the above-mentioned technical problems through the following technical solution: a method for diagnosing leakage current faults in transformer substations based on fault recording, comprising:

[0012] Under normal operation, the instantaneous values ​​of various electrical quantities on the low-voltage side of the transformer area are continuously collected at sampling intervals; wherein, the electrical quantities include at least three-phase current and residual current;

[0013] When the preset abnormal event triggering conditions are met, the moment when the conditions are first met is determined as the analysis baseline moment; acquire and save a segment of event waveform data including the analysis baseline moment, wherein the recording start point of the waveform data is earlier than the analysis baseline moment;

[0014] Based on the event recording data, within the first analysis window containing the analysis reference time, the fundamental phase angle difference between the residual current and the current of each phase is calculated, and a phase angle consistency score is calculated based on the fundamental phase angle difference; the corresponding phase is determined to be a leakage phase based on the phase angle consistency score.

[0015] Using the aforementioned analysis reference time as the boundary, a pre-trigger sub-window and a post-trigger sub-window are defined, which together constitute the second analysis window. Based on the instantaneous values ​​of each electrical quantity, a statistical sequence of each electrical quantity within the second analysis window is generated, aggregated according to a fixed time period. Based on the statistical sequence, the mean transition value of the residual current, the synchronization parameter between the residual current and the total three-phase current, and the duration of the residual current exceeding the threshold are calculated. The leakage current type is determined based on the mean transition value, the synchronization parameter, and the duration.

[0016] This invention employs a method of calculating the fundamental phase angle difference and constructing a phase angle consistency score for phase identification. Phase angle information is insensitive to amplitude ratio mismatch and certain waveform distortions, primarily reflecting the phase relationship of the current. This fundamentally overcomes the fatal weaknesses of traditional amplitude correlation methods, which are affected by sampling channel gain errors and reverse polarity, greatly enhancing the stability and robustness of phase identification. This invention uses the moment when soft conditions are met as the analysis reference time, eliminating the need to wait for the protection device to trip. This allows diagnostic actions to be performed online in the early stages of a fault, providing maintenance personnel with early warning and response time, transforming "post-event tracing" into "online analysis."

[0017] This invention establishes a first analysis window (based on event recordings, used for precise transient analysis) and a second analysis window (based on statistical sequences, used for steady-state trend analysis). This design simultaneously captures the instantaneous transient characteristics and long-term steady-state behavior of faults, providing a comprehensive data foundation for accurately distinguishing between transient wiring errors and persistent insulation damage, and solving the problem of "difficulty in classification due to excessively short windows".

[0018] This invention employs a hybrid data management model combining "normal lightweight statistics" and "event-triggered detailed recording." Under normal conditions, only a single circular buffer needs to be maintained to generate / save lightweight periodic statistics, resulting in a minimal data volume. When trigger conditions are met, the pre-triggered data in the buffer is immediately locked and transferred, while subsequent data continues to be recorded, forming a complete high-frequency waveform file containing the critical fault process. This model, while ensuring the complete preservation of fault details, completely avoids the enormous storage and communication overhead caused by long-term indiscriminate storage of all high-frequency waveform data, achieving optimal allocation of data storage resources.

[0019] Furthermore, the electrical quantity also includes zero-ground voltage, and the preset abnormal event triggering condition includes at least one of the following:

[0020] The instantaneous value of the residual current or the value in the statistical sequence exceeds a set first threshold.

[0021] The rate of change of the residual current exceeds a preset second threshold.

[0022] The difference between the mean values ​​of residual current statistics in adjacent time units exceeds a preset third threshold.

[0023] The sudden change in zero-ground voltage at adjacent sampling times exceeds the preset fourth threshold.

[0024] This invention integrates four criteria: amplitude, rate of change (transient), statistical mutation (steady-state trend), and zero-ground voltage (correlation feature). It can cross-verify the authenticity of the fault from different physical dimensions, ensuring that waveform recording and analysis are triggered only when a real and suspected leakage event occurs, which greatly improves the reliability of the system.

[0025] Amplitude exceeding limits: directly captures steady-state leakage current; Rate of change exceeding limits: sensitively responds to sudden changes in step-type faults (such as wiring errors); Statistical mean abrupt change: effectively identifies the trend changes in slowly changing or intermittent faults (such as slow insulation degradation); Zero-to-ground voltage abrupt change: serves as an important supplement and confirmation to current criteria, especially suitable for the early detection of certain special grounding faults. This combination enables the invention to adapt to various leakage fault modes that may occur in transformer substations, providing broad diagnostic coverage.

[0026] Further, the calculation of the fundamental phase angle difference between the residual current and each phase current, and the calculation of the phase angle consistency score based on the fundamental phase angle difference, includes:

[0027] For each power frequency cycle within the first analysis window, calculate the residual current and the fundamental complex phasor of each phase current;

[0028] Calculate the corresponding fundamental complex phase angles based on the residual current and the fundamental complex phasors of each phase current;

[0029] Calculate the fundamental phase angle difference between the residual current and the current in each phase, and perform a back-flipping process;

[0030] Based on the results of the turnaround process, the phase angle consistency score is calculated.

