Fault positioning method and device based on threshold adaptive calibration

Abstract

The invention provides a fault positioning method and device based on threshold adaptive calibration, and relates to the technical field of fault positioning, and the method comprises the steps: obtaining a current signal sequence, calculating a harmonic pollution index, and recognizing the stable state of a load; extracting a current signal probability density distribution curve and a quantile set; dynamically updating high and low quantile thresholds, and triggering anomaly detection to calculate fault confidence; when the confidence exceeds a threshold, a fault indicator is triggered to act; a fault section is determined by weighted fusion of the plurality of indicator information. The fault positioning accuracy can be improved, and the false alarm rate can be reduced.

Description

Technical Field

[0001] This invention relates to the field of fault location, and in particular to a fault location method and apparatus based on threshold adaptive calibration. Background Technology

[0002] With the continuous development of power distribution network construction, the structure of power distribution networks is becoming increasingly complex, and the frequency of faults is increasing, posing a huge challenge to the operation and maintenance of power distribution systems. Fault indicators, as commonly used fault monitoring devices in power distribution lines, can monitor line current signals in real time and detect faults. They are of great significance for quickly locating faulty sections, shortening fault handling time, and improving power supply reliability.

[0003] Traditional fault indicator technology typically relies on fixed threshold judgments; that is, a fault alarm is triggered when the monitored current signal exceeds a preset fixed threshold. However, this method has several problems in practical applications. First, load fluctuations in power distribution lines are time-varying and random, making it difficult for fixed thresholds to adapt to complex and changing operating environments, easily leading to false alarms or missed alarms. Second, harmonic pollution in power distribution systems is becoming increasingly serious, with more interference signals, making it difficult for traditional fault indicators to effectively distinguish between normal load fluctuations and fault states, reducing the accuracy of fault detection. Third, existing fault indicators usually operate independently, lacking an effective information fusion mechanism, and cannot comprehensively utilize information from multiple fault indicators for collaborative judgment, affecting the accuracy and reliability of fault location. Summary of the Invention

[0004] The fault location method, apparatus, electronic device, storage medium, and computer program products based on threshold adaptive calibration provided by the embodiments of the present invention improve the accuracy of fault location in the distribution network to a certain extent.

[0005] In a first aspect, the present invention provides a fault location method based on threshold adaptive calibration, the method comprising:

[0006] The current signal sequence collected by the fault indicator in the power distribution line is obtained, the harmonic pollution index is calculated based on the harmonic components and fundamental components of the current signal sequence, and the load stable operation status of the line monitored by the fault indicator is identified in combination with the stability criteria.

[0007] For the time period corresponding to the stable operation state of the load, the probability density distribution curve of the current signal is extracted, and the set of quantiles of the probability density distribution curve is calculated.

[0008] Based on the set of quantiles, a dynamic quantile tracking algorithm is used to update the high quantile threshold representing the upper bound of normal current fluctuation and the low quantile threshold representing the lower bound of normal current fluctuation in real time. When the current signal exceeds the high quantile threshold or falls below the low quantile threshold, anomaly detection is triggered, and the fault confidence is calculated based on the magnitude and duration of the exceedance.

[0009] When the fault confidence level exceeds the preset confidence threshold, the fault indicator is triggered and the location and time of the fault in the power distribution line are recorded.

[0010] The fault confidence level of the fault indicator is weighted and fused with the action time. Based on the spatial distribution gradient of the fault confidence level, the propagation direction and attenuation law of the fault current are determined. Based on the propagation direction and the attenuation law, the fault occurrence section in the power distribution line is determined.

[0011] In one embodiment of the present invention, a current signal sequence collected by a fault indicator in a power distribution line is acquired, a harmonic pollution index is calculated based on the harmonic components and fundamental components of the current signal sequence, and a stability criterion is combined to identify the load stability operation status of the line monitored by the fault indicator, including:

[0012] The current signal sequence is subjected to frequency domain transformation to extract the amplitude and phase of the fundamental component and each harmonic component, and the energy ratio of the harmonic component to the fundamental component is calculated.

[0013] The energy of each harmonic in the harmonic component is weighted and summed with the energy of the fundamental component. The harmonic pollution index is calculated based on the ratio of the weighted sum to the energy of the fundamental component.

[0014] A time series of the harmonic pollution index is constructed in chronological order, and the difference between the harmonic pollution index at adjacent moments in the time series is calculated to obtain a sequence of the rate of change of the harmonic pollution index.

[0015] A sliding window statistical analysis was performed on the harmonic pollution index change rate sequence to calculate the variance of the harmonic pollution index change rate within the sliding window;

[0016] The variance is compared with a preset stability criterion. When the variance is less than the stability criterion, the time period corresponding to the sliding window is determined to be the load stable operation state. When the variance is greater than or equal to the stability criterion, the time period corresponding to the sliding window is determined to be the load switching state.

[0017] In one embodiment of the present invention, for a time period corresponding to the stable operation state of the load, the probability density distribution curve of the current signal is extracted, and the set of quantiles of the probability density distribution curve is calculated, including:

[0018] The amplitude samples of the current signal are extracted from the time period corresponding to the stable operation state of the load, and the amplitude samples are numerically sorted to obtain an ordered amplitude sequence.

[0019] Based on the ordered amplitude sequence, a probability density distribution curve of the current signal is constructed using a kernel density estimation method. The kernel density estimation method obtains a continuous probability density function by applying a kernel function to each amplitude sample in the ordered amplitude sequence and superimposing them; the cumulative distribution function is obtained by integrating the probability density function.

[0020] Based on the cumulative distribution function, the quantile values ​​corresponding to multiple preset probability levels are calculated. The multiple preset probability levels include a high probability level representing the upper bound of the current signal and a low probability level representing the lower bound of the current signal. The quantile values ​​corresponding to the multiple preset probability levels are combined to form the quantile set.

[0021] In one embodiment of the present invention, based on the set of quantiles, a dynamic quantile tracking algorithm is used to update the high quantile threshold representing the upper bound of normal current fluctuations and the low quantile threshold representing the lower bound of normal current fluctuations in real time, including:

[0022] Construct a sliding time window for the amplitude of the current signal. Within the sliding time window, count the number of times the amplitude of the current signal exceeds the initial value of a preset high quantile threshold and the number of times it falls below the initial value of a preset low quantile threshold. Calculate the high-end default rate and the low-end default rate based on the number of times the amplitude exceeds the initial value of a preset high quantile threshold and the number of times it falls below the initial value of a preset low quantile threshold, respectively.

[0023] The high-end default rate is compared with a preset target high-end default rate to calculate the high-end default rate deviation. A high-end threshold correction amount is constructed based on the high-end default rate deviation. The high-end threshold correction amount is proportional to the high-end default rate deviation. When the high-end default rate is greater than the target high-end default rate, the high-end threshold correction amount is positive to increase the high percentile threshold. When the high-end default rate is less than the target high-end default rate, the high-end threshold correction amount is negative to decrease the high percentile threshold.

[0024] The low-end default rate is compared with a preset target low-end default rate to calculate the low-end default rate deviation. A low-end threshold correction amount is constructed based on the low-end default rate deviation. The low-end threshold correction amount is proportional to the low-end default rate deviation. When the low-end default rate is greater than the target low-end default rate, the low-end threshold correction amount is negative to reduce the low percentile threshold. When the low-end default rate is less than the target low-end default rate, the low-end threshold correction amount is positive to increase the low percentile threshold.

[0025] The high-end threshold correction is superimposed on the initial value of the high quantile threshold to obtain the updated high quantile threshold, and the low-end threshold correction is superimposed on the initial value of the low quantile threshold to obtain the updated low quantile threshold.

[0026] In one embodiment of the present invention, calculating the fault confidence level based on the excess amplitude and duration includes:

[0027] Obtain the magnitude of the current signal amplitude exceeding the high quantile threshold, calculate the ratio of the magnitude of the excess amplitude to the high quantile threshold, and obtain the normalized excess amplitude characterizing the degree of excess.

[0028] The duration for which the amplitude of the current signal exceeds the high quantile threshold is obtained. The ratio of the duration to the preset reference duration is nonlinearly transformed. The nonlinear transformation uses a logarithmic function mapping to obtain a normalized duration index characterizing the persistence.

[0029] The normalized exceedance magnitude and the normalized duration index are weighted and multiplied, with the weight coefficient of the normalized exceedance magnitude being greater than the weight coefficient of the normalized duration index, to obtain the fault confidence.

[0030] In one embodiment of the present invention, the number of fault indicators is multiple, and the method further includes:

[0031] For each fault indicator, a binary feature vector containing the fault confidence and the action time is constructed, and the spatial distance between any two fault indicators is calculated based on the installation location coordinates of the fault indicator.

