Method and device for high impedance fault event detection

The method analyzes harmonic energy data using a leaky integrator to detect high impedance faults in power systems, addressing detection challenges and improving fault identification and system reliability.

WO2024240682A1PCT designated stage expired Publication Date: 2024-11-28EATON INTELLIGENT POWER LTD

Patent Information

Application Number
PCT/EP2024/063777
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Priority Date
2023-05-19
Filing Date
2024-05-17
Publication Date
2024-11-28

AI Technical Summary

Technical Problem

High impedance faults in power distribution systems are challenging to detect due to low conductor current magnitude below traditional overcurrent relay thresholds and difficulty in differentiation from non-fault events, posing risks such as the Dixie Fire incident.

Method used

A method using a processor to analyze time series data of harmonic energy, determining indicators of random components, and employing a leaky integrator to identify high impedance faults by comparing these indicators to thresholds, thereby generating a signal indicative of a fault event.

Benefits of technology

Effectively detects high impedance faults by distinguishing them from other events, reducing false alarms and ensuring timely isolation of affected power lines, thus enhancing system reliability and safety.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure EP2024063777_28112024_PF_FP_ABST
    Figure EP2024063777_28112024_PF_FP_ABST
Patent Text Reader

Abstract

A method for high impedance fault detection in a power system is described. It comprises: obtaining (S1000) time series data representative of harmonic energy of one or more electrical signals in the system; determining (S1001) values over time of an indicator of presence of random components in the harmonic energy; comparing (S1002) each indicator value or values to a first threshold (Tc, Ts, Tr) to determine the presence of random components integrating (S1003) a result of the comparison over time using a leaky integrator (1204); determining (S1004) whether an output of the integrator is above a second threshold (Te, Trms) for a duration greater than a third threshold (Di, Dr) and if the determination is positive, generating a signal (1206, 1505) indicating a high impedance fault event. A device implementing the method and a computer-implemented medium with code for carrying out the method are also described.
Need to check novelty before this filing date? Find Prior Art

Description

Title of the InventionMETHOD AND DEVICE FOR HIGH IMPEDANCE FAULT EVENT DETECTIONTechnical Field

[0001] The field of the disclosure relates generally to a method and a device for identification of high impedance fault events in electrical systems such as power distribution systems.Background Art

[0002] Power distribution systems can be monitored for the occurrence of potentially abnormal changes within the system, using appropriate monitoring devices. Monitoring may be performed based on changes in system characteristics. Events may signal a degradation, a fault, a cyber-intrusion or an impending failure, possibly with catastrophic consequences. However, there is a need to distinguish among detected events through appropriate classification. In particular, identification of a high impedance fault has long been a challenge in maintaining the reliability of power distribution systems. High impedance faults typically occur when a downed powerline conductor falls on a high impedance surface, such as grass, sand, asphalt or gravel, or when a tree limb falls on the conductor. The Dixie Fire of 2021, second largest in California history, was started when a tree fell on a power line. High impedance faults hence need to be distinguished from other types of abnormal events. When a high impedance fault occurs, the conductor current magnitude may be lower than the pickup threshold of a traditional overcurrent relay. Therefore, it cannot be detected by an over current relay. Moreover, differentiation from non-fault i.e. normal, events (such as e.g. capacitor bank switches, branch openings, induction motor soft starts...) is challenging. Identifying high impedance faults is crucial to enable elimination of the dangerous situation generated by the fault, e.g. by isolating the power line on which the fault occurred from an active power source.Summary of the Invention

[0003] A first aspect concerns a method for high impedance fault event detection in a power system, said method being carried out by a device comprising aprocessor, said method comprising: o obtaining time series data representative of harmonic energy of one or more electrical signals in the system; o determining values over time of an indicator of presence of random components in the harmonic energy; o comparing each indicator value or values to a first threshold to determine the presence of random components; o integrating a result of the comparison over time using a leaky integrator; o determining whether an output of the integrator is above a second threshold for a duration greater than a third threshold and if the determination is positive, generating a signal indicative of a high impedance fault event.

[0004] According to one or more embodiments, further comprising determining the presence of an anomaly event as a function of the time series data comprising: determining instantaneous harmonic energy changes (AE) as a function of the time series data; determining whether a difference between an instance of the instantaneous harmonic energy changes and a moving average of the instantaneous harmonic energy changes is greater than a fourth threshold, wherein the fourth threshold is function of a mean of said difference; as a result of the determination, providing an input to a leaky integrator configured to count data points for which the difference is greater than the fourth threshold; determining whether an output of said integrator is greater than a fifth threshold and in the affirmative, determining that an anomaly event is present.

[0005] The instantaneous harmonic energy changes are also referred to as ‘delta energy’.

[0006] According to one or more embodiments, the indicator of presence of random components in the harmonic energy is a correlation coefficient resulting from a correlation of the time series data over two adjacent time windows of same lengthsubstantially equal to an integer multiple of a fundamental period of the one or more electrical signals.

[0007] According to one or more embodiments, the two adj acent time windows are shifted in time by a shifting step for the determination of successive values of the indicator.

[0008] According to one or more embodiments, the comparison of the indicator with the first threshold indicates the presence of random components if the indicator is below the first threshold.

[0009] According to one or more embodiments, the indicator of presence of random components in the harmonic energy is function of a scaled root of the sum of the energy of harmonic components.

[0010] According to one or more embodiments, to minimize the influence of background noise in comparison of random components, a statistic function named t- score was implemented using student t-test. The similarity of the distributions of harmonic energy within two consecutive windows are measured. The threshold is optional based upon the customer selections: confidence level includes 95%, 98% and 99%.