[0031] The core criterion of this invention is shifted from the traditional amplitude correlation to the fundamental phase angle difference. The phase angle information is not sensitive to the gain error of the sampling channel, small amplitude fluctuations, and certain waveform distortions. It mainly reflects the phase relationship of the current, thereby completely overcoming the problem of misjudgment caused by inconsistent characteristics of field transformers and wiring polarity deviations, making the judgment result inherently stable.

[0032] By using the phase angle reversal operator to constrain the phase angle difference within the range of [-π,π) or [0,2π), the computational ambiguity caused by the full-cycle phase rotation is eliminated, ensuring that the phase angle difference reflects the true phase lead or lag relationship.

[0033] Phase angle consistency scoring is not a simple averaging of phase angle differences, but rather a percentage of the effective value of the phase current. The weights are used for weighted averaging. This gives the phase with heavier load and larger current a more important place in the score, which is consistent with the physical fact that "leakage current is more likely to be generated from the phase with a larger load", further enhancing the physical rationality and accuracy of the judgment results.

[0034] Further, determining whether a corresponding phase is a leakage phase based on the phase angle consistency score includes:

[0035] The maximum and second-largest values ​​are determined from the phase angle consistency scores of phases A, B, and C;

[0036] If the maximum value is greater than or equal to the first scoring threshold, and the difference between the maximum value and the second largest value is greater than or equal to the second scoring threshold, then the phase corresponding to the maximum value is determined to be the leakage phase.

[0037] This invention not only requires that the score of the suspected leakage phase be sufficiently high (maximum value ≥ first scoring threshold), but also that it be significantly better than other phases (maximum value - second largest value ≥ second scoring threshold). This "relative advantage" criterion ensures that only when the phase angle consistency of a phase is significantly outstanding is it identified as a leakage phase. This effectively prevents ambiguity caused by factors such as three-phase load imbalance and measurement noise, which can lead to similar scores for two phases and make it difficult to choose, thus avoiding misjudgments based on apparent similarity.

[0038] The first scoring threshold (absolute threshold) ensures that candidate phases have a basic level of phase angle consistency, excluding cases where scores are accidentally inflated due to random interference. The second scoring threshold (relative threshold) ensures that the leading advantage of candidate phases is sufficiently significant, fundamentally eliminating the possibility of making arbitrary selections when multiple phase scores are at a critical state. The combination of these two thresholds ensures that the final "leakage phase" determination has high confidence and strong resistance to interference.

[0039] This invention can intelligently identify edge scenarios where the score is high but the advantage is not obvious. When the dual scoring threshold is not met, the algorithm may not output a definite phase distinction, but instead give a conclusion of "phase distinction uncertain". This conservative strategy of "better to remain doubtful than to make a wrong judgment" is crucial for applications with high reliability requirements such as fault diagnosis, avoiding erroneous positioning based on weak evidence that could lead to misoperation by maintenance personnel.

[0040] Further, based on the instantaneous values ​​of each electrical quantity, a statistical sequence of each electrical quantity within the second analysis window, aggregated according to a fixed time period, is generated, including:

[0041] Within the second analysis window, multiple consecutive and non-overlapping time windows are divided using the fixed time period as the unit; for each time window, the effective value is calculated based on all instantaneous values ​​of the electrical quantity within it, which serves as the statistical quantity of that electrical quantity in that time window; the statistical quantities of each time window are arranged in chronological order to form the statistical quantity sequence.

[0042] The formula for calculating the corresponding statistic for a given time window is as follows:

[0043] ;

[0044] in, This represents the statistical value of electrical quantity x within the m-th time window; N represents the p-th instantaneous value of electrical quantity x within the m-th time window; N = T s / △t, N represents the number of sampling points in the m-th window, T s The value represents the size of the time window, and Δt represents the sampling interval.

[0045] This invention calculates effective values ​​using a fixed time period as a window, compressing and refining massive amounts of instantaneous data into a series of representative values ​​characterizing the energy level within that time period. This process significantly reduces the amount of data to be processed, lowering the computational burden. Furthermore, the effective values ​​themselves filter out high-frequency noise and waveform distortion, highlighting the main trends in current change. This makes subsequent macroscopic features such as "mean transitions" and "synchronicity" clearer, more stable, and easier to extract and judge.

[0046] This invention establishes a unified, equally spaced time scale for the data throughout the second analysis window. This makes the data "before triggering" and "after triggering" have a completely comparable time basis; the calculation of the "mean transition" (comparison of the mean values ​​of the statistics before and after the window) becomes meaningful and computationally fair; and the measurement of "duration" can be accurate to integer time period units. This normalization is a prerequisite for accurate and quantitative time series feature analysis.

[0047] Further, the calculation of the mean transition of the residual current includes:

[0048] Calculate the first average value of the residual current statistics sequence in the sub-window before triggering and the second average value of the residual current statistics sequence in the sub-window after triggering;

[0049] Calculate the absolute difference between the first average value and the second average value; the absolute difference is the mean transition amount.

[0050] This invention avoids direct analysis of complex transient waveforms, instead focusing on the average level of residual current in two relatively steady-state phases before and after a fault event. By calculating the absolute difference between the average values ​​of these two phases, the overall rise (or fall) of the DC or fundamental component of the residual current introduced by the fault event is directly and quantitatively measured. The mean transition is a key indicator for determining whether a fault causes structural or persistent changes.