[0032] Extract the action times of multiple fault indicators, pair the action times together, calculate the time difference between the action times of each pair of fault indicators, and associate the time difference with the corresponding spatial distance; based on the association between the time difference and the spatial distance, use the least squares fitting method to establish a linear relationship model between the time difference and the spatial distance; the slope of the linear relationship model characterizes the propagation speed of the fault current in the power distribution line.

[0033] In one embodiment of the present invention, the propagation direction and attenuation law of the fault current are determined based on the spatial distribution gradient of the fault confidence, and the fault occurrence section in the power distribution line is determined based on the propagation direction and the attenuation law, including:

[0034] Based on the propagation speed, the fault indicator with the earliest action time is selected as the initial reference point. Starting from the initial reference point, along the negative gradient direction of the gradient vector field of the fault confidence, and combining the action time of other fault indicators with the propagation speed, the theoretical propagation time from the initial reference point to the position of each fault indicator is calculated. The time difference between the theoretical propagation time and the actual action time is compared to construct a time deviation index that characterizes the consistency between theoretical propagation and actual action.

[0035] The fault confidence values ​​of each fault indicator are weighted and corrected according to the time deviation index, and the weighted and corrected comprehensive fault confidence is calculated.

[0036] The comprehensive fault confidence is spatially mapped on the distribution line topology to identify the peak region where the comprehensive fault confidence occurs. Within the peak region, the reverse propagation path of the fault current is determined according to the negative gradient direction of the gradient vector field of the fault confidence.

[0037] Along the reverse propagation path, combining the propagation speed with the action time of each fault indicator, trace back to the position where the action time is extrapolated to zero, determine the position as the fault source point, and determine the power distribution line section where the fault source point is located as the fault occurrence section.

[0038] Secondly, the present invention provides a fault location device based on threshold adaptive correction, the fault location device based on threshold adaptive correction comprising:

[0039] The status identification module is used to acquire the current signal sequence collected by the fault indicator in the power distribution line, calculate the harmonic pollution index based on the harmonic components and fundamental components of the current signal sequence, and identify the load stable operation status of the line monitored by the fault indicator in combination with the stability criteria.

[0040] The quantile calculation module is used to extract the probability density distribution curve of the current signal for the time period corresponding to the stable operation state of the load, and calculate the set of quantiles of the probability density distribution curve.

[0041] The confidence calculation module is used to update the high quantile threshold representing the upper bound of normal current fluctuation and the low quantile threshold representing the lower bound of normal current fluctuation in real time based on the quantile set and the dynamic quantile tracking algorithm. When the current signal exceeds the high quantile threshold or falls below the low quantile threshold, anomaly detection is triggered, and the fault confidence is calculated based on the magnitude and duration of the exceedance.

[0042] The action triggering module is used to trigger the fault indicator and record the location and time of the fault in the power distribution line when the fault confidence exceeds the preset confidence threshold.

[0043] The fault location module is used to perform weighted fusion of the fault confidence level and the action time of the fault indicator, determine the propagation direction and attenuation law of the fault current based on the spatial distribution gradient of the fault confidence level, and determine the fault occurrence section in the power distribution line based on the propagation direction and the attenuation law.

[0044] Thirdly, the present invention provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the fault location method based on threshold adaptive correction as described above.

[0045] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the fault location method based on threshold adaptive correction as described in any of the preceding claims.

[0046] Fifthly, the present invention provides a computer program product, including a computer program, which, when executed by a processor, causes the computer to perform the fault location method based on threshold adaptive correction as described in any of the preceding claims.

[0047] The technical solution provided by this invention identifies the stable operating state of the load by calculating the harmonic pollution index and combining it with stability criteria, avoiding misjudgments caused by threshold calibration under non-steady-state conditions and improving the accuracy of fault identification. Based on the probability density distribution curve of the current signal, a quantile set is extracted, and a dynamic quantile tracking algorithm is used to update the high and low quantile thresholds in real time, enabling the fault judgment threshold to be adaptively adjusted, overcoming the shortcomings of traditional fixed threshold methods that cannot adapt to dynamic load changes.

[0048] By introducing a fault confidence calculation mechanism that comprehensively considers both the magnitude and duration of current anomalies, false alarms caused by instantaneous disturbances are reduced, improving the reliability of fault identification. Through weighted fusion analysis of information from multiple fault indicators, and utilizing the spatial distribution gradient of fault confidence and the time difference of action, the propagation direction and attenuation law of fault current can be accurately inferred, achieving more precise fault segment location, reducing the fault inspection range, and shortening fault handling time. Attached Figure Description

[0049] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention, and those skilled in the art can obtain other embodiments based on these drawings without creative effort.

[0050] Figure 1 This is a flowchart illustrating a fault location method based on threshold adaptive calibration provided in an embodiment of the present invention.

[0051] Figure 2 This is a schematic diagram of the fault location process in a power distribution network according to an embodiment of the present invention.

[0052] Figure 3 This is a schematic diagram of the dynamic quantile threshold update process provided in an embodiment of the present invention.

[0053] Figure 4 This is a schematic diagram of a fault location method based on threshold adaptive calibration provided in an embodiment of the present invention.

[0054] Figure 5 This is a schematic diagram of an electronic device provided according to an embodiment of the present invention. Detailed Implementation

[0055] Embodiments of the present invention will now be described in more detail with reference to the accompanying drawings. While some embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of the present invention. It should be understood that the drawings and embodiments of the present invention are for illustrative purposes only and are not intended to limit the scope of protection of the present invention.

[0056] Please see Figure 1 and Figure 2 The present invention provides a fault location method based on threshold adaptive calibration. This method may include the following steps.

[0057] Step S110: Obtain the current signal sequence collected by the fault indicator in the power distribution line, calculate the harmonic pollution index based on the harmonic components and fundamental components of the current signal sequence, and identify the load stable operation status of the line monitored by the fault indicator in combination with the stability criterion.

[0058] In this embodiment, the fault indicator refers to a monitoring device installed on the power distribution line to collect electrical signals such as line current and voltage, and to trigger actions (such as alarms or recordings) according to preset conditions. The fault indicator is installed at a critical node of the power distribution line, and collects the current signal flowing through that node in real time through a current sensor. The current sensor continuously samples the current signal at a fixed sampling frequency to form time-series data. The selection of the sampling frequency needs to satisfy the Nyquist sampling theorem to ensure accurate capture of high-frequency harmonic components in the current signal. The collected current signal sequence contains a fundamental component and multiple harmonic components; the fundamental component corresponds to the power frequency current, while the harmonic components are generated by nonlinear loads.

[0059] In this embodiment, the current signal sequence refers to the set of continuous current data collected by the fault indicator at a fixed sampling frequency, which is the basic data for subsequent analysis. Harmonic components are components in the current signal whose frequency is an integer multiple of the fundamental frequency (such as the 3rd harmonic and the 5th harmonic), generated by factors such as nonlinear loads, and reflect the degree of distortion of the current signal. The fundamental component is the component in the current signal whose frequency is the rated frequency of the power grid (50Hz in my country), and is the main component of the current signal. The harmonic pollution index is an indicator that quantifies the degree of harmonic interference to the fundamental frequency, reflecting the quality and stability of the current signal. The calculation method of the harmonic pollution index can be expressed by formula (1):

[0060] Formula (1)

[0061] in, Let H be the effective value of the h-th harmonic voltage, where h = 2, 3, ..., H, and H is the highest harmonic order. This is the fundamental current.

[0062] Specifically, for example, the acquired current signal sequence is subjected to Fourier transform processing to convert the time-domain signal into a frequency-domain representation. In the frequency domain, the amplitude of the fundamental component and the amplitudes of each harmonic component are separated. The fundamental component corresponds to the power frequency, and the harmonic components correspond to integer multiples of the power frequency. The amplitudes of the second to fifteenth harmonics are extracted, the sum of the squares of each harmonic amplitude is calculated, and the square root of this sum is taken to obtain the total harmonic amplitude. The total harmonic amplitude is divided by the fundamental amplitude to obtain the harmonic pollution index. The harmonic pollution index reflects the proportion of harmonic components to the fundamental component in the current signal; the larger the index, the more severe the harmonic pollution of the current signal.

[0063] In this embodiment, the stability criterion refers to a preset threshold used to determine the load operating state, which serves as the basis for distinguishing between stable and switching load states. A stable load operating state refers to an operating state where parameters such as load power and current of the distribution line fluctuate relatively little; the current data at this time can be used to construct a reliable threshold. That is, when the harmonic pollution index is greater than the preset stability criterion, it indicates that the current distribution network is in an unstable operating state; conversely, when the harmonic pollution index is less than or equal to the preset stability criterion, it indicates that the current distribution network is in a stable operating state.

[0064] Step S120: For the time period corresponding to the stable operation state of the load, extract the probability density distribution curve of the current signal and calculate the quantile set of the probability density distribution curve.