[0011] The scaled root of the sum of the energy of harmonic components, which is equivalent to the RMS of the selected harmonic components when the scaling factor is 0.5.

[0012] According to one or more embodiments, the indicator is equal to a difference between A, respectively A / B and a moving average of A, respectively A / B, with A equal to Al or A2, andwhere X^is the scaled root of the sum of the energy of at least one harmonic calculated for time series data in a first time window shifted among N positions within a second time window, with k = 1 to N and N>1.

[0013] According to one or more embodiments, the comparison of the indicator with the first threshold indicates the presence of random components if the indicator is greater than the first threshold.

[0014] According to one or more embodiments, the first threshold is function of the variance of past values of A, respectively A / B.

[0015] According to one or more embodiments, the method comprises generating a control signal of the power system as a function of the signal indicative of a high impedance fault event.

[0016] Another aspect concerns a device for high impedance fault detection in a power system, said device comprising a processor and memory, said memory comprising software code, the processor causing the device to perform one of the methods described herein.

[0017] Another aspect concerns a computer-readable storage medium comprising code which when executed by a processor of a device causes the device to perform one of the methods described herein.Brief Description of Drawings

[0018] Non-limiting and non-exhaustive embodiments are described with reference to the following figures, wherein like reference numerals refer to like parts throughout the various views unless otherwise specified.

[0019] Fig. l is a schematic diagram of a system overview according to one or more exemplary embodiments.

[0020] Fig. 2 is a flow chart of a method for anomaly event detection according to one or more exemplary embodiments.

[0021] Fig. 3 is a block diagram illustrating an exemplary embodiment of a first step of the method of figure 2.

[0022] Fig. 4 is a block diagram illustrating an exemplary embodiment of a second step of the method of figure 2.

[0023] Fig. 5 is a flow chart of detailed steps of an exemplary method for anomaly event detection such as the method of Figure 1.

[0024] Fig. 6a is a graph representing time series data points and illustrating the abnormal event detection process in a non-limiting manner for the detection of a so-called high-impedance fault in a power line.

[0025] Fig. 6b is a graph representing an output of an integrator used in the abnormal event detection based on data points shown in Fig. 6b.

[0026] Fig. 7 is a block diagram of an abnormal event detection device according to one or more exemplary embodiments.

[0027] Fig. 8 is a flow chart of a method for event classification according to one or more exemplary embodiments.

[0028] Fig. 9 is a flowchart of a non-limiting embodiment of a first method for high impedance fault detection.

[0029] Fig. 10 is a more detailed flowchart of the method of Fig. 8.

[0030] Fig. I l a graph schematically illustrating an example of temporal windows which can be used for a correlation calculation in the method of Fig. 9.

[0031] Fig. 12 is a diagram illustrating the various functional blocks implementing the decision logic of the first method according to an exemplary embodiment.

[0032] Fig. 13 is a graph schematically illustrating an example of time windows which can be used in a non-limiting embodiment of a second method for high impedance fault detection.

[0033] Fig. 14 is a flowchart of the second method according to a non-limiting embodiment.

[0034] Fig. 15 is a block diagram of the functional blocks of a device implementing part of the second method according to a non-limiting embodiment.

[0035] Fig. 16 is a block diagram of a decision logic of Fig. 15 according to a non-limiting embodiment.

[0036] Fig. 17a and Fig. 17b are respective exemplary graphs illustrating signals and outputs when applying the first embodiment respectively to a high impedance fault event situation and to a normal event situation.Description of Embodiments

[0037] Various exemplary embodiments will now be described more fully withreference to the accompanying drawings. However, specific structural and functional details disclosed herein are merely representative for purposes of describing the example embodiments. The exemplary embodiments may be embodied in many alternate forms and should not be construed as limited to only the embodiments set forth herein. It should be understood that there is no intent to limit example embodiments to the particular forms disclosed.

[0038] It should be appreciated by those skilled in the art that any functions, engines, block diagrams, flow diagrams, state transition diagrams and / or flowcharts herein represent conceptual views of illustrative circuitry embodying the principles of the exemplary embodiments. Functions, actions or steps described herein can be implemented in hardware or software or any combination thereof. If implemented in software, the functions, blocks of the block diagrams and / or flowchart illustrations can be implemented e.g. using software code executed by a processor or a processing device.

[0039] Fig. 1 is a schematic diagram of an exemplary architecture of an electrical system with associated abnormal event detection and control functions. The system comprises an event detection and classification device 101. In the frame of the present description, a power grid is used as a non-limiting example of an electrical system for which device 101 provides an event detection function. In Fig. 1, the power grid is represented by power line 102. Sensor 103 provides measurements of at least one time-varying signal representative of the electrical system to be monitored. A control circuit 104 allows device 101 to control the electrical system. For example, control circuit 104 may reduce or cut power on the power line 102 under control of device 101. In the frame of a power grid application, processing device 101 may be an Intelligent Electronic Device (TED’), i.e. a processor-based device which controls power grid functions, such as reclosers, circuit breakers, transformers and capacitor banks. Depending on implementation, sensor 103 and control circuit 104 may be integral part of device 101 or separate modules.

[0040] Sensor 103 provides time series data of e.g. currents, voltages or energies in the power grid 102, which are functioning variables of the electrical system. According to one embodiment, for identification of high impedance faults, time series data representative of harmonic components of these representative signalsare used. In particular, the energy magnitude of a set of selected harmonics can be used. These may be determined e.g. through appropriate processing by device 101 of the signals provided by sensor 103. For other types of events, other data, may be used.

[0041] Device 101 carries out an event detector function 105 and an event classifier function 106. Event detector 105 and event classifier 106 are shown in Fig. 1 as distinct boxes for illustrative purposes only - in an actual implementation, the functions may be carried out by the same or overlapping or disjoint hardware, as needed.