[0051] For incorrect user ground / neutral wiring: the fault usually occurs at the moment of equipment switching, manifested as a sudden jump in residual current from one steady-state level to another significantly different steady-state level, resulting in a large difference in the average value before and after (large mean jump). For insulation damage: leakage may develop gradually, or although it occurs suddenly, the current value is relatively small and fluctuates, and the change in the average value before and after may not be obvious (small mean jump). Therefore, the mean jump is a very effective first and most intuitive filter for distinguishing between these two types of faults.

[0052] This invention achieves the ability to smooth noise and suppress instantaneous spike interference by calculating the average value of the sequence within a sub-window and then comparing them. Abnormal fluctuations at a single sampling point will not have a decisive impact on the average value, thus ensuring that the characteristic value of "mean jump" can stably reflect the overall trend change caused by the fault, rather than random disturbances, thereby improving the robustness of the criterion.

[0053] Furthermore, the synchronization parameters of the residual current and the total three-phase current are calculated as follows:

[0054] Calculate the correlation coefficient between the residual current statistics sequence and the total three-phase current statistics sequence within the sub-window after the trigger.

[0055] This invention uses the correlation coefficient as a synchronicity parameter. The correlation coefficient directly and quantitatively measures the degree of linear correlation between the changing trends of two statistical sequences within the sub-window after triggering. It can reveal whether the change in residual current is "synchronous and in the same direction" with the change in the three-phase total load current, thereby transforming the abstract concept of "synchronicity" into a clear numerical criterion between -1 and 1, providing accurate and comparable input for automated discrimination.

[0056] For incorrect grounding connections by the user: the residual current essentially originates from the incorrect shunting of phase current (such as flowing to the ground wire). Therefore, its magnitude directly and strongly depends on the load current of the corresponding phase. When the total load current fluctuates, this "incorrectly shunted" current will also fluctuate synchronously, resulting in a high positive correlation between the residual current and the total three-phase current sequence (correlation coefficient close to 1).

[0057] For insulation damage: leakage current mainly depends on the degree of insulation resistance degradation and the voltage to ground at the fault point, and has a weak correlation with changes in load current, manifesting more as a relatively independent or continuous leakage. Therefore, its correlation coefficient with the total three-phase current is usually low or not significantly correlated.

[0058] By using correlation coefficients, we can directly distinguish between "current-following" faults (wiring errors) and "voltage-driven / fixed leakage" faults (insulation damage) based on data correlation.

[0059] Furthermore, the duration for which the residual current exceeds the threshold refers to the number of residual current statistics in the residual current statistics sequence that exceed the set first threshold within the sub-window after triggering.

[0060] This invention transforms the qualitative concept of "long / short duration" into a precisely calculable and unambiguous numerical value—the number of statistical quantities that meet the conditions. This avoids subjective judgment and makes "duration" a hard indicator that can be directly and fairly compared with preset thresholds (the seventh and eighth thresholds), providing crucial numerical evidence for automated judgment.

[0061] Furthermore, the leakage current type is determined based on the mean transition amount, synchronization parameters, and duration, including:

[0062] If the mean transition amount is greater than or equal to the preset fifth threshold, the synchronization parameter is greater than or equal to the preset sixth threshold, and the duration is less than or equal to the seventh threshold, then the leakage current type is determined to be a user ground connection error.

[0063] If the mean transition amount is less than the preset fifth threshold, the synchronization parameter is less than the preset sixth threshold, and the duration is greater than or equal to the eighth threshold, then the leakage type is determined to be insulation failure.

[0064] The eighth threshold is greater than the seventh threshold.

[0065] This invention abandons the one-sidedness of single-feature decision-making and creatively combines three features that characterize the nature of a fault from different dimensions—mean transition magnitude (step amplitude), synchronicity parameter (correlation with load), and duration (event length)—into a joint logical judgment. A judgment is only made when all three features simultaneously conform to the typical phenomena of a certain type of fault. This multi-dimensional cross-validation mechanism minimizes misclassification caused by accidental matching of individual features, ensuring high accuracy in type determination.

[0066] For user grounding wiring errors: its typical characteristics are defined as "large amplitude (≥ fifth threshold), strong correlation (≥ sixth threshold), and short duration (≤ seventh threshold)". This perfectly corresponds to the physical process of "a large amount of phase current mistakenly flowing into the ground wire due to wiring errors at the moment of equipment switching".

[0067] For insulation damage: its typical characteristics are defined as "small amplitude, weak correlation, and long duration (≥ the eighth threshold, and the eighth threshold > the seventh threshold)". This accurately describes the fault mode of "continuous deterioration of equipment insulation leading to long-term and relatively stable leakage current to ground".

[0068] By setting the eighth threshold to be greater than the seventh threshold, a clear and non-overlapping boundary is established for the two types of faults in the time dimension, avoiding the situation where they cannot be distinguished due to the duration being in the intermediate ambiguity zone.