[0065] In some embodiments, the probability density distribution curve refers to a continuous curve describing the probability of current amplitude samples appearing in different numerical ranges, reflecting the distribution pattern of current amplitude. The quantile set is a set of current amplitude values ​​corresponding to multiple preset probability levels, providing an initial reference for the dynamic threshold.

[0066] In this embodiment, after identifying all time periods corresponding to the stable operation of the load, the current signal amplitude data within these time periods are extracted. The current signal amplitude data from all stable operation time periods are aggregated to form a current amplitude sample set representing the normal operating state. Statistical analysis is performed on this sample set to calculate its probability density distribution. The probability density distribution describes the frequency of current amplitude occurrences in different numerical ranges. The sample data is smoothed using a kernel density estimation method to obtain a continuous probability density distribution curve. The horizontal axis of this curve represents the current amplitude, and the vertical axis represents the probability density of the corresponding amplitude occurrence.

[0067] Step S130: Based on the set of quantiles, the high quantile threshold representing the upper limit of normal current fluctuation and the low quantile threshold representing the lower limit of normal current fluctuation are updated in real time using a dynamic quantile tracking algorithm. When the current signal exceeds the high quantile threshold or falls below the low quantile threshold, anomaly detection is triggered, and the fault confidence is calculated based on the magnitude and duration of the exceedance.

[0068] In this embodiment, the dynamic quantile tracking algorithm refers to an algorithm that can update the quantile threshold based on real-time current data, thereby achieving adaptive adjustment of the threshold. The high quantile threshold represents the upper limit of normal current fluctuation (e.g., 99th quantile); if the current exceeds this threshold, a fault may exist. The low quantile threshold represents the lower limit of normal current fluctuation (e.g., 1% quantile); if the current is below this threshold, a fault may exist (e.g., a wire breakage fault).

[0069] In this embodiment, to enable the thresholds to adapt to slow load changes, a dynamic quantile tracking algorithm is employed to update the high and low quantile thresholds in real time. The dynamic quantile tracking algorithm is based on a sliding time window mechanism. A sliding window of several hours is set, and this window moves forward continuously over time. Whenever a new stable operating period is identified, the current amplitude data for that period is added to the sliding window, while the oldest data of equal magnitude in the window is removed. The probability density distribution curve and quantile set are recalculated for the data within the updated sliding window, thus obtaining the updated high and low quantile thresholds. This update mechanism ensures that the thresholds can track seasonal changes and long-term trends in the load, while maintaining robustness to short-term fluctuations.

[0070] After the threshold update, the real-time acquired current signal amplitude is compared with the high-quantile threshold and the low-quantile threshold. When the current signal amplitude exceeds the high-quantile threshold at a certain moment, this moment is recorded as the start of the anomaly, and the excess amplitude is calculated. The excess amplitude equals the difference between the actual current amplitude and the high-quantile threshold. The current signal is continuously monitored at subsequent moments, and when the current signal amplitude falls back below the high-quantile threshold, this moment is recorded as the end of the anomaly. The difference between the end of the anomaly and the start of the anomaly is the duration. Similarly, when the current signal amplitude is below the low-quantile threshold, the same anomaly detection process is performed, calculating the below amplitude and duration.

[0071] In this embodiment, the fault confidence score is an index that quantifies the probability that an abnormal current is a fault, with a value ranging from 0 to 1. The higher the value, the higher the probability of a fault. The fault confidence score can be calculated by combining the magnitude and duration of the current exceeding the threshold, and quantifying the probability of a fault through a weighted product.

[0072] In this embodiment, calculating the fault confidence level requires considering both the exceedance magnitude and duration. The exceedance magnitude is divided by the high quantile threshold to obtain the normalized exceedance magnitude, which represents the proportion of the exceedance relative to the upper bound of normal fluctuations. The duration is divided by a preset baseline duration to obtain the duration ratio. The baseline duration is determined based on the operating time characteristics of the distribution line's protection devices. A logarithmic transformation is performed on the duration ratio to obtain the normalized duration index. The logarithmic transformation converts the linear growth of the duration into logarithmic growth, avoiding excessive influence on the fault confidence level when the duration is too long. The normalized exceedance magnitude and normalized duration index are weighted and combined, with the exceedance magnitude weighting coefficient set to 0.7 and the duration weighting coefficient set to 0.3. The weighted product of the two is used as the initial fault confidence level. The initial fault confidence level is truncated; when it exceeds a preset upper confidence level limit, it is limited to that upper limit to obtain the final fault confidence level.

[0073] Step S140: When the fault confidence level exceeds the preset confidence threshold, the fault indicator is triggered and the location and time of the fault in the power distribution line are recorded.

[0074] In this embodiment, the preset confidence threshold refers to a pre-defined fault confidence threshold (e.g., 0.8) that triggers the fault indicator. In other words, when the fault confidence is greater than or equal to the preset confidence threshold, it indicates a high probability of a fault occurring in the power distribution line, initiating fault probability analysis and recording the location and time of the fault occurrence in the power distribution line. Conversely, when the fault confidence is less than the preset confidence threshold, it indicates a low probability of a fault occurring in the power distribution line.

[0075] In this embodiment, the confidence threshold is determined based on statistical analysis of historical fault data to ensure effective identification of real faults while suppressing false alarms. When the fault confidence level exceeds the confidence threshold, the fault indicator triggers an action, initiating the fault recording process. The recorded information includes the fault indicator's number, geographical coordinates, fault confidence level, anomaly detection start time, anomaly detection end time, and waveform characteristic parameters of the current signal.

[0076] Step S150: The fault confidence of the fault indicator and the action time are weighted and fused. Based on the spatial distribution gradient of the fault confidence, the propagation direction and attenuation law of the fault current are determined. Based on the propagation direction and the attenuation law, the fault occurrence section in the power distribution line is determined.

[0077] In this embodiment, the confidence spatial distribution gradient refers to the rate of change and direction of change of the fault confidence in the distribution line space, reflecting the propagation trend of the fault current. The fault current propagation direction refers to the direction in which the fault current propagates from the fault point to both ends of the line after the fault occurs.

[0078] In this embodiment, the attenuation law refers to the variation of the amplitude of the fault current as it decreases with distance during propagation. The fault occurrence section refers to the line segment in the distribution line where the fault source point is located, which is the core area for fault handling. The propagation direction and attenuation law of the fault current can be determined by analyzing the propagation direction and amplitude attenuation characteristics of the fault current based on the spatial distribution gradient of confidence level, and by tracing back to the fault source point, the fault location section can be determined.

[0079] In the above embodiments, by dynamically updating the threshold, the problem that traditional fixed thresholds cannot adapt to load fluctuations is solved, significantly reducing the false alarm and missed alarm rates. Furthermore, the introduction of a fault confidence metric to quantify the fault probability improves the scientific rigor and reliability of fault judgment, avoiding malfunctions caused by instantaneous fluctuations. Secondly, by integrating the spatiotemporal information of multiple fault indicators, accurate inference of the fault current propagation direction and attenuation pattern is achieved, providing a reliable basis for fault segment location.

[0080] In some embodiments, acquiring a current signal sequence collected by a fault indicator in a power distribution line, calculating a harmonic pollution index based on the harmonic and fundamental components of the current signal sequence, and identifying the load stability operation status of the line monitored by the fault indicator in conjunction with a stability criterion, may include:

[0081] The current signal sequence is subjected to frequency domain transformation to extract the amplitude and phase of the fundamental component and each harmonic component, and the energy ratio of the harmonic component to the fundamental component is calculated.

[0082] The energy of each harmonic in the harmonic component is weighted and summed with the energy of the fundamental component. The harmonic pollution index is calculated based on the ratio of the weighted sum to the energy of the fundamental component.

[0083] A time series of the harmonic pollution index is constructed in chronological order, and the difference between the harmonic pollution index at adjacent moments in the time series is calculated to obtain a sequence of the rate of change of the harmonic pollution index.

[0084] A sliding window statistical analysis was performed on the harmonic pollution index change rate sequence to calculate the variance of the harmonic pollution index change rate within the sliding window;

[0085] The variance is compared with a preset stability criterion. When the variance is less than the stability criterion, the time period corresponding to the sliding window is determined to be the load stable operation state. When the variance is greater than or equal to the stability criterion, the time period corresponding to the sliding window is determined to be the load switching state.

[0086] In some cases, fault indicators serve as crucial monitoring devices in power distribution line monitoring, collecting current signals from the lines. Analyzing these signals enables accurate monitoring of the line load status, supporting the safe and stable operation of the power distribution network.

[0087] In this embodiment, the acquired current signal sequence is subjected to frequency domain transformation to extract the amplitude and phase of the fundamental component and each harmonic component. Frequency domain transformation commonly employs the Fourier transform method to convert the time-domain signal to the frequency domain for analysis. The amplitude and phase information of each frequency component in the current signal can be obtained through Fourier transform. In practice, considering computational efficiency, the Fast Fourier Transform algorithm is typically used for processing.