[0042] An event detected by event detector 105 may be processed either by classifier 106 for feature extraction and fault classification or sent to a server or cloud processing platform 107 for that purpose. The latter solution allows to reduce the complexity of device 101. As a function of fault classification results determined by device 101 itself or based on return information from a remote device, device 101 applies appropriate control measures to power grid 102 through control circuit 104. For that purpose, a control signal 108 can be generated.In what follows, two embodiments for high impedance fault event detection will be described. The first embodiment uses instantaneous harmonic energy changes (also referred to as ‘delta energy’) for event detection and further uses the total harmonic energy for event classification, whereas the second embodiment is based on the analysis of a scaled root of the sum of the energy of harmonic components, which is equivalent to the RMS of the selected harmonic components when the scaling factor is 0.5.First EmbodimentEvent detection

[0043] Fig. 2 is a flowchart of a method 200 for anomaly event detection according to the first embodiment. An anomaly event comprises normal system operation events and abnormal events. Examples of normal events are transformer inrush and capacitor bank switching. An example of an abnormal event is a highimpedance fault event. Both normal events and abnormal events are distinctly different from normal baseline operation. Once an anomaly event has been detected, further processing may be applied to distinguish high impedance fault events.

[0044] The method comprises acquiring sensor data / processing (at 201), out- of-boundary transient detection (at 202) and integrator-based event detection (at 203).

[0045] Sensor data may be processed in various ways depending on the requirements of a given application. According to the present embodiment, sensor data is processed to obtain time series data representative of harmonic energy, which will be utilized to determine the instantaneous harmonic changes, also called delta energy (the delta energy being denoted AE) . AE is the difference between the sum of the selected harmonic energies and its mean value. The equation for AE is as follows:If Ek> Ek-thenEk= Ek-1 + (ffc - ffc-1elseEk= Ek— Ekwhere Ekis the sum of the selected harmonic energy at step k. The equation uses two time constants for estimating the mean harmonic energy, tau lower, and tau upper, which define the behavior respectively when Ek> Ek-and when Ek< Ek-.

[0046] According to the present embodiment, the out-of-boundary transient detection comprises the determination of the residual error between the delta energy and its baseline reference with regard to a threshold. It also comprises an estimation of the threshold. The baseline reference is calculated by determining a moving average of the delta energy, i.e. an average over a number of E^samples within a time window.

[0047] Fig. 3 is a diagram detailing the out-of-boundary transient detection of Fig. 2 according to one embodiment.

[0048] The residual error at a step k is obtained by subtracting (at 301) the baseline reference at step k from the acquired sensor data X(k) at step k. According to the process of Fig. 3, the baseline reference is calculated by applying a filter 302. Filter 302 can for example be an infinite impulse response (IIR) filter. Other filters may be used. The equation for an extended IIR filter can be as follows:

[0049] In equation 1, Xkrepresents the baseline reference at step k and k is an integer equal or greater than one. At is the time step. Tminand Tmaxare the minimum and maximum time constants for Xk. Pt is similar as to a probability density function of Xkevaluated under distribution equation (3) which varies in a range of [0, 1], An example can be as below:In equation 2, ak-is the estimated standard deviation at step k-1. When Tmax=min, the modified exponential weighted average (EWA) will be the same as traditional EWA. It is to be noted that the notation X(k) and Xkare equivalent. The resulting residual error A(k) is:A (k) = abs(Xk— XkEq. 3The time constant of filter 302 dynamically changes in the range of [Tmin, Tmax] depending on whether the current value of Xkis within the normal range or not.

[0050] According to a variant embodiment, the equation for an IIR filter can be as follows:The value of Tmean represents the time constant of filter 302, which functions as an exponentially weighted moving average. The value Tvarfor filter 305 is selected much larger than the Tmeanfor filter 302, considering that the threshold usually varies slowly over time. As illustrative and non-limiting example,meancan be of the scale of few seconds.

[0051] A threshold, or margin, is estimated to determine which values of Xklie outside of the boundary compared to the baseline reference. The threshold is estimated dynamically as a function of an average of residual errors at previous steps. The threshold is thus adaptive.

[0052] According to the present embodiment, residual errors which are potentially abnormal are discarded, i.e. not stored for determining the adaptive threshold. This will be described in more detail later in conjunction with Fig. 4.

[0053] In the diagram of Fig. 3, the M residual error values Ai, A2, ... M taken into account for the threshold estimation are stored in a buffer 303. According to one embodiment, a flag called ‘freeze’ is introduced which, when true, does not allow a residual value to be added to the buffer. The mean variation, mV, of the residual values is determined. In Fig. 3, this is shown at 304 (mean calculation) and 305 (variation calculation). The choice of M determines how sensitive to residual error variations the current signal Xkvalue is.

[0054] Block 305 can for example be implemented using a filter similar to filter302, i.e. an HR filter:

[0055] The threshold (i.e. margin) is obtained by multiplying the mean variation mV by a coefficient p, also called boundary parameter, p depends on the context of an actual implementation. In a normal distribution, when p = 2, it approximately equals to twice the standard deviance which corresponds to 95% of the confidence interval. For noisier systems, p could be larger. When the absolute value of a residual error is greater than the threshold and other block conditions are met, (test at 306), the corresponding data value is identified as an out-of-boundary transient and feeds the integrator of process 203. The value Tvarfor filter 305 is selected much larger than the Tminand Tmaxfor filter 302, considering that the threshold usually varies slowly over time. As illustrative and non-limiting examples, Tvarcould be equal to a few seconds or minutes whereas Tmiriand Tmnxcan be in the scale of few seconds. Optionally, the other block conditions in 306 can be defined based upon the needs of different applications. For instance, in a power line system, the percentage of changesfor an anomaly event may need to be higher than a certain value. Or it could be that the mean variation mV has to be larger than the system noise level if it is available. At 307, an appropriate output of process 202 is generated based on the output of the threshold comparison at 306 to feed process 203.