[0069] This invention not only provides a clear binary decision (user ground wiring error or insulation failure), but its criteria themselves also possess strong physical interpretability. Maintenance personnel can intuitively understand that an event judged as a wiring error must simultaneously satisfy three verifiable characteristics: "significant sudden increase in current," "change synchronized with load," and "rapid disappearance." This enhances the credibility and acceptability of the diagnostic results and provides clear clues for subsequent troubleshooting.

[0070] Based on the same concept, the present invention also provides an electronic device, including a memory, a processor, and a computer program or instructions stored in the memory, wherein the processor executes the computer program or instructions to implement the fault recording-based transformer leakage fault diagnosis method as described above.

[0071] Based on the same concept, the present invention also provides a computer-readable storage medium having a computer program or instructions stored thereon, which, when executed by a processor, implements the above-described method for diagnosing transformer leakage faults based on fault recording waveforms.

[0072] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0073] This invention employs event-triggered waveform data for precise phase angle analysis, and determines the leakage phase through phase angle consistency scoring. This overcomes the weakness of traditional amplitude correlation methods, which are easily affected by sampling deviations, resulting in more stable and accurate phase identification. Simultaneously, steady-state characteristic analysis (mean transition, synchronicity, and duration) is performed using a fixed-time-period statistical sequence, effectively distinguishing between wiring errors and insulation damage, making type identification more reliable.

[0074] Normally, only lightweight statistical sequences need to be processed / stored, and short-term high-frequency waveforms are only saved when triggered. This intelligent hybrid data management mode greatly saves the storage and communication resources of terminal devices, allowing advanced diagnostic functions to be deployed directly on existing transformer terminals without hardware modifications, making it highly economical and feasible in terms of engineering.

[0075] This invention automates the entire process, from data acquisition and event triggering to phase and type determination, ultimately arriving at a clear conclusion. The logic is clear and straightforward. This changes the inefficient troubleshooting model that relies on manual experience. The output ("X phase, wiring error / insulation damage") directly guides precise maintenance, significantly improving operation and maintenance efficiency and power supply reliability. Attached Figure Description

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

[0077] Figure 1 This is a flowchart of the transformer substation leakage fault diagnosis method based on fault recording in an embodiment of the present invention;

[0078] Figure 2 This is a comparison chart of phase angle consistency scores for leakage scenario one in this embodiment of the invention;

[0079] Figure 3 This is a comparison chart of phase angle consistency scores for leakage current scenario two in this embodiment of the invention;

[0080] Figure 4 This is a comparison chart of phase angle consistency scores for leakage scenario three in this embodiment of the invention;

[0081] Figure 5 This is a comparison chart of phase angle consistency scores for leakage scenario four in this embodiment of the invention;

[0082] Figure 6 This is a comparison chart of phase angle consistency scores for leakage scenario five in this embodiment of the invention. Detailed Implementation

[0083] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0084] The technical solution of the present invention will be described in detail below with reference to specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0085] Example 1

[0086] like Figure 1 As shown, the method for diagnosing transformer leakage faults based on fault recording provided in this embodiment of the invention includes the following steps:

[0087] Step S1: Under normal operation, continuously collect the instantaneous values ​​of various electrical quantities on the low-voltage side of the transformer area at sampling intervals.

[0088] Under normal operating conditions (i.e., the low-voltage distribution lines in the transformer area are in normal power supply mode and there are no leakage faults or abnormal residual currents), this invention continuously collects the instantaneous values ​​of various electrical quantities on the low-voltage side of the transformer area at fixed sampling intervals. The specific implementation process is as follows:

[0089] Data collected includes at least the A-phase current, B-phase current, C-phase current, residual current, and neutral-to-ground voltage (optional) on the low-voltage side of the distribution area. These electrical quantities are obtained through existing current transformers, residual current transformers, and voltage transformers in the distribution terminal or monitoring device.

[0090] Sampling interval: To ensure the accuracy of subsequent analysis, the sampling interval should be significantly smaller than the power grid frequency cycle (e.g., if the power grid frequency is 50Hz, the power grid frequency cycle is 20ms). In a preferred embodiment, the sampling interval is 10ms, 5ms, or less to ensure that enough sampling points can be obtained within each power grid frequency cycle, thereby accurately reconstructing the waveform.

[0091] Sampling mode: The sampling process is continuous, equally spaced real-time sampling, which does not depend on any triggering conditions and continues throughout the entire normal operation time.

[0092] Circular buffer (or rolling buffer): The instantaneous values ​​of each electrical quantity acquired are written into a fixed-length first-in-first-out (FIFO) circular buffer. The length of this circular buffer (i.e., the length of time it can cover) should be greater than the preset pre-trigger time (e.g., 5~10 seconds) to ensure that when an abnormal event is triggered, the original data for a period of time before the event occurs can be saved.

[0093] Persistent storage strategy: To conserve storage resources, all original high-frequency instantaneous values ​​are not stored permanently under normal circumstances. Instantaneous values ​​are mainly used for the following two purposes:

[0094] Real-time calculation of statistics: used to generate the statistical series (such as second-level valid values) required for subsequent steps in real time.

[0095] Provides event analysis data source: When an abnormal event is triggered, the instantaneous value in the current loop buffer will be locked and transferred as the pre-trigger data part of the "event recording" for high-precision transient analysis.