[0088] After completing the frequency domain transformation, the energy ratio of the harmonic components to the fundamental component is calculated. For current signals, the energy is proportional to the square of the amplitude. The proportion of each harmonic energy to the total energy is calculated as a preliminary characterization of the degree of harmonic pollution.

[0089] Next, the energy of each harmonic component is weighted and summed with the energy of the fundamental component to calculate the harmonic pollution index (HPI). The weighting coefficients are determined based on the degree of influence of each harmonic on the system; typically, higher-order harmonics are given larger weights because they are more harmful to equipment. The HPI is defined as the ratio of the weighted sum to the fundamental energy.

[0090] In practical applications, the weighting coefficients can be adjusted according to the specific characteristics of the power distribution system. For example, for some systems that are particularly sensitive to the third harmonic, the weighting coefficient for the third harmonic can be increased, so that the pollution index can better reflect the actual condition of the system.

[0091] After obtaining the harmonic pollution index, it is constructed into a time series in chronological order. Assuming that the harmonic pollution index is calculated every certain time interval (e.g., one second) during a monitoring period, a harmonic pollution index sequence varying over time can be obtained: {HPI1, HPI2, ..., HPI...} n}

[0092] The difference in harmonic pollution index between adjacent time points is calculated to obtain the harmonic pollution index change rate sequence. The change rate ΔHPI at time t is then calculated. t = HPI t – HPI t-1 This yields the rate of change sequence {ΔHPI2, ΔHPI3, ..., ΔHPI}. n The rate of change of the harmonic pollution index can more intuitively reflect the dynamic changes in load conditions.

[0093] In this embodiment, sliding window statistical analysis refers to an analysis method that selects a fixed-length time window, slides the window along the time axis, and calculates the variance of the data within the window to extract local features of the data. Variance better reflects the degree of data change. By using the variance of the harmonic pollution index over a period of time to replace the harmonic pollution index and compare it with the stability criterion, the interference of a single detection anomaly on the analysis results can be avoided.

[0094] In this embodiment, the length of the sliding window can be set according to actual needs; for example, ten seconds can be selected as the window length. For each window, the variance of the rate of change sequence within the window is calculated to measure the load stability during that time period. It is understood that when performing sliding window statistical analysis on the harmonic pollution index rate of change sequence, there is partial overlap between adjacent sliding windows. For example, when selecting ten seconds as the window length, the overlap time can be two seconds, three seconds, five seconds, etc. Generally, the overlap time does not exceed half the sliding window time length.

[0095] In this embodiment, the calculated variance is compared with a preset stability criterion. The stability criterion is a preset threshold determined based on historical data analysis and expert experience. When the variance is less than the stability criterion, the time period corresponding to the sliding window is determined to be in a stable load operation state; when the variance is greater than or equal to the stability criterion, it is determined to be in a load switching state. The load switching state refers to an operating state where parameters such as power and current of the distribution line load fluctuate significantly (e.g., the start-up and shutdown of high-power user equipment), and the current data in this state is not suitable for threshold calibration.

[0096] The above embodiments realize the effective identification of the stable operating status of distribution line loads based on the current signal sequence collected by the fault indicator, providing important technical support for the safe and stable operation of the distribution network. This method makes full use of the variation characteristics of harmonic components in the current signal, and has the advantages of simple implementation, low computational load, and strong real-time performance, making it suitable for online monitoring systems in smart distribution networks.

[0097] In some embodiments, extracting the probability density distribution curve of the current signal for a time period corresponding to the stable operation state of the load, and calculating the quantile set of the probability density distribution curve, may include:

[0098] The amplitude samples of the current signal are extracted from the time period corresponding to the stable operation state of the load, and the amplitude samples are numerically sorted to obtain an ordered amplitude sequence.

[0099] Based on the ordered amplitude sequence, a probability density distribution curve of the current signal is constructed using a kernel density estimation method. The kernel density estimation method obtains a continuous probability density function by applying a kernel function to each amplitude sample in the ordered amplitude sequence and superimposing them; the cumulative distribution function is obtained by integrating the probability density function.

[0100] Based on the cumulative distribution function, the quantile values ​​corresponding to multiple preset probability levels are calculated. The multiple preset probability levels include a high probability level representing the upper bound of the current signal and a low probability level representing the lower bound of the current signal. The quantile values ​​corresponding to the multiple preset probability levels are combined to form the quantile set.

[0101] In this embodiment, amplitude samples of the current signal are extracted from the time period corresponding to the stable operation of the load. In a power system, current data during load operation can be collected by a current sensor, and the time period of stable operation can be selected through a preprocessing module. For example, current data from one hour of normal operation of the power equipment can be selected, with a sampling frequency of 100Hz, resulting in 360,000 amplitude samples. These amplitude samples are then numerically sorted to obtain an ordered amplitude sequence. The sorting can employ algorithms such as quicksort or heapsort to arrange the original current amplitudes from smallest to largest, facilitating subsequent analysis.

[0102] In this embodiment, a probability density distribution curve of the current signal is constructed based on an ordered amplitude sequence using kernel density estimation. Kernel density estimation is a non-parametric estimation technique that does not rely on specific distribution assumptions about the data. A kernel function is applied to each amplitude sample in the ordered amplitude sequence and then superimposed to obtain a continuous probability density function. Commonly used kernel functions include the Gaussian kernel function and the Epanechnikov kernel function. Taking the Gaussian kernel function as an example, its expression is simple and computationally efficient; for each sample point in the current amplitude sequence... The contribution after applying the kernel function is centered on The normal distribution function at that location.

[0103] In practical applications, the choice of bandwidth for the kernel function is crucial, as it determines the smoothness of the probability density estimation. Empirical rules such as the Silverman rule can be used to automatically determine the bandwidth parameter h. For example, for electrical equipment with current amplitudes ranging from 5A to 20A, an appropriate bandwidth parameter might be between 0.1 and 0.5A, which can be adjusted based on the sample size and data dispersion.

[0104] In this embodiment, the kernel function contributions of all sample points are summed and normalized to obtain the overall probability density function. In actual calculations, a sufficiently dense grid of points can be selected within the range of the ordered amplitude sequence to calculate the probability density value at each grid point, forming a discrete probability density function representation. For example, 1000 points can be uniformly selected within the current amplitude range for estimation to obtain a high-precision probability density curve.

[0105] In this embodiment, the probability density function is integrated to obtain the cumulative distribution function. Integration can be performed using numerical integration methods such as the trapezoidal rule or Simpson's rule. For discrete representations of the probability density function, the estimated value of the cumulative distribution function at each grid point can be obtained by summing the probability density values ​​of adjacent grid points and multiplying them by the grid spacing. This method is simple, efficient, and suitable for most practical applications.

[0106] In this embodiment, quantile values ​​corresponding to multiple preset probability levels are calculated based on the cumulative distribution function. The preset probability levels include a high probability level representing the upper bound of the current signal and a low probability level representing the lower bound of the current signal. For example, probability levels such as 0.01, 0.05, 0.1, 0.9, 0.95, and 0.99 can be selected. For a given probability level p, the corresponding current amplitude whose cumulative distribution function value is equal to or closest to p is found; this is the p-quantile. If a more precise quantile value is required, linear interpolation can be performed between adjacent grid points.

[0107] In this embodiment, quantiles can effectively characterize the distribution characteristics and fluctuation range of current signals. For example, for a normally operating motor load, the 95th percentile might be 15A, indicating that the current typically does not exceed this value; the 5th percentile might be 8A, indicating that the current typically does not fall below this value. These quantile sets can be used for subsequent load anomaly detection and energy efficiency assessment.

[0108] In the above embodiments, the quantile calculation method based on probability density distribution is used in power systems. Compared with the traditional threshold setting method, it can more objectively reflect the statistical characteristics of current signals and reduce the influence of human factors. Furthermore, because it uses kernel density estimation, it does not make strong assumptions about the distribution of the original data, making it suitable for various complex load current patterns. The quantile set can be used as a load characteristic index to establish a statistical model for normal load operation, providing a basis for anomaly detection and predictive maintenance.

[0109] In practical applications, appropriate preset probability levels and kernel function types can be flexibly selected based on the characteristics of different types of loads to obtain the optimal expression of current signal characteristics. This method is applicable to the steady-state operation characteristic analysis of various power loads, providing an effective tool for the refined management and energy efficiency optimization of power systems.

[0110] Please see Figure 3 In some embodiments, based on the set of quantiles, a dynamic quantile tracking algorithm is used to update the high quantile threshold representing the upper bound of normal current fluctuations and the low quantile threshold representing the lower bound of normal current fluctuations in real time, which may include:

[0111] Construct a sliding time window for the amplitude of the current signal. Within the sliding time window, count the number of times the amplitude of the current signal exceeds the initial value of a preset high quantile threshold and the number of times it falls below the initial value of a preset low quantile threshold. Calculate the high-end default rate and the low-end default rate based on the number of times the amplitude exceeds the initial value of a preset high quantile threshold and the number of times it falls below the initial value of a preset low quantile threshold, respectively.