[0056] In case equation 1’ of the variant embodiment is used,meanis the estimation of time constant of filter 302 which is between TmIlWiriit and Tmnx. Tvnris then selected much larger than Tmean-

[0057] Fig. 4 is a diagram detailing the integrator-based event detection process of Fig. 2 according to one embodiment. The process uses a leaky integrator which accumulates out-of-boundary transients detected by process 202, but also leaks over time. According to one embodiment, the behavior of integrator 401 can be described as follows:If OBk> Tk-thenwhere OBkrepresents the output of the out-of-boundary transient detection process 202 for the current signal value at step k, current transient, Ikis the integrator output at step k, At is the time step, tauupperis a first time constant applied when an out-of- boundary transient is detected, tauioweris a second time constant applied when no out-of-boundary transient is detected. tauupperdefmes how fast the integrator accumulates out-of-boundary transients, whereas taiqowerdefmes how fast the integrator leaks.

[0058] According to the present non-limiting embodiment, OBkis defined (at 402) as follows: if an out-of-boundary transient is detected at step k, then OBk= 1, else OBk= 0. Since the input to the integrator is either 0 or 1, the output signal of the integrator will be in the range [0, 1],

[0059] At 403, the output Ikof the integrator is compared to a threshold ‘thr’.If / fc is greater than thr, an event is detected and a corresponding alarm signal is generated (at 404).

[0060] A mechanism is furthermore implemented to reduce or eliminate the impact of signal values following event detection on the calculation of the threshold mV*p used to determine whether a value is an out-of-boundary transient or not. Indeed, given relatively large oscillations during or following an event, signal values may lie within the margin defined by the threshold mV*p around the baseline reference, but such signal values may modify the margin calculation in an unpredictable manner. In the frame of the present embodiment, such signal values are blocked from contributing to the margin calculation (i.e. the corresponding residual error is not stored in buffer 303) during a time period following the detection of an event. This can, for example, be performed using a countdown timer T1 and a flag (‘Freeze’), as already mentioned in conjunction with the description of process 202 and Fig. 3. When the integrator output is greater than threshold thr, countdown timer T1 is triggered (405) to start counting down and the Freeze flag is set to ‘True’ (406). When the integrator output is lower than threshold thr, i.e. no event is detected, a check is carried out whether the timer T1 is on (still counting down) or off (countdown is over, or timer was not triggered). If the timer is off, then the flag is set to ‘false’ . While the flag is True, residual errors - whether corresponding or not to out-of-boundary transients - are not stored in buffer 303 and hence do not contribute to the determination of mean variation mV and consequently to threshold mV*p. Timer T1 duration is function of a specific application, e.g. function of the duration of an event and its oscillations.

[0061] Fig. 5 is a flowchart of an event detection method according to one or more non-limiting embodiments. The process comprises:S500: Start event detection process.S501 : Measure system variable, for example using a sensor such as sensor 103 or an edge device. According to the present embodiment, the sensor signal or signals are further processed to obtain the input data X(k), e.g. delta energy.S502: Determine residual error At(k)=abs(X(k)-F(X(k))) where F is a filter function, e.g. based on equation 1.S503: Determine mean variance mV by filtering the mean values of At(k), k=l, 2,...M, where M historical samples of At(k) are stored in a buffer.S504: If (At(k)>mV*p, then X(k) is an out-of-boundary transient.S505: If X(k) is an out-of-boundary transient, update integrator with input 1, else update with input 0.S506: If integrator output < alarm threshold, keep timer T1 off, else trigger timer Tl. S507: If timer Tl is on or if residual error At(k) is larger than the margin mV*p, do not update buffer, else update buffer with At(k) and flush oldest value in buffer if the buffer is full.S508: If integrator output > alarm threshold, trigger event detection alarm.

[0062] Fig. 6a and Fig. 6b are graphs illustrating the detection of an event corresponding to a high impedance fault in a power line system, applying the processes disclosed herein. In the case of Figs. 6a and 6b, the HiZ fault occurs at 5s. Fig. 6a is a graph showing the change in harmonic energy over time (X(k)) for an exemplary HiZ fault, the associated baseline reference (light grey line) and the estimated margin around the baseline reference (two dashed lines, respectively above and below the baseline reference). Data points considered as out-of-boundary transients, i.e. lying beyond the margin, are identified by fatter dots. Fig. 6b is a graph showing the output of the integrator, with an event detection threshold of 0.3, identified by a horizontal dashed line. A vertical dashed line identifies the moment when the integrator output crosses the event detection threshold. In Fig. 6, the time series data points representative of the change in harmonic energy oscillate around the baseline reference until the HiZ fault appears. It is to be noted that although the data points cross the estimated margin before the HiZ event, which leads to out-of-boundary transients being detected, this does not result in the integrator providing an output greater than the event detection threshold. This is obtained through the leaking of the integrator. Only when the out-of-boundary transients occur substantially continuously due to the HiZ fault does the integrator output reach the event detection threshold and trigger the event alarm. The time series data points following the event detection can be recorded and stored for further feature extraction, HiZ fault classification and appropriate action.