[0096] Through step S1, the present invention only needs to maintain a circular cache and perform lightweight computation under normal conditions, realizing efficient and low-resource management of raw data, and providing a reliable and continuous data foundation for the entire online diagnostic method.

[0097] Step S2: Determine whether the preset abnormal event triggering conditions are met, and when the preset abnormal event triggering conditions are met, determine the moment when the conditions are first met as the analysis reference time; acquire and save a segment of event waveform data including the analysis reference time.

[0098] The system continuously monitors the electrical quantity data (especially residual current) collected in step S1, and calculates in real time the criteria related to the preset abnormal event triggering conditions, such as:

[0099] Does the instantaneous value or the value in the statistical sequence of the residual current exceed the set first threshold? (e.g., 0.4) , The rated residual operating current of the leakage current protection device for the transformer substation is typically 300mA.

[0100] Does the rate of change (differential or slope) of the residual current exceed a preset second threshold? (e.g., 0.1) );

[0101] Does the difference between the mean values ​​of residual current statistics in adjacent time units (e.g., seconds) exceed a preset third threshold? (e.g. 0.2) );

[0102] Does the abrupt change (i.e., absolute difference) of the zero-ground voltage at adjacent sampling times exceed a preset fourth threshold (e.g., 15V)?

[0103] When any of the above criteria first changes from "not satisfied" to "satisfied," the system immediately records the precise moment of this transition as t0 and determines it as the analysis baseline moment for this abnormal event analysis. The determination of the analysis baseline moment does not depend on protection tripping, thus achieving proactive diagnosis.

[0104] The difference between the mean values ​​of residual current statistics in adjacent time units refers to the absolute difference between the arithmetic mean of all instantaneous residual current values ​​in the current time unit (e.g., 1 second) and the arithmetic mean of all instantaneous residual current values ​​in the previous time unit.

[0105] While determining the analysis reference time, the system automatically executes the process of generating and saving event waveform data:

[0106] The event waveform data originates from the circular buffer used to store the most recent instantaneous values ​​during normal operation, as well as the instantaneous values ​​of various electrical quantities acquired in real time after t0. The saved "event waveform data" is a continuous high-sampling-rate waveform data, with its recording start point earlier than the analysis reference time t0 and its recording end point later than the analysis reference time t0. Specifically, it includes:

[0107] Pre-triggered data segment: The system extracts and fixes the data from the circular buffer at time (t0-). The instantaneous value data from t0 to t0. Among them, The preset pre-trigger time length (e.g., 5 seconds) is used to capture the steady-state state before the fault occurs and the start-up transient process.

[0108] Post-trigger data segment: The system continues to collect and record instantaneous data from time t0 in real time until the waveform recording termination condition is met (e.g., the residual current is lower than the return threshold (0.3)). The recording continues for a period of time (e.g., 2 seconds) or is forcibly terminated when the maximum recording duration (e.g., 10 seconds) is reached. After recording stops, a trailing time (e.g., 0.2 seconds) of data can be optionally appended to fully capture the state after the transient process ends.

[0109] The pre-triggered data segment and the triggered data segment (including trailing data) are spliced ​​together in chronological order to form a complete event waveform data file containing information before and after time t0, and then saved to non-volatile memory.

[0110] After an event recording ends, the system can start a configurable re-entry suppression timer (e.g., 30 seconds). Before the re-entry suppression timer expires, even if the triggering condition is detected to be met again, the system will not start a new event recording to avoid repeatedly recording the same continuous or recurring leakage process, thus saving storage resources.

[0111] Through step S2, the present invention ensures that each analysis is based on a unified starting point (analysis reference time t0) and a complete waveform (including the process before and after the fault), providing a reliable and unified data foundation for subsequent high-precision phase identification.

[0112] Step S3: Based on the event recording data, within the first analysis window containing the analysis reference time, calculate the fundamental phase angle difference between the residual current and the current of each phase, and calculate the phase angle consistency score based on the fundamental phase angle difference; determine whether the corresponding phase is a leakage phase based on the phase angle consistency score.

[0113] Step S3, based on the event recording data, performs leakage current phase determination within the first analysis window.

[0114] Taking the analysis reference time t0 as the boundary, the first analysis window can be divided into a pre-trigger analysis window and a post-trigger analysis window. The pre-trigger analysis window can be represented as follows: The analysis window after triggering can be represented as The first analysis window can be represented as .in, This indicates a backward lookup from the baseline time of the analysis. The duration of a power frequency cycle (i.e., the length of the pre-trigger event). This indicates the preset action buffer time (e.g., 0.1s).

[0115] For example, suppose , ,but , , .

[0116] In a specific embodiment of the present invention, the fundamental phase angle difference between the residual current and each phase current is calculated, and a phase angle consistency score is calculated based on the fundamental phase angle difference, including:

[0117] Step S3.1: For each power frequency cycle within the first analysis window, use the discrete Fourier transform or integer cycle integral algorithm to calculate the residual current and the fundamental complex phasor of each phase current.