[0112] The high-end default rate is compared with a preset target high-end default rate to calculate the high-end default rate deviation. A high-end threshold correction amount is constructed based on the high-end default rate deviation. The high-end threshold correction amount is proportional to the high-end default rate deviation. When the high-end default rate is greater than the target high-end default rate, the high-end threshold correction amount is positive to increase the high percentile threshold. When the high-end default rate is less than the target high-end default rate, the high-end threshold correction amount is negative to decrease the high percentile threshold.

[0113] The low-end default rate is compared with a preset target low-end default rate to calculate the low-end default rate deviation. A low-end threshold correction amount is constructed based on the low-end default rate deviation. The low-end threshold correction amount is proportional to the low-end default rate deviation. When the low-end default rate is greater than the target low-end default rate, the low-end threshold correction amount is negative to reduce the low percentile threshold. When the low-end default rate is less than the target low-end default rate, the low-end threshold correction amount is positive to increase the low percentile threshold.

[0114] The high-end threshold correction is superimposed on the initial value of the high quantile threshold to obtain the updated high quantile threshold, and the low-end threshold correction is superimposed on the initial value of the low quantile threshold to obtain the updated low quantile threshold.

[0115] In this embodiment, historical data of the current signal is acquired, and statistical analysis is performed on the historical data to determine the set of quantiles of the current signal amplitude. This set contains multiple quantile values, each corresponding to a current amplitude distribution with different probabilities.

[0116] In this embodiment, based on the determined set of quantiles, a dynamic quantile tracking algorithm is used to update the high quantile threshold representing the upper bound of normal current fluctuations and the low quantile threshold representing the lower bound of normal current fluctuations in real time. This update process includes the following steps:

[0117] A sliding time window is constructed to measure the amplitude of the current signal. This window can be set to the current data of the most recent N sampling points. For example, if N=1000, it means that the most recent 1000 current sampling points constitute a time window. Within this sliding time window, the number of times the current signal amplitude exceeds a preset high quantile threshold (initial value) and the number of times it falls below a preset low quantile threshold (initial value) are counted. For example, if the initial high quantile threshold is set to 10A and the low quantile threshold is set to 2A, then the number of points with current values ​​greater than 10A within the window is counted as the number of times the amplitude exceeds the limit, and the number of points with current values ​​less than 2A is counted as the number of times the amplitude falls below the limit.

[0118] The high-end default rate and low-end default rate are calculated based on the number of times the limit was exceeded and the number of times the limit was exceeded, respectively. The high-end default rate is calculated by dividing the number of times the limit was exceeded by the total number of sampling points within the window; the low-end default rate is calculated by dividing the number of times the limit was exceeded by the total number of sampling points within the window. Taking the above example, if 50 points within the window exceed 10A, the high-end default rate is 50 / 1000 = 5%; if 30 points are below 2A, the low-end default rate is 30 / 1000 = 3%.

[0119] The high-end default rate is compared with a preset target high-end default rate to calculate the high-end default rate deviation. The target high-end default rate can be set according to actual application needs, for example, 2%. The high-end default rate deviation equals the actual high-end default rate minus the target high-end default rate. In the example above, the high-end default rate deviation is 5% - 2% = 3%.

[0120] A high-end threshold correction amount is constructed based on the deviation in the high-end default rate, and this correction amount is proportional to the deviation in the high-end default rate. A proportionality coefficient can be set. This makes the high-end threshold correction equal to Multiply by the high-end default rate bias, for example =0.5A / %. When the high-end default rate is greater than the target high-end default rate, the high-end threshold correction is positive, used to increase the high-percentile threshold; when the high-end default rate is less than the target high-end default rate, the high-end threshold correction is negative, used to decrease the high-percentile threshold. In the above example, the high-end threshold correction is 0.5A / % × 3% = 1.5A.

[0121] The low-end default rate is compared with a preset target low-end default rate to calculate the low-end default rate deviation. The target low-end default rate can also be set according to actual application needs, for example, 2%. The low-end default rate deviation equals the actual low-end default rate minus the target low-end default rate. In the example above, the low-end default rate deviation is 3% - 2% = 1%.

[0122] A low-end threshold correction amount is constructed based on the deviation of the low-end default rate; this correction amount is proportional to the deviation of the low-end default rate. A proportionality coefficient can be set. This makes the low-end threshold correction equal to Multiply by the low-end default rate bias, for example =0.3A / %. When the low-end default rate is greater than the target low-end default rate, the low-end threshold correction is negative, used to lower the low-quantile threshold; when the low-end default rate is less than the target low-end default rate, the low-end threshold correction is positive, used to raise the low-quantile threshold. In the example above, the low-end threshold correction is -0.3A / % × 1% = -0.3A (the negative sign indicates that the threshold needs to be lowered).

[0123] The high-end threshold correction is added to the initial value of the high-quantile threshold to obtain the updated high-quantile threshold. In the example above, the updated high-quantile threshold is 10A + 1.5A = 11.5A. The low-end threshold correction is added to the initial value of the low-quantile threshold to obtain the updated low-quantile threshold. In the example above, the updated low-quantile threshold is 2A - 0.3A = 1.7A.

[0124] In practical applications, a maximum step size for threshold updates can be set to avoid excessive threshold changes due to single data fluctuations. For example, the threshold change in each update can be limited to no more than 10% of the original threshold. Furthermore, an effective range for the threshold can be set to prevent updates to unreasonable intervals.

[0125] Through this dynamic adjustment mechanism, the high-quantile and low-quantile thresholds can adaptively adjust as the statistical characteristics of the current signal change, thus more accurately representing the normal fluctuation range of the current. When the actual current signal exceeds this dynamically updated range, it can be determined as a current anomaly, triggering corresponding alarms or protection measures in a timely manner.

[0126] In electric vehicle charging systems, this method can effectively identify abnormal current fluctuations during charging, detect potential faults in charging equipment in advance, and ensure charging safety. Since the current characteristics vary depending on the charging equipment and battery state, static thresholds are difficult to adapt to all situations. This method, however, dynamically adjusts the threshold to adapt to changes in current characteristics at different charging stages, improving the accuracy of anomaly detection.

[0127] In some embodiments, calculating the fault confidence level based on the excess amplitude and duration may include:

[0128] Obtain the magnitude of the current signal amplitude exceeding the high quantile threshold, calculate the ratio of the magnitude of the excess amplitude to the high quantile threshold, and obtain the normalized excess amplitude characterizing the degree of excess.

[0129] The duration for which the amplitude of the current signal exceeds the high quantile threshold is obtained. The ratio of the duration to the preset reference duration is nonlinearly transformed. The nonlinear transformation uses a logarithmic function mapping to obtain a normalized duration index characterizing the persistence.

[0130] The normalized exceedance magnitude and the normalized duration index are weighted and multiplied, with the weight coefficient of the normalized exceedance magnitude being greater than the weight coefficient of the normalized duration index, to obtain the fault confidence.

[0131] In this embodiment, the high quantile threshold is obtained through statistical analysis of the current signal under stable load operation, representing the upper limit of normal current fluctuations. When the amplitude of the real-time monitored current signal exceeds this threshold, it indicates a possible fault or abnormal event. When the amplitude exceeds the threshold, the current signal amplitude at the current moment is acquired by the data acquisition unit. This amplitude is obtained by analog-to-digital conversion from the current sensor. The high quantile threshold is subtracted from the current signal amplitude to obtain the excess amplitude value. The magnitude of the excess amplitude value directly reflects the degree to which the current signal deviates from the normal range; a larger value indicates a more severe abnormality.

[0132] In this embodiment, to eliminate the impact of differences in high quantile thresholds under different power distribution lines and operating conditions, it is necessary to normalize the excess amplitude value. The normalized excess amplitude is obtained by dividing the excess amplitude value by the high quantile threshold itself. The normalized excess amplitude is a dimensionless relative quantity, with its value starting from zero and theoretically having no upper limit. Normalization ensures the comparability of fault confidence calculated by different fault indicators.

[0133] In this embodiment, while recording the amplitude exceeding the threshold, it is also necessary to record the duration for which the current signal continuously exceeds the high quantile threshold. The duration is calculated using a timestamp mechanism. When the current signal amplitude first exceeds the high quantile threshold, the timestamp of that moment is recorded as the start time of the exceedance. In subsequent sampling periods, the current signal amplitude is continuously monitored, and as long as the current signal amplitude remains above the high quantile threshold, the timestamp of the current moment is continuously updated. When the current signal amplitude falls back below the high quantile threshold, the timestamp of that moment is recorded as the end time of the exceedance. The duration is the difference between the end time of the exceedance and the start time of the exceedance.