[0063] Fig. 7 is a schematic diagram of an event detection device 101. Device 101 comprises a processor or microcontroller 701, a working memory 702, a long term memory 703 and a collection of interfaces 704. Blocks 701 to 704 are connected through bus 705. The long term memory 703 holds software code. The interfaces 704 comprise the interfaces with bus 705, sensor 103 and control circuit 104. When the processor 701 executes the software code, it causes the device 101 to carry out one or more of the methods described herein. For example, processor 701 may execute software code causing device 101 to carry out out-of-boundary data detection and processing. Device 101 may comprise additional functional blocks, such as a user interface or a display.

[0064] The proposed method and device enable monitoring of a baseline reference and noise variance of the system, and identifies abnormal changes. The processing can be performed in real time on high-rate data. The method can easily be implemented in a device able to perform simple data processing, such as an ZED. Only the data corresponding to detected events need be stored and processed for further analysis in real time. Complexity and cost is thus maintained at a reasonable level.

[0065] The use of a fixed threshold or margin to detect out-of-boundary transients in the data points can lead to the selection of a relatively high value to avoid false alarms. This can however result in a failure to detect subtle changes. The embodiments disclosed herein implement an adaptive threshold a.k.a margin combined with a leaky integrator used as an over-threshold event counter. This allows limiting false alarms while still enabling detection of subtle changes that persist over a relatively long time.

[0066] One advantage of utilizing a leaky integrator instead of a traditional counter is its ability to perform a weighted sum. Unlike simple counters, which disregard the timing pattern of the items being counted, a leaky integrator gives more weight to recent events compared to those that occurred in the distant past. In addition, the integrator leaks over time. Sparse events are thus forgotten over time before being able to contribute to reaching a given integrator output value that would trigger event detection. This feature can help prevent noise from triggering a false alarm.

[0067] Methods that use low pass filters to smooth signals may be used to reduce false alarms and detect subtle changes. However, these methods limit the high frequency characteristics, and may thus fail to detect events associated with high frequency changes. The embodiments described herein do not rely on low pass filtering.

[0068] When an anomaly event has been detected by event detector 101, the subsequent alarm can trigger classification of the event, e.g. whether the event corresponds to a high impedance fault or not.High Impedance Fault identification

[0069] Fig. 8 is a flow chart of a method 800 for event classification according to a first embodiment. The method comprises extraction of features from harmonic signal components (at 801) and event classification based on the extracted features (at 802). A buffer 803 stores the signal data that was used for the event detection.

[0070] Feature extraction determines specific features in the signal which can subsequently be used to distinguish a high impedance fault from other events.

[0071] According to the present embodiment, feature extraction is performed while an event is being detected. In the frame of the previous section, the alarm signal provided by detector 105 can be used to determine when feature extraction is to be carried out. Feature extraction uses spectral components obtained from the sensor signals. Various sensor signals may be used for this purpose. In a three phase, three wire electrical system, device 101 will obtain measurements for each of the three phases. In a three phase, four wire electrical system, device 101 will obtain measurements for each of the three phases and of the neutral - or ground - channel. In a single phase system, measurements for the single phase are available. According to some embodiments, for the purpose of detecting high impedance faults, it has been found that using the current of the neutral channel, the sum of currents of the three phases when no neutral channel is available, or the current of the single phase signal provides good results. Corresponding voltages may also be used. In Fig. 8, the exemplary input of a neutral current harmonic energy HINxis shown. The featureextraction may also optionally receive a user configuration as input (not shown). User configuration can for example identify one or more harmonics to be part of a set used for processing.

[0072] The functional blocks of Fig. 8 can be implemented using a device such as the one of Fig. 7.

[0073] According to one or more embodiments, a sensor signal is digitally sampled and filtered in a manner know per se to obtain time series data of the raw electrical signal. Time series data characterizing the spectrum of the signal can be obtained for example using a spectral analyzer. The spectral analyzer may for example be based on a discrete Fourier transform or a wavelet transform acting upon data within a sliding temporal window. Spectral data can include the magnitude, phase and / or energy of the spectrum or other parameters. The values are calculated at regular intervals and correspond to the data Ekin the section above. In the present nonlimiting embodiment, the harmonic energy magnitude of a chosen set of harmonics is used. When several harmonics are considered, the harmonic energy magnitude used for event detection and classification can for example be the sum of the energy magnitudes of the individual harmonics, as mentioned in conjunction with one of the embodiments in the event detection section. Since a high impedance fault is usually associated with bursts of harmonic energy, the primary input to the event detection and classification is typically data representative of change in the harmonic energy of single phase or grounding currents, e.g. the 3rdand the 5thharmonics. According to a variant embodiment the energy magnitude of a single harmonic is used as input.

[0074] According to one or more embodiments, fault classification is based on the detection of signal random components in the frequency domain, associated with a duration criterion of the period during which such randomness is detected. Indeed: o Arcing is the most prevalent physical signature of a high impedance fault. Arcing events are associated in the frequency domain with sudden bursts of harmonic energy. During arcing, the impedance characteristic of the arc can vary rapidly. Between two successive electrical power cycles, the current will show distortions - this will not be the case in normal circumstances. Theresulting randomness in the sensed signal or signals can be evaluated for example on a cycle-to-cycle basis. A ‘cycle’ corresponds to the fundamental frequency of the sensed signal or signals. This frequency is e.g. 50 or 60 Hz for alternative current in a power line. o Events linked to high impedance faults last longer than events linked to normal operations. Indeed, high impedance faults are persistent until cleared, which is not the case for events linked to normal operation, which are hence of shorter duration.

[0075] In one or more embodiments, it is thus proposed to (a) quantify random signal components and (b) quantify the length of time during which these components appear. An event is classified as a high impedance fault if random signal components are present for at least a certain time.

