An arc fault intelligent detection method and detection device

Abstract

This invention relates to the field of arc fault detection technology, specifically to an intelligent arc fault detection method and device. The method includes: acquiring electrical data from monitoring points; filtering current and voltage surge peaks based on the amplitude distribution of electrical data in the current and historical periods; constructing a bipartite graph with the current and voltage surge peaks as vertices; using the time interval between the current and voltage surge peaks as a matching weight to match the current and voltage surge peaks, obtaining matching pairs; obtaining the relative standard deviation of the ratio based on the overall distribution characteristics of the voltage amplitude to current amplitude ratio in the matching pairs; obtaining the proportion of isolated current surges based on the proportion of unmatched current surge peaks; and determining the current arc state by combining the relative standard deviation and the proportion of isolated current surges. This invention improves the accuracy of arc fault detection results.

Description

Technical Field

[0001] This invention relates to the field of arc fault detection technology, and specifically to an intelligent method and device for arc fault detection. Background Technology

[0002] The electrical environment at construction sites, temporary power systems, and the ends of some industrial power distribution networks is highly complex. These scenarios widely feature nonlinear loads such as tower cranes, construction hoists, and various inverter welding machines controlled by frequency converters. When these power electronic devices are operating, the high-frequency switching of their internal power devices continuously injects broadband pulse noise into the power grid. This noise manifests as spike pulses in the time domain and covers a frequency range from several kilohertz to several megahertz, highly overlapping with the characteristics of random high-frequency noise generated by series arcing faults caused by aging insulation or poor contact in the lines.

[0003] Existing arc fault detection technologies primarily rely on single-channel spectral analysis of branch current signals or simple time-domain energy integration. However, in the aforementioned high-noise environments, current signals alone are insufficient to effectively distinguish conducted interference generated during normal equipment operation from genuine fault arcs. From an electrical principle perspective, the interference generated by frequency converters is essentially current-source noise, injecting high-frequency current into the power grid. However, due to the typically low internal resistance of the power supply side of the distribution network, these current pulses usually do not cause significant voltage drops at the common bus. Conversely, series arc faults are resistive blocking faults, especially at the end of low-voltage distribution networks or in long-distance cable power supply circuits. Due to the distributed impedance of the lines, the instantaneous current change during arc occurrence will generate synchronous fluctuations (voltage drops or distortions) in the port voltage of the monitoring point. Existing technologies have not effectively utilized this electrical characteristic difference in voltage-current synchronous response, or simply employ simple time window overlap for judgment. In situations where multiple devices operate simultaneously and pulses are dense, simple time window methods are prone to mismatches, such as incorrectly associating the current pulse of device A with the voltage fluctuation of device B, resulting in low accuracy of arc fault detection results. Summary of the Invention

[0004] To address the issue of low accuracy in existing methods for detecting arc faults, this invention aims to provide an intelligent method and device for arc fault detection. The specific technical solution adopted is as follows: In a first aspect, the present invention provides an intelligent method for detecting electric arc faults, the method comprising the following steps: Acquire electrical data from monitoring points, including current and voltage signals; Based on the amplitude distribution of electrical data in the current and historical periods, peak points of current and voltage sudden changes are selected respectively. A bipartite graph is constructed with the peak points of current and voltage sudden changes as vertices. The time interval between the peak points of current and voltage sudden changes is used as the matching weight to match the peak points of current and voltage sudden changes, and each matching pair is obtained. Based on the overall distribution characteristics of the ratio of voltage amplitude to current amplitude in the matching pair, the relative standard deviation of the ratio is obtained; based on the proportion of current surge peaks in unmatched pairs, the proportion of isolated current surges is obtained. The current arc state is determined by combining the relative standard deviation and the proportion of isolated current mutations.

[0005] Preferably, the peak points of sudden current changes are screened, including: The current difference sequence is obtained by performing neighborhood difference calculation on the current signals of corresponding sampling points in the current period and the historical period; Local peak detection is performed on the current differential sequence, and local maxima points whose amplitudes exceed a preset current amplitude threshold are selected as candidate points for current abrupt changes. All current mutation candidate points are sorted in descending order of amplitude, and the first preset number of current mutation candidate points are taken as the current mutation peak points.

[0006] Preferably, the step of performing neighborhood difference calculation on the current signals of corresponding sampling points in the current period and historical periods to obtain a current difference sequence includes: For any sampling point in the current cycle, calculate the absolute value of the difference between the current amplitude of the sampling point and the same index point in the historical cycle and its adjacent sampling points before and after it, and take the minimum value of the three absolute values ​​as the current difference value of the sampling point. The current difference values ​​of all sampling points within the current period constitute the current difference sequence.

[0007] Preferably, the step of constructing a bipartite graph with current and voltage surge peaks as vertices, using the time interval between current and voltage surge peaks as matching weights, and matching current and voltage surge peaks to obtain matching pairs includes: Construct a bipartite graph with the peak point of current change as the left vertex and the peak point of voltage change as the right vertex; If the time interval between the current surge peak point and the voltage surge peak point is greater than the preset maximum synchronization tolerance threshold, then the edge weight between the corresponding current surge peak point and the corresponding voltage surge peak point is set to a preset value; if the time interval between the current surge peak point and the voltage surge peak point is less than or equal to the preset maximum synchronization tolerance threshold, then the time interval is used as the edge weight between the corresponding current surge peak point and the corresponding voltage surge peak point. The edge weights between all current surge peaks and voltage surge peaks constitute a weight matrix; Based on the weight matrix, the peak points of current sudden changes and the peak points of voltage sudden changes are matched to obtain several matching pairs.

[0008] Preferably, obtaining the ratio of the voltage