[0134] In this embodiment, directly using the duration ratio presents a problem: when the duration is very long, the ratio becomes very large, potentially distorting the calculation of fault confidence. To address this issue, a nonlinear transformation is applied to the duration ratio. A logarithmic function is used for mapping, with the duration ratio as the argument of the logarithmic function. The logarithmic function has the property of compressing large values ​​and amplifying small values, effectively balancing the contribution of different durations to fault confidence. Taking the natural logarithm of the duration ratio yields the logarithmically transformed value.

[0135] In this embodiment, after obtaining the normalized exceedance magnitude and normalized duration indices, they need to be fused to calculate the comprehensive fault confidence. The fusion process uses a weighted product approach. Weighted product can simultaneously consider information from both exceedance magnitude and duration dimensions, and when the value of either dimension is small, the product result will decrease accordingly, consistent with the logic of fault judgment. In the weighted product, weight coefficients need to be set for the normalized exceedance magnitude and normalized duration indices. The weight coefficients are set based on the importance of the two factors to fault judgment. The current exceedance magnitude directly reflects the severity of the fault and is a more critical criterion; therefore, the weight coefficient for the normalized exceedance magnitude should be larger. Although duration is also important, in some transient faults the duration may be short; therefore, the weight coefficient for the normalized duration index is set relatively smaller.

[0136] In some embodiments, the number of fault indicators is multiple, and the fault location method based on threshold adaptive calibration may further include:

[0137] For each fault indicator, a binary feature vector containing the fault confidence and the action time is constructed, and the spatial distance between any two fault indicators is calculated based on the installation location coordinates of the fault indicator.

[0138] Extract the action times of multiple fault indicators, pair the action times together, calculate the time difference between the action times of each pair of fault indicators, and associate the time difference with the corresponding spatial distance; based on the association between the time difference and the spatial distance, use the least squares fitting method to establish a linear relationship model between the time difference and the spatial distance; the slope of the linear relationship model characterizes the propagation speed of the fault current in the power distribution line.

[0139] In this embodiment, basic information of multiple fault indicators in the power distribution line is first obtained, including installation location coordinates, fault confidence level, and activation time. Fault confidence level reflects the degree of certainty the fault indicator has about the occurrence of a fault, and is typically calculated based on the degree of abnormality of electrical parameters such as current and voltage. Activation time records the specific point in time when the fault indicator detects the fault and triggers its action.

[0140] In this embodiment, the binary feature vector reflects the intensity and temporal characteristics of the fault information detected by the fault indicator, serving as the basis for subsequent analysis. For example, for fault indicator A, its feature vector can be represented as a vector containing two elements: fault confidence CA and action time TA.

[0141] In this embodiment, calculating the spatial distance between any two fault indicators based on their installation coordinates can be achieved by setting the position coordinates of fault indicator i to be in three-dimensional space. , , The coordinates of the fault indicator j are: , , The spatial distance between the two points can be obtained by taking the square root of the sum of the squares of the coordinate differences between the two points. This calculation takes into account the actual physical distribution of the fault indicator on the power distribution line.

[0142] In this embodiment, the action times of multiple fault indicators are extracted, and the action times are paired up to calculate the time difference between the action times of each pair of fault indicators. For fault indicators i and j, the time difference is calculated as follows: minus The results, among which and These represent the activation times of fault indicators i and j, respectively. The time difference calculation must consider time synchronization accuracy to ensure clock synchronization among all fault indicators, which can be achieved using GPS clocks or network time protocols.

[0143] In this embodiment, the calculated time difference is correlated with the corresponding spatial distance. A spatial distance matrix D and a time difference matrix T are established, where the elements in matrix D... The elements in matrix T represent the spatial distance between fault indicators i and j. This represents the time difference between the actions of fault indicators i and j. These two matrices contain information about the spatial and temporal characteristics of fault propagation.

[0144] In this embodiment, based on the correlation between time difference and spatial distance, a linear relationship model between time difference and spatial distance is established using the least squares fitting method. This linear relationship can be described as: time difference equals spatial distance divided by propagation speed plus an error term. The optimal propagation speed value is obtained by minimizing the sum of squared errors of all fault indicator pairs. Specifically, the calculation process involves summing the squared differences between the actual time difference and the theoretical time difference (spatial distance divided by propagation speed) of each fault indicator pair, differentiating the propagation speed with respect to the propagation speed, and setting the derivative to zero to obtain the optimal estimate of the propagation speed.

[0145] In this embodiment, by utilizing the established linear relationship model and the obtained propagation velocity, combined with the action time and spatial location of the fault indicator, the location of the fault point can be deduced. Assuming the fault occurs at... The activation time of fault indicator i is Then the distance from the fault point to the fault indicator i can be estimated as the propagation speed multiplied by the time difference ( reduce By establishing multiple such distance equations, an overdetermined system of equations is formed, and the coordinates of the fault point are solved using the least squares method.

[0146] In practical applications, fault indicator data, including fault confidence, activation time, and location information, can be obtained through data acquisition terminals in power distribution lines. Data preprocessing requires eliminating outliers and time synchronization errors, which can be achieved using median filtering and time calibration algorithms. During the fitting of the linear relationship model, weighted least squares can be used, determining the weight coefficients based on the fault confidence value to improve fitting accuracy.

[0147] In this embodiment, based on the established linear relationship model, the status of the power distribution line can be dynamically evaluated, and the fault location strategy can be adjusted in real time. When a change in line parameters is detected that causes a change in propagation speed, such as a change in line impedance exceeding a preset threshold, the linear model parameters are automatically recalculated to maintain the accuracy of fault location. The model parameter update can use a sliding window method, comprehensively considering the weights of historical data and the latest data.

[0148] In the above embodiments, by fusing the confidence and time information of multiple fault indicators, a spatiotemporal model of fault propagation is established, which enables precise location of the fault point. The location error can be controlled within a percentage range of the total line length, improving the efficiency and accuracy of fault location and reducing power outage time and maintenance costs.

[0149] In some embodiments, determining the propagation direction and attenuation law of the fault current based on the spatial distribution gradient of the fault confidence, and determining the fault occurrence section in the distribution line based on the propagation direction and the attenuation law, may include:

[0150] Based on the propagation speed, the fault indicator with the earliest action time is selected as the initial reference point. Starting from the initial reference point, along the negative gradient direction of the gradient vector field of the fault confidence, and combining the action time of other fault indicators with the propagation speed, the theoretical propagation time from the initial reference point to the position of each fault indicator is calculated. The time difference between the theoretical propagation time and the actual action time is compared to construct a time deviation index that characterizes the consistency between theoretical propagation and actual action.

[0151] The fault confidence values ​​of each fault indicator are weighted and corrected according to the time deviation index, and the weighted and corrected comprehensive fault confidence is calculated.

[0152] The comprehensive fault confidence is spatially mapped on the distribution line topology to identify the peak region where the comprehensive fault confidence occurs. Within the peak region, the reverse propagation path of the fault current is determined according to the negative gradient direction of the gradient vector field of the fault confidence.

[0153] Along the reverse propagation path, combining the propagation speed with the action time of each fault indicator, trace back to the position where the action time is extrapolated to zero, determine the position as the fault source point, and determine the power distribution line section where the fault source point is located as the fault occurrence section.

[0154] In this embodiment, the activation time and fault confidence value of all fault indicators on the distribution line are obtained. The fault confidence value can be calculated from characteristic quantities such as the current amplitude, phase angle change rate, and harmonic content measured by the fault indicator. The value range is usually from 0 to 1, and the larger the value, the higher the probability of a fault occurring at that location. For example, if the short-circuit current amplitude measured by a fault indicator is 5 times the rated current, its fault confidence value can be set to 0.85.

[0155] In this embodiment, the propagation speed of the fault current in the line is determined by analyzing the sequence of actions of each fault indicator. In overhead lines, the propagation speed of the fault current is close to 0.9 times the speed of light, approximately 2.7 × 10^8 meters per second; in cable lines, the propagation speed is approximately 0.6 times the speed of light, approximately 1.8 × 10^8 meters per second. In mixed lines, the average propagation speed can be calculated based on the ratio of overhead lines to cables.

[0156] In this embodiment, the fault indicator with the earliest activation time is selected as the initial reference point. For example, if there are 5 fault indicators sequentially... =0ms、 =0.15ms =0.25ms =0.35ms If the action occurs at 0.5ms, then select... The indicator for the moment of action serves as the initial reference point.

[0157] In this embodiment, for fault indicator pairs whose spatial distance is less than a preset neighborhood threshold, the ratio of the difference in fault confidence between the two fault indicators to the spatial distance is calculated to obtain a confidence gradient scalar characterizing the spatial rate of change of fault confidence. Specifically, the calculation method is as follows: take the fault confidence values ​​of fault indicators i and j... and The difference, divided by the spatial distance between the two. The confidence gradient scalar is obtained. The preset neighborhood threshold takes into account the topology of the power distribution line and the propagation characteristics of the fault current, and can usually be set as a multiple of the average span of the line.