[0076] Fig. 10 is a more detailed flowchart of the method of Fig. 8. According to Fig. 10, time series data representative of harmonic energy of one or more electrical signals in a power system are obtained (SI 000). Then, values of an indicator quantifying the presence of random components in the harmonic energy data are calculated iteratively over time (S1001). For each value of the indicator, it is determined whether it indicates the presence or not of random components. This is achieved by comparing each value to a first threshold at SI 002. Within the present embodiment, the first threshold is fixed and the comparison is carried out when an event is being detected (e.g. in the frame of the example of Fig. 4, based on output 404). The result of the comparison is integrated over time (SI 003). E.g. an out of boundary value count is performed over time. It is then determined at SI 004 whether an output of the integrator is above a second threshold for a duration greater than a third threshold and if the determination is positive, a signal indicative of a high impedance fault is generated.

[0077] The integration is performed using a leaky integrator, i.e. indicator values indicating the presence of random components at a level that could be associated with high impedance fault events are counted over time - thanks to the leaking, sparsely appearing values will however not enable reaching the integrator output threshold, i.e. the second threshold, limiting the impact of noise. Lastly, aduration criterion is applied with a third threshold, to distinguish between the duration of a normal event and that of a high impedance fault event.

[0078] Fault classification according to the first embodiment is based on the evaluation of energy correlation. More precisely, the temporal correlation of the energy magnitude of harmonics of a signal is evaluated. The correlation may for example be determined over two adjacent time windows each of the same length, of one or more fundamental frequency cycles. Correlation coefficients are determined for respective shifted positions of the windows. The correlation coefficients are input to a decision logic to provide the classification result with regard to a high impedance fault. The correlation coefficient is the indicator of presence of random components mentioned earlier.

[0079] To minimize the impact of background noise on the evaluation of energy correlation, the statistics t-test, named t-score, is applied on two consecutive- window harmonics. It determines whether the randomness is purely due to background noise. The t-score module will check whether the two consecutive windowed energy waveforms are filling into the same distribution. It mainly compares the mean values of the two windowed energies.

[0080] For example, the total energy (i.e. the sum of the magnitudes) of the 3rdand 5thharmonics of the neutral current signal is used. It is to be noted that other signals and corresponding frequency domain features can be used.

[0081] Fig. 9 is a flowchart of the first method according to a present nonlimiting embodiment. Energy correlation processing is performed during the time when event detection integrator output 403 is greater than threshold thr. A first buffer A holds recorded time series data which is used to calculate harmonic components. A second buffer B holds the harmonic component magnitudes (or the sums of the magnitudes of selected harmonic components) stored over time and used for determining the energy correlation and t-score over two adjacent time windows. E.g. a new value of harmonic component magnitude may be added to the buffer every quarter of a cycle. The two time windows have a same length equal to an integer number of cycles. E.g. if each window has a length of two cycles, the two windows will cover 2*8 values in the second buffer. The second buffer may have a sizecorresponding to the number of values covered by the windows.

[0082] The correlation of the harmonic energy and t-score between the two adjacent time windows is determined (S901) - the outputs are a correlation coefficient and a t-score. The correlation coefficient is compared to a first threshold Tc (S902-1). A coefficient below the threshold indicates low correlation, i.e. high randomness. The t-score is compared to the statistic threshold referenced from the student t-test table Ts (S902-2). The number of degrees of freedom (df) is calculated as follows:Where tol is the tolerance to avoid the denominator being equal to 0. For example, Tol = le-5. df is not always an integer - hence the rounded value of df will be used as the final degree of the freedom. A t-score above the threshold Ts indicates that the randomness is not purely from background noise. The combination of the two comparisons is integrated (S903). If the integrator output is above a second threshold Te for a duration greater than a threshold Di (S904), an output fault signal is generated (S905). Else, the windows are translated by a step k and the next coefficient is calculated at S901.

[0083] Fig. 11 is a diagram that illustrates adjacent windows XI and X2, each having the length of two cycles and that can be used to explain the correlation calculation according to the present example. The correlation coefficient Y 1 between the harmonic energy data falling within the two windows is determined. The sinusoids shown in the figure do not represent the data itself - they are simply indicated to illustrate the temporal cycles covered by the windows for the purpose of clarity. Then the two windows are shifted to a new position and a new correlation coefficient is calculated (Y2 in Fig. 11). The shifting step determines the overlap of windows between two coefficient calculations and the resolution of the calculation, i.e. the smaller the step, the higher the resolution. The step may for example be a quarter of a cycle, i.e. a coefficient is output each quarter of a cycle.

[0084] According to a variant embodiment, the correlation is determined using energy of higher order harmonics or high pass filtered current time series.

[0085] For evaluation of the correlation, various methods can be used. In the frame of the present embodiment, a Pearson correlation is applied. Pearson correlationis known per se. The Pearson correlation coefficient is the covariance of the two variables divided by the product of their standard deviations. It measures the linear correlation between two data sets (i.e. the harmonic energy magnitude time series data in windows XI and X2).

[0086] The resulting coefficients, t-score as well as the event alarm of Fig. 4 are used as input to a decision logic. The decision logic provides an output indicating whether the analyzed data characterizes a high impedance fault or not.

[0087] Fig. 12 is a diagram illustrating the various functional blocks implementing the decision logic of the first method according to an exemplary embodiment. The correlation coefficients produced by energy correlator 1201 as described above are input to a comparator 1202. Comparator 1202 compares each coefficient to a coefficient threshold Tc. If a given coefficient is below the threshold Tc, it is determined that there is little correlation between the data that was used to calculate the given coefficient and that the cycle-to-cycle randomness is high. The t- score produced by 1203 compares the distribution of two consecutive-window energies. If a given t-score is above the threshold Ts, it is determined that the cycle- to-cycle randomness measure through correlation is not purely due to the background noise. If an event alarm is raised, correlation coefficients and t-score are calculated and the output is processed by 1205 - this is symbolically represented by a logic AND. Integrator 1206 counts the true logics from 1205. The integrator can for example be a leaky integrator summing up the number of coefficients below threshold Tc. The integrator may be implemented as already described in the event detection section, with the same benefits.