[0118] Taking the whole-cycle integration algorithm as an example, the specific formulas for the fundamental complex phasors of the residual current and each phase current are as follows:

[0119] , (1)

[0120] in, The fundamental complex phasor of the k-th current within the n-th power frequency cycle is represented; T represents the power frequency cycle. This represents the instantaneous value of the k-th current at the t-th sampling time; Indicates the rated frequency of the power grid. = 1 / T; t represents the time variable; j represents the imaginary unit; These correspond to the currents of phases A, B, and C, respectively. Corresponding to the residual current. When k is A. Let k be the fundamental complex phasor of phase A current in the nth power frequency cycle; when k is hour, Let be the fundamental complex phasor of the residual current in the nth power frequency cycle.

[0121] Step S3.2: Based on the residual current and the fundamental complex phasors of each phase current. Calculate the corresponding fundamental complex phase angles respectively, using the following formulas:

[0122] , (2)

[0123] in, Let represent the fundamental complex phase angle of the k-th current within the n-th power frequency cycle, and arg represent the phase angle taken as a complex number.

[0124] Step S3.3: Calculate the fundamental phase angle difference between the residual current and the current in each phase, and perform a foldback process. The specific formula is as follows:

[0125] , (3)

[0126] in, This represents the fundamental phase angle difference between the residual current and the k-th phase current in the nth power frequency cycle after the reversal process. The complex phase angle of the residual current in the nth power frequency cycle represents the fundamental phase angle; wrap represents the phase angle foldback operator, used to fold the phase angle difference back to the point with a 2π period. Within the range, to eliminate the ambiguity caused by integer phase shift.

[0127] Step S3.4: Based on the return processing results, calculate the phase angle consistency score. The specific formula is as follows:

[0128] , (4)

[0129] (5)

[0130] (6)

[0131] in, The phase angle consistency score of the k-th phase is represented, and its value ranges from [0,1]. The larger the value, the more consistent the phase current and the fundamental phase of the residual current are within the first analysis window. The weight of phase k in the nth power frequency cycle is defined as the proportion of the effective value of phase k current in the total effective value of the three-phase current. This represents the effective value of the k-th phase current within the nth power frequency cycle. , , These represent the effective values ​​of the currents in phases A, B, and C respectively within the nth power frequency cycle; This indicates the number of power frequency cycles within the first analysis window.

[0132] Step S3.5: Determine whether the corresponding phase is a leakage phase based on the phase angle consistency score.

[0133] The phase consistency score of the three phases is , , Find the maximum value among them. and the second largest value ;

[0134] like First scoring threshold (e.g., 0.6) and The second scoring threshold (e.g., 0.2) then the maximum value The corresponding phase is the leakage phase; otherwise, the output phase is uncertain.

[0135] Through step S3, this invention utilizes short-term, high-precision waveform recording data from the initial stage of a fault, and through rigorous phasor analysis and statistical scoring, achieves rapid, stable, and automated determination of the leakage phase.

[0136] Step S4: Define a pre-trigger sub-window and a post-trigger sub-window, with the analysis reference time as the boundary. The pre-trigger sub-window and the post-trigger sub-window together constitute the second analysis window. Based on the instantaneous values ​​of each electrical quantity, generate a statistical sequence of each electrical quantity in the second analysis window aggregated according to a fixed time period. According to the statistical sequence, calculate the mean transition of the residual current, the synchronization parameter of the residual current and the total three-phase current, and the duration of the residual current exceeding the threshold. Determine the leakage current type based on the mean transition, synchronization parameter, and duration.

[0137] Step S4 is based on the instantaneous value data collected during normal operation, and performs leakage current type determination in the second analysis window.

[0138] Using the analysis baseline time t0 as the boundary, define the pre-trigger sub-windows respectively. and triggered child window The second analysis window is composed of the pre-trigger sub-window and the post-trigger sub-window. .in, This indicates the observation time before the event (e.g., 10s to 60s), used to characterize the steady-state background before the event. This indicates the observation duration after the event is triggered, used to capture the ongoing state after the event (e.g., 30s to 180s).

[0139] Within the second analysis window, the instantaneous value sequences of three-phase current, residual current, and zero-ground voltage are aggregated:

[0140] Time block division: The second analysis window is divided into M consecutive and non-overlapping time windows, using a fixed time period (preferably 1 second). The time range of the m-th time window is [t]. m ,t m +T s ), where t1 = t0 - .

[0141] For each time window m and each electrical quantity x, the effective value is calculated based on all instantaneous values ​​collected at sampling intervals Δt within the time window, and is used as the statistic for that time window. The calculation formula is:

[0142] (7)

[0143] in, This represents the statistical value of electrical quantity x within the m-th time window; N represents the p-th instantaneous value of electrical quantity x within the m-th time window; N = T s / △t, N represents the number of sampling points in the m-th window, T s Indicates the size of the time window.

[0144] Arrange the statistics of each electrical quantity in chronological order across all time windows to obtain the phase A current I. A [m], Phase B current I B [m], C-phase current I C [m], Residual Current I Δ [m] is a statistical sequence (m=1,2,...,M). Simultaneously, the statistical sequence I of the total three-phase current within each time window is calculated. Σ [m]: .