[0158] A gradient vector field representing the spatial distribution gradient of fault confidence is constructed by combining the confidence gradient scalar with the spatial direction vector connecting two fault indicators. The spatial direction vector is a unit vector pointing from fault indicator i to fault indicator j, which can be calculated by dividing the coordinate difference between the two points by the spatial distance. Multiplying this direction vector by the previously calculated confidence gradient scalar yields a directional gradient vector indicating the direction in which fault confidence increases in space. Then, starting from the initial reference point, analysis is performed along the negative gradient direction of the fault confidence gradient vector field. The fault confidence gradient vector field represents the rate of change and direction of fault confidence in space; the negative gradient direction points to the direction of the fastest increase in fault confidence, typically pointing to the fault source point.

[0159] In this embodiment, the theoretical propagation time from the initial reference point to the location of each fault indicator can be calculated based on the physical location of each indicator and the known propagation speed of the fault current. The theoretical time required for the fault current to propagate from the initial reference point to each indicator can be calculated. For example, if the second indicator is 40 kilometers from the initial reference point and the current propagation speed is 2.7 × 10^8 meters / second, then the theoretical propagation time is 0.148 milliseconds.

[0160] In this embodiment, the time difference between the theoretical propagation time and the actual action time is compared to construct a time deviation index. The time deviation index can be expressed as: ,in For the actual action time of the i-th indicator, The action time at the initial reference point, This represents the theoretical propagation time. The smaller the time deviation, the more consistent the indicator's operation is with the fault current propagation pattern.

[0161] In this embodiment, the fault confidence of each fault indicator is weighted and corrected based on the time deviation index. An exponential weighting method can be used. ,in The original fault confidence level. This is the weighted and adjusted confidence level. This is the weighting coefficient, typically ranging from 5 to 10. This is a time deviation indicator.

[0162] In this embodiment, the weighted and corrected overall fault confidence can be calculated by using a distance-inverse weighted average method to calculate the overall fault confidence of each node on the line, based on the corrected confidence of surrounding fault indicators. ,in Let x be the distance from node x to the i-th fault indicator.

[0163] In this embodiment, the overall fault confidence is spatially mapped onto the distribution line topology to form a confidence distribution curve. The confidence level can be represented by the color intensity or height, visually showing the area where a fault may occur. For example, on a 10 kV distribution line, the overall fault confidence at different locations is represented by a color gradient from blue (low confidence) to red (high confidence).

[0164] In this embodiment, the peak region where the overall fault confidence level shows a peak is typically the location where the fault is most likely to occur. Within the peak region, the reverse propagation path of the fault current is determined based on the negative gradient direction of the gradient vector field of the fault confidence level. The negative gradient direction can be calculated from the fault confidence level difference between adjacent nodes.

[0165] In this embodiment, the propagation path is reversed, and the propagation speed and the action time of each fault indicator are combined to trace back to the position where the extrapolated action time is zero. Specifically, starting from the peak point, the process proceeds step by step along the negative gradient direction. For each point, the theoretical action time corresponding to that point is extrapolated based on its distance from each fault indicator and the action time of the indicator. When the extrapolated theoretical action time is close to zero, that point is considered to be the fault source point.

[0166] In this embodiment, the section of the power distribution line where the fault source is located is obtained, and this section is determined as the fault occurrence section, thereby completing the fault location. For example, if the calculated fault source is located between towers 12 and 13 of the power distribution line, this section is determined as the fault occurrence section, providing accurate location information for subsequent maintenance work.

[0167] By using the above methods, the spatial distribution and time series information of fault confidence can be comprehensively utilized to achieve accurate location of faulty sections of power distribution lines, improve fault handling efficiency, reduce power outage time, and improve power supply reliability.

[0168] Please see Figure 4 One embodiment of the present invention provides a fault location device based on threshold adaptive correction. The fault location device based on threshold adaptive correction may include: a state recognition module, a quantile calculation module, a confidence calculation module, an action triggering module, and a fault location module.

[0169] The status identification module is used to acquire the current signal sequence collected by the fault indicator in the power distribution line, calculate the harmonic pollution index based on the harmonic components and fundamental components of the current signal sequence, and identify the load stable operation status of the line monitored by the fault indicator in combination with the stability criteria.

[0170] The quantile calculation module is used to extract the probability density distribution curve of the current signal for the time period corresponding to the stable operation state of the load, and calculate the set of quantiles of the probability density distribution curve.

[0171] The confidence calculation module is used to update the high quantile threshold representing the upper bound of normal current fluctuation and the low quantile threshold representing the lower bound of normal current fluctuation in real time based on the quantile set and the dynamic quantile tracking algorithm. When the current signal exceeds the high quantile threshold or falls below the low quantile threshold, anomaly detection is triggered, and the fault confidence is calculated based on the magnitude and duration of the exceedance.

[0172] The action triggering module is used to trigger the fault indicator and record the location and time of the fault in the power distribution line when the fault confidence exceeds the preset confidence threshold.

[0173] The fault location module is used to perform weighted fusion of the fault confidence level and the action time of the fault indicator, determine the propagation direction and attenuation law of the fault current based on the spatial distribution gradient of the fault confidence level, and determine the fault occurrence section in the power distribution line based on the propagation direction and the attenuation law.

[0174] The specific functions and effects of the fault location device based on threshold adaptive correction can be explained by referring to other embodiments in this specification, and will not be repeated here. Each module in the fault location device based on threshold adaptive correction can be implemented entirely or partially through software, hardware, or a combination thereof. Each module can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each module.

[0175] Please see Figure 5 One embodiment of the present invention can provide an electronic device, the electronic device comprising:

[0176] A memory, and one or more processors communicatively connected to the memory;

[0177] The memory stores instructions that can be executed by the one or more processors, which, when executed by the one or more processors, enable the one or more processors to implement the fault location method based on threshold adaptive correction as described in any of the above embodiments.

[0178] One embodiment of the present invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the fault location method based on threshold adaptive correction as described in any of the above embodiments.

[0179] This specification also provides a computer program product containing instructions that, when executed by a computer, cause the computer to perform the fault location method based on threshold adaptive correction as described in any of the above embodiments.

[0180] It is understood that the specific examples in this document are only intended to help those skilled in the art better understand the embodiments described herein, and are not intended to limit the scope of the invention.

[0181] It is understood that in the various embodiments described in this specification, the sequence number of each process does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments described in this specification.

[0182] It is understood that the various implementation methods described in this specification can be implemented individually or in combination, and the implementation methods in this specification are not limited in this respect.

[0183] Unless otherwise stated, all technical and scientific terms used in the embodiments of this specification have the same meaning as commonly understood by one of ordinary skill in the art. The terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the scope of this specification. The term "and / or" as used in this specification includes any and all combinations of one or more of the associated listed items. The singular forms "a," "the," and "the" as used in the embodiments of this specification and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise.

[0184] It is understood that the processor in this invention can be an integrated circuit chip with signal processing capabilities. In implementation, each step of the above method implementation can be completed by the integrated logic circuitry in the processor's hardware or by instructions in software form. The processor can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this specification. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this specification can be directly embodied in the execution of a hardware decoding processor, or executed by a combination of hardware and software modules in the decoding processor. The software modules can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. This storage medium is located in memory; the processor reads information from the memory and, in conjunction with its hardware, completes the steps of the above method.

[0185] It is understood that the memory in this invention can be volatile memory or non-volatile memory, or may include both. Specifically, the non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory can be random access memory (RAM). It should be noted that the memory in the systems and methods described herein is intended to include, but is not limited to, these and any other suitable types of memory.

[0186] The user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in the embodiments of the present invention are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, use and processing of the relevant data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and corresponding operation entry points are provided for users to choose to authorize or refuse.

[0187] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this specification.

[0188] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the aforementioned method implementations, and will not be repeated here.

[0189] In the several embodiments provided in this specification, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0190] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0191] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0192] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solutions of this specification, in essence, or the parts that contribute to the prior art, or parts of the technical solutions, can be embodied in the form of software products. These computer software products are stored in a storage medium and include several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this specification. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0193] The above description is merely a specific embodiment of this specification, but the scope of protection of this invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this specification should be included within the scope of protection of this specification. Therefore, the scope of protection of this invention should be determined by the scope of the claims.

Claims

1. A fault location method based on threshold adaptive calibration, characterized in that, include: The current signal sequence collected by the fault indicator in the power distribution line is obtained, the harmonic pollution index is calculated based on the harmonic components and fundamental components of the current signal sequence, and the load stable operation status of the line monitored by the fault indicator is identified in combination with the stability criteria. For the time period corresponding to the stable operation state of the load, the probability density distribution curve of the current signal is extracted, and the set of quantiles of the probability density distribution curve is calculated. Based on the set of quantiles, a dynamic quantile tracking algorithm is used to update the high quantile threshold representing the upper bound of normal current fluctuation and the low quantile threshold representing the lower bound of normal current fluctuation in real time. When the current signal exceeds the high quantile threshold or falls below the low quantile threshold, anomaly detection is triggered, and the fault confidence is calculated based on the magnitude and duration of the exceedance. When the fault confidence level exceeds the preset confidence threshold, the fault indicator is triggered and the location and time of the fault in the power distribution line are recorded. The fault confidence level of the fault indicator is weighted and fused with the action time. Based on the spatial distribution gradient of the fault confidence level, the propagation direction and attenuation law of the fault current are inferred and determined. Based on the propagation direction and attenuation law, the fault occurrence section in the power distribution line is determined.