[0088] Optionally, the counting of true logics by integrator 1206 can be blocked by a blocking signal also input to the logic AND 1205. Blocking may be useful e.g. when an event corresponding to normal operation is taking place.

[0089] Output of integrator 1206 is compared to an integrator threshold Te by comparator 1207. If the output of the integrator is greater than threshold Te, for a duration threshold Di, then a high impedance fault is detected and a corresponding fault signal 1208 is generated.Second Embodiment

[0090] An embodiment of a second method for event detection and fault classification will now be described. The second method evaluates cycle-to-cycle randomness based on the temporal ratio of transient and static current values. Depending on the implementation, the evaluation is based on the true root mean square of the sensed signal, e.g. current, or the scaled root of the sum of the energy of harmonic components (this scaled root being called harmonic RMS) of the sensed signal. In the context of the present embodiment, the latter will be considered. The evaluation takes into account either the variance of the harmonic RMS or to the coefficient of variation of the harmonic RMS, calculated over a time window.

[0091] The description of the present embodiment uses harmonic energy of one or more electrical signals in the system as an exemplary basis for determining the presence of random components, the equations provided being based on that parameter. However, it is clear to the person skilled in the art that he / she can equivalently use directly a current and / or voltage signal as input and adapt the equations as necessary.

[0092] Fig. 13 is a graph schematically showing a time window 1301 (Tong window’) for which a ratio as described above is to be calculated. Harmonic RMS valuesare calculated for the temporal signal data corresponding to each position of a window 1302 (‘short window’), with k an integer between 1 and N. Short window 1302 is shifted by a shifting step in the direction of the arrow in Fig. 13 for each increment in k, within long window 1301. The shifting step can for example be of one quarter cycle. N represents the number of short windows 1302 fitting into long window 1301 with this shifting step. The N harmonic RMS values, {x1,x2, ... xk, . . xw}, obtained for a long window are stored in a buffer.

[0093] The true RMS is the RMS of the signal, which is equivalent to the root of the sum of the energy of all harmonics and the fundamental, divided by 2.

[0094] Based on these N harmonic RMS values in the buffer, an output value can be either A or A / B if a normalized output is desired, with A equal to Al or A2:or2 = sqrtand withB = mean(x1,x2ixN) where X^ is the harmonic RMS.

[0095] According to a variant embodiment, only one combination among Al, A2, Al / B, A2 / B is implemented, but depending on the context, more combinations may be offered for automatic selection or selection by a user.

[0096] The flowchart of Fig. 10 described in conjunction with the first embodiment also applies to the second embodiment, with the difference that the threshold used to determine the presence of random components as a function of the indicator is adaptive.

[0097] Fig. 14 is a flowchart of the second method according to the present non-limiting embodiment. The process starts at S1401. A value for either A or A / B (depending on implementation) is obtained at S1402, as previously explained. A difference with a baseline reference is obtained. This difference is the indicator of the presence of random components mentioned earlier. The baseline reference results from the smoothing of past values of A or A / B, In the present embodiment, it is a moving average of past values over a defined time window, and it may be implemented in a way similar to what has been described in conjunction with the event detection of the first embodiment. At S1403, the difference is compared to a threshold Tr. A difference above the threshold Tr indicates high randomness, i.e. the presence of random components that may indicate a high impedance fault event. The output of the comparison is integrated (S1404). If the integrator output is above a threshold Tims for a duration greater than a threshold Dr (S1405), an output fault signal is generated (S1406). Else, the window 1301 is moved (S1407) and the next difference is calculated at S1402. If the integrator output is lower than threshold Tims for a duration greater than a threshold Dr, then the process also goes to S1407.

[0098] According to one embodiment, threshold Tr is adaptive, in a fashion similar to what was described in conjunction with the event detection method and Fig. 3. In the present embodiment, the threshold is defined as a function of the mean ofbuffered differences, multiplied by a parameter p. A ‘freeze’ flag can be implemented to avoid unwanted out-of-boundary difference values impacting the determination of this threshold. A timer can also be implemented to exclude difference values from the difference value buffer.

[0099] The integrator can for example be a leaky integrator summing up the number of differences above threshold Tr. The integrator may be implemented as already described in the event detection section of the first embodiment, with the same benefits. E.g. it may count zeros and ones as a function of the output of the threshold comparison with Tr at S1403.

[0100] Fig. 15 is a block diagram of the functional blocks of a device implementing part of the second method. A multiplexer 1501 controlled by a selector at an input ‘c’ receives either the true root means square values, or the harmonic RMS values calculated for the short windows as previously explained. The selected values are stored in a buffer 1502. Calculator 1503 iteratively determines A / B (or A) values used by decision logic 1504 to generate a fault signal function of the A / B (or A) values. A blocking signal may optionally be applied to the decision logic.

[0101] Fig. 16 is a block diagram of the decision logic of Fig. 15 according to a non-limiting embodiment. The decision logic will check if the instantaneous variation of the A / B (or A) value is abnormal, i.e. whether the difference between the instantaneous A / B value and its baseline reference (calculated similarly as 302 in Fig. 3) is greater than the adaptive threshold Tr estimated from variance calculator 1603. This is done using out of boundary detector 1601. Both the out-boundary detector 1601 and the variance calculator 1603 are designed similarly to what has been described in conjunction with Fig. 3 (i.e. in particular references 303 to 309). The absolute value of the difference between the current A / B (or A) value and the baseline reference is fed to variance calculator 1603, unless the freeze flag has been set or the out of boundary value is true. The freeze flag is generated similarly to the freeze flag of Fig. 4. An event detection logic 1602 integrates the output of the out-of-boundary detector 1601 in a fashion similar to what was described in conjunction with Fig. 4, with duration Dr replacing duration dur thr.