[0145] Based on the generated statistical sequence, three key features are calculated:

[0146] Mean transition Calculate the first average value of the residual current statistics sequence within the sub-window before triggering and the second average value of the residual current statistics sequence within the sub-window after triggering. The absolute difference between the first average value and the second average value is the mean transition value. The specific formula is:

[0147] (8)

[0148] in, This represents the second average value of the sequence of residual current statistics within the sub-window after triggering. This represents the first average value of the sequence of residual current statistics within the sub-window before triggering.

[0149] Synchronization parameter: defined as the correlation coefficient between the residual current statistics sequence and the total three-phase current statistics sequence within the sub-window after triggering, such as the Pearson correlation coefficient.

[0150] Duration: Counts the remaining current statistics within the sub-window after triggering when the value exceeds the set first threshold. The number of residual current statistics (or time windows).

[0151] The three calculated key features are compared with preset thresholds, and the leakage type is determined according to the following rules:

[0152] User neutral / ground connection error criterion: If the following conditions are met simultaneously:

[0153] If the mean transition is greater than or equal to the preset fifth threshold (0.3) If the synchronization parameter is greater than or equal to the preset sixth threshold (e.g., 0.7) and the duration is less than or equal to the seventh threshold (e.g., 10s), then the leakage current type is determined to be a user's neutral ground wiring error.

[0154] Insulation failure criterion: If the following conditions are met simultaneously:

[0155] If the mean transition amount is less than the preset fifth threshold, the synchronization parameter is less than the preset sixth threshold, and the duration is greater than or equal to the eighth threshold (e.g., 60s), then the leakage type is determined to be insulation failure.

[0156] Other scenarios: If the key features do not meet all the conditions of any of the above groups, the type of this event will be marked as "pending review". It is recommended to combine manual analysis or supplementary data for further judgment.

[0157] Through step 4, this invention utilizes long-term steady-state statistical characteristics to effectively distinguish between instantaneous step-type wiring errors and long-term continuous insulation damage, thus achieving accurate classification of leakage faults.

[0158] Step S5: Output the leakage phase and leakage type.

[0159] Based on the determination results of steps S3 and S4, output the leakage phase and leakage type.

[0160] Table 1 shows five preset leakage scenarios. The type of each scenario, the determination of the leakage phase, and the description of the leakage condition help in the classification and location of leakage faults. The scenarios in Table 1 are set with different combinations of loads, fault types, and phases. These settings can simulate various leakage fault conditions and provide data support for fault location.

[0161] During the positioning process, the phase angle consistency score is calculated based on the phase difference between the residual current and the fundamental phase of each phase current. Determine the phase in which the leakage occurred (e.g.) Figures 2 to 6 (As shown). Figures 2 to 6 The following five leakage current scenarios were demonstrated for phases A, B, and C. Contrast; when 0.6 and When the value is 0.2, the corresponding phase is taken as the leakage phase; otherwise, the output phase is uncertain.

[0162] Table 1 Five Leakage Scenarios

[0163]

[0164] Tests were conducted on the five typical leakage scenarios predefined in Table 1. The method of this invention achieved correct diagnosis in all scenarios, with a location accuracy rate of 100% and a false alarm rate and false negative rate of 0% (see Table 2). The test results show that the method of this invention can achieve accurate location and reliable diagnosis of leakage faults in transformer substations, and has excellent engineering applicability and accuracy.

[0165] Table 2 Positioning Result Accuracy

[0166]

[0167] Example 2

[0168] This invention also provides an electronic device, which includes a memory, a processor, and a computer program or instructions stored in the memory. The processor executes the computer program or instructions to implement the transformer substation leakage fault diagnosis method based on fault recording in this invention.

[0169] Although not shown, the electronic device includes a processor that can perform various appropriate operations and processes based on programs and / or data stored in read-only memory (ROM) or loaded from a storage portion into random access memory (RAM). The processor can be a multi-core processor or may contain multiple processors. In some embodiments, the processor may include a general-purpose main processor and one or more specialized coprocessors, such as a central processing unit, graphics processing unit (GPU), neural network processor (NPU), digital signal processor (DSP), etc. Various programs and data required for device operation are also stored in RAM. The processor, ROM, and RAM are interconnected via a bus. Input / output (I / O) interfaces are also connected to the bus.

[0170] The processor and memory described above are used together to execute programs / instructions stored in the memory. When the program / instructions are executed by the computer, they can implement the methods, steps, or functions described in the above embodiments.

[0171] Although not shown, embodiments of the present invention also provide a computer-readable storage medium having a computer program or instructions stored thereon, which, when executed by a processor, implements the transformer substation leakage fault diagnosis method based on fault recording in embodiments of the present invention.