2. The method according to claim 1, characterized in that, Acquire the current signal sequence collected by the fault indicator in the power distribution line, calculate the harmonic pollution index based on the harmonic and fundamental components of the current signal sequence, and identify the load stability operation status of the line monitored by the fault indicator by combining the stability criteria, including: The current signal sequence is subjected to frequency domain transformation to extract the amplitude and phase of the fundamental component and each harmonic component, and the energy ratio of the harmonic component to the fundamental component is calculated. The energy of each harmonic in the harmonic component is weighted and summed with the energy of the fundamental component. The harmonic pollution index is calculated based on the ratio of the weighted sum to the energy of the fundamental component. A time series of the harmonic pollution index is constructed in chronological order, and the difference between the harmonic pollution indexes at adjacent moments in the time series is calculated to obtain a sequence of the rate of change of the harmonic pollution index. A sliding window statistical analysis was performed on the harmonic pollution index change rate sequence to calculate the variance of the harmonic pollution index change rate within the sliding window; The variance is compared with a preset stability criterion. When the variance is less than the stability criterion, the time period corresponding to the sliding window is determined to be the load stable operation state. When the variance is greater than or equal to the stability criterion, the time period corresponding to the sliding window is determined to be the load switching state.

3. The method according to claim 1, characterized in that, For the time period corresponding to the stable operation state of the load, the probability density distribution curve of the current signal is extracted, and the set of quantiles of the probability density distribution curve is calculated, including: The amplitude samples of the current signal are extracted from the time period corresponding to the stable operation state of the load, and the amplitude samples are numerically sorted to obtain an ordered amplitude sequence. Based on the ordered amplitude sequence, a probability density distribution curve of the current signal is constructed using a kernel density estimation method. The kernel density estimation method obtains a continuous probability density function by applying a kernel function to each amplitude sample in the ordered amplitude sequence and superimposing them; the cumulative distribution function is obtained by integrating the probability density function. Based on the cumulative distribution function, the quantile values ​​corresponding to multiple preset probability levels are calculated. The multiple preset probability levels include a high probability level representing the upper bound of the current signal and a low probability level representing the lower bound of the current signal. The quantile values ​​corresponding to the multiple preset probability levels are combined to form the quantile set.

4. The method according to claim 1, characterized in that, Based on the aforementioned quantile set, a dynamic quantile tracking algorithm is used to update the high quantile threshold representing the upper bound of normal current fluctuations and the low quantile threshold representing the lower bound of normal current fluctuations in real time, including: Construct a sliding time window for the amplitude of the current signal. Within the sliding time window, count the number of times the amplitude of the current signal exceeds the initial value of a preset high quantile threshold and the number of times it falls below the initial value of a preset low quantile threshold. Calculate the high-end default rate and the low-end default rate based on the number of times the amplitude exceeds the initial value of a preset high quantile threshold and the number of times it falls below the initial value of a preset low quantile threshold, respectively. The high-end default rate is compared with a preset target high-end default rate to calculate the high-end default rate deviation. A high-end threshold correction amount is constructed based on the high-end default rate deviation. The high-end threshold correction amount is proportional to the high-end default rate deviation. When the high-end default rate is greater than the target high-end default rate, the high-end threshold correction amount is positive to increase the high percentile threshold. When the high-end default rate is less than the target high-end default rate, the high-end threshold correction amount is negative to decrease the high percentile threshold. The low-end default rate is compared with a preset target low-end default rate to calculate the low-end default rate deviation. A low-end threshold correction amount is constructed based on the low-end default rate deviation. The low-end threshold correction amount is proportional to the low-end default rate deviation. When the low-end default rate is greater than the target low-end default rate, the low-end threshold correction amount is negative to reduce the low percentile threshold. When the low-end default rate is less than the target low-end default rate, the low-end threshold correction amount is positive to increase the low percentile threshold. The high-end threshold correction is superimposed on the initial value of the high quantile threshold to obtain the updated high quantile threshold, and the low-end threshold correction is superimposed on the initial value of the low quantile threshold to obtain the updated low quantile threshold.

5. The method according to claim 1, characterized in that, The fault confidence level is calculated based on the magnitude and duration of the exceedance, including: Obtain the magnitude of the current signal amplitude exceeding the high quantile threshold, calculate the ratio of the magnitude of the excess amplitude to the high quantile threshold, and obtain the normalized excess amplitude characterizing the degree of excess. The duration for which the amplitude of the current signal exceeds the high quantile threshold is obtained. The ratio of the duration to the preset reference duration is nonlinearly transformed. The nonlinear transformation uses a logarithmic function mapping to obtain a normalized duration index characterizing the persistence. The normalized exceedance magnitude and the normalized duration index are weighted and multiplied, with the weight coefficient of the normalized exceedance magnitude being greater than the weight coefficient of the normalized duration index, to obtain the fault confidence.

6. The method according to claim 1, characterized in that, The number of fault indicators is multiple, and the method further includes: For each fault indicator, a binary feature vector containing the fault confidence and the action time is constructed, and the spatial distance between any two fault indicators is calculated based on the installation location coordinates of the fault indicator. Extract the action times of multiple fault indicators, pair the action times together, calculate the time difference between the action times of each pair of fault indicators, and associate the time difference with the corresponding spatial distance; based on the association between the time difference and the spatial distance, use the least squares fitting method to establish a linear relationship model between the time difference and the spatial distance; the slope of the linear relationship model characterizes the propagation speed of the fault current in the power distribution line.

7. The method according to claim 6, characterized in that, Based on the spatial distribution gradient of the fault confidence level, the propagation direction and attenuation law of the fault current are determined, and based on the propagation direction and attenuation law, the fault occurrence section in the power distribution line is determined, including: Based on the propagation speed, the fault indicator with the earliest action time is selected as the initial reference point. Starting from the initial reference point, along the negative gradient direction of the gradient vector field of the fault confidence, and combining the action time of other fault indicators with the propagation speed, the theoretical propagation time from the initial reference point to the position of each fault indicator is calculated. The time difference between the theoretical propagation time and the actual action time is compared to construct a time deviation index that characterizes the consistency between theoretical propagation and actual action. The fault confidence values ​​of each fault indicator are weighted and corrected according to the time deviation index, and the weighted and corrected comprehensive fault confidence is calculated. The comprehensive fault confidence is spatially mapped on the distribution line topology to identify the peak region where the comprehensive fault confidence occurs. Within the peak region, the reverse propagation path of the fault current is determined according to the negative gradient direction of the gradient vector field of the fault confidence. Along the reverse propagation path, combining the propagation speed with the action time of each fault indicator, trace back to the position where the action time is extrapolated to zero, determine the position as the fault source point, and determine the power distribution line section where the fault source point is located as the fault occurrence section.

8. A fault location device based on threshold adaptive correction, characterized in that, The fault location device based on threshold adaptive correction includes: The status identification module is used to acquire the current signal sequence collected by the fault indicator in the power distribution line, calculate the harmonic pollution index based on the harmonic components and fundamental components of the current signal sequence, and identify the load stable operation status of the line monitored by the fault indicator in combination with the stability criteria. The quantile calculation module is used to extract the probability density distribution curve of the current signal for the time period corresponding to the stable operation state of the load, and calculate the set of quantiles of the probability density distribution curve. The confidence calculation module is used to update the high quantile threshold representing the upper bound of normal current fluctuation and the low quantile threshold representing the lower bound of normal current fluctuation in real time based on the quantile set and the dynamic quantile tracking algorithm. When the current signal exceeds the high quantile threshold or falls below the low quantile threshold, anomaly detection is triggered, and the fault confidence is calculated based on the magnitude and duration of the exceedance. The action triggering module is used to trigger the fault indicator and record the location and time of the fault in the power distribution line when the fault confidence exceeds the preset confidence threshold. The fault location module is used to perform weighted fusion of the fault confidence level and the action time of the fault indicator, determine the propagation direction and attenuation law of the fault current based on the spatial distribution gradient of the fault confidence level, and determine the fault occurrence section in the power distribution line based on the propagation direction and the attenuation law.

9. An electronic device, characterized in that, It includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the fault location method based on threshold adaptive calibration as described in any one of claims 1 to 7.

10. A computer storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the fault location method based on threshold adaptive calibration as described in any one of claims 1 to 7.

Images