[0102] While in the above embodiments concerning the first and secondmethods a fault signal is generated separately for each method, a variant embodiment implements both methods. Depending on user preference, an output of a fault signal may be generated when either or both methods indicate that a fault signal is to be generated.

[0103] Following the detection of a high impedance fault event, device 101 can issue a control command to trip or disconnect a switch to isolate the power source feeding the part of the system or the line for which the fault was detected.

[0104] The methods described in the second embodiment can be implemented using the device of Fig. 7.

[0105] Two embodiments for detecting a high impedance fault have been described. Each of these embodiments can be used separately. It is also possible to use both embodiments in parallel and to combine fault signal 1206 and fault signal 1505. For example, a high impedance fault event alarm signal is generated only if the output of both embodiments yields a coherent result.

[0106] Fig. 17a and Fig. 17b are a collection of graphs illustrating signals and outputs when applying the first embodiment to ground current signals obtained respectively for the case of a conductor falling on grass (i.e. a high impedance fault event) and for a motor start (i.e. an event occurring during normal operation). The output of the integrator is shown in the second graph from the top, whereas the output of the correlation calculation is shown in the third graph. Thresholds are indicated by dashed horizontal lines. The high impedance fault event alarm signal is provided in the bottom graph. The integrator in the event detection (second row) detects both the fault and non-fault cases. The detected events will be further processed for feature extraction (third row). In this example, the feature shown in the third row is the harmonic energy correlation. The behavior of this features is much different between these two cases: due to the randomness in high impedance fault case the correlation fell below the threshold (dashed line) for a long period time whereas the correlation in the motor start case only fell transiently. The fault event alarm is raised within 2seconds after the start of the fault event. It is never raised for the normal event.

Claims

CLAIMS1. A method for high impedance fault event detection in a power system, said method being carried out by a device comprising a processor, said method comprising:- obtaining (SI 000) time series data representative of harmonic energy of one or more electrical signals in the system;- determining (SI 001) values over time of an indicator of presence of random components in the harmonic energy;- comparing (SI 002) each indicator value or values to a first threshold (Tc,Ts, Tr) to determine the presence of random components;- integrating (SI 003) a result of the comparison over time using a leaky integrator (1204);- determining (SI 004) whether an output of the integrator is above a second threshold (Te, Trms) for a duration greater than a third threshold (Di, Dr) and if the determination is positive, generating a signal (1206, 1505) indicative of a high impedance fault event.

2. The method of claim 1, further comprising determining the presence of an anomaly event as a function of the time series data, comprising:- determining instantaneous harmonic energy changes (AE) as a function of the time series data. determining (202, S502-S504) whether a difference (A(k)) between an instance of the instantaneous harmonic energy change (X(k)) and a moving average of the instantaneous harmonic energy changes is greater than a fourth threshold (T4), wherein the fourth threshold is function of a mean of said difference; as a result of the determination, providing (S505) an input to a leaky integrator (401) configured to count data points for which the difference is greater than the fourth threshold;determining (S508) whether an output of said integrator is greater than a fifth threshold (thr) and in the affirmative, determining (404, S508) that an anomaly event is present.

3. The method of claim 1 or 2, wherein the indicator of presence of random components in the harmonic energy is a correlation coefficient resulting from a correlation of the time series data over two adjacent time windows of same length substantially equal to an integer multiple of a fundamental period of the one or more electrical signals.

4. The method of claim 3, wherein the two adjacent time windows are shifted in time by a shifting step for the determination of successive values of the indicator.

5. The method according to one of the claims 3 or 4, wherein the comparison of the indicator with the first threshold (Tc) indicates the presence of random components if the indicator is below the first threshold.

6. The method according to one of the claims 1 to 5, wherein the indicator of presence of random components in the harmonic energy is function of a scaled root of time series data representative of the sum of at least one harmonic energy.

7. The method according to claim 6, wherein the indicator is equal to a difference between A, respectively A / B and a moving average of A, respectively A / B, with A equal to Al or A2, and712where X^is the scaled root of the sum of the energy of at least one harmonic calculated for time series data in a first time window shifted among N positions within a second time window, with k = 1 to N and N>1.

8. The method according to one of the claims 6 or 7, wherein the comparison of the indicator with the first threshold (Tr) indicates the presence of random components if the indicator is greater than the first threshold.

9. The method according to claim 8, wherein the first threshold (Tr) is function of the variance of past values of A, respectively A / B.

10. The method according to one of the claims 1 to 9, comprising: generating a control signal of the power system as a function of the signal indicative of a high impedance fault event.

11. Device (101) for high impedance fault detection in a power system (102), said device comprising a processor (701) and memory (703), said memory comprising software code, the processor causing the device to perform one of the methods according to claims 1 to 9 when executing said software code.

12. Computer-readable storage medium comprising code which when executed by a processor of a device causes the device to perform the method according to one of the claims 1 to 10.

Citation Information

Patent Citations

  • High-impedance fault detection method based on Pearson's correlation coefficient algorithm

    CN116125174A

  • Data analytic engine towards the self-management of complex physical systems

    US20140108324A1

  • Method and system for testing a signal path having an operational signal

    WO2004084033A2

Cited By

  • High impedance fault location in electric power distribution systems

    EP4722740A1