[0172] Readable storage media include both permanent and non-permanent, removable and non-removable media that can store information by any method or technology. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0173] The above description only discloses specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or modifications that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for diagnosing transformer substation leakage faults based on fault recording, characterized in that, The diagnostic method includes: Under normal operation, the instantaneous values ​​of various electrical quantities on the low-voltage side of the transformer area are continuously collected at sampling intervals; wherein, the electrical quantities include at least three-phase current and residual current; When the preset abnormal event triggering conditions are met, the moment when the conditions are first met is determined as the analysis baseline moment; acquire and save a segment of event waveform data including the analysis baseline moment, wherein the recording start point of the waveform data is earlier than the analysis baseline moment; Based on the event recording data, within the first analysis window containing the analysis reference time, the fundamental phase angle difference between the residual current and the current of each phase is calculated, and a phase angle consistency score is calculated based on the fundamental phase angle difference; the corresponding phase is determined to be a leakage phase based on the phase angle consistency score. Using the aforementioned analysis reference time as a boundary, a pre-trigger sub-window and a post-trigger sub-window are defined, which together constitute the second analysis window. Based on the instantaneous values ​​of each electrical quantity, a statistical sequence of each electrical quantity within the second analysis window is generated, aggregated according to a fixed time period. Based on the statistical sequence, the mean transition value of the residual current, the synchronization parameter between the residual current and the total three-phase current, and the duration of the residual current exceeding the threshold are calculated. The leakage current type is determined based on the mean transition value, the synchronization parameter, and the duration. The calculation of the fundamental phase angle difference between the residual current and each phase current, and the calculation of the phase angle consistency score based on the fundamental phase angle difference, includes: For each power frequency cycle within the first analysis window, calculate the residual current and the fundamental complex phasor of each phase current; Calculate the corresponding fundamental complex phase angles based on the residual current and the fundamental complex phasors of each phase current; Calculate the fundamental phase angle difference between the residual current and the current in each phase, and perform a back-flipping process; Calculate the phase angle consistency score based on the return processing results; The determination of whether a corresponding phase is a leakage phase based on the phase angle consistency score includes: The maximum and second-largest values ​​are determined from the phase angle consistency scores of phases A, B, and C; If the maximum value is greater than or equal to the first scoring threshold, and the difference between the maximum value and the second largest value is greater than or equal to the second scoring threshold, then the phase corresponding to the maximum value is determined to be the leakage phase.

2. The method for diagnosing transformer substation leakage faults based on fault recording as described in claim 1, characterized in that, The electrical quantity also includes zero-ground voltage, and the preset abnormal event triggering condition includes at least one of the following: The instantaneous value of the residual current or the value in the statistical sequence exceeds a set first threshold. The rate of change of the residual current exceeds a preset second threshold. The difference between the mean values ​​of residual current statistics in adjacent time units exceeds a preset third threshold. The sudden change in zero-ground voltage at adjacent sampling times exceeds the preset fourth threshold.

3. The method for diagnosing transformer substation leakage faults based on fault recording as described in claim 1, characterized in that, Based on the instantaneous values ​​of each electrical quantity, a statistical sequence of each electrical quantity within the second analysis window, aggregated according to a fixed time period, is generated, including: Within the second analysis window, multiple consecutive and non-overlapping time windows are divided using the fixed time period as the unit; for each time window, the effective value is calculated based on all instantaneous values ​​of the electrical quantity within it, which serves as the statistical quantity of that electrical quantity in that time window; the statistical quantities of each time window are arranged in chronological order to form the statistical quantity sequence. The formula for calculating the corresponding statistic for a given time window is as follows: ； in, This represents the statistical value of electrical quantity x within the m-th time window; N represents the p-th instantaneous value of electrical quantity x within the m-th time window; N = T s / △t, N represents the number of sampling points in the m-th window, T s The value represents the size of the time window, and Δt represents the sampling interval.

4. The method for diagnosing transformer substation leakage faults based on fault recording as described in claim 1, characterized in that, The calculation of the mean transition of the residual current includes: Calculate the first average value of the residual current statistics sequence in the sub-window before triggering and the second average value of the residual current statistics sequence in the sub-window after triggering; Calculate the absolute difference between the first average value and the second average value; the absolute difference is the mean transition amount.

5. The method for diagnosing transformer substation leakage faults based on fault recording as described in claim 1, characterized in that, The synchronization parameters of the residual current and the total three-phase current are calculated as follows: Calculate the correlation coefficient between the residual current statistics sequence and the total three-phase current statistics sequence within the sub-window after the trigger.

6. The method for diagnosing transformer substation leakage faults based on fault recording as described in claim 1, characterized in that, The leakage current type is determined based on the mean transition value, synchronization parameters, and duration, including: If the mean transition amount is greater than or equal to the preset fifth threshold, the synchronization parameter is greater than or equal to the preset sixth threshold, and the duration is less than or equal to the seventh threshold, then the leakage current type is determined to be a user ground connection error. If the mean transition amount is less than the preset fifth threshold, the synchronization parameter is less than the preset sixth threshold, and the duration is greater than or equal to the eighth threshold, then the leakage type is determined to be insulation failure. The eighth threshold is greater than the seventh threshold.

7. An electronic device comprising a memory, a processor, and a computer program or instructions stored in the memory, characterized in that, The processor executes the computer program or instructions to implement the transformer substation leakage fault diagnosis method based on fault recording as described in any one of claims 1 to 6.

8. A computer-readable storage medium having a computer program or instructions stored thereon, characterized in that, When the computer program or instructions are executed by the processor, they implement the method for diagnosing leakage faults in transformer substations based on fault recording as described in any one of claims 1 to 6.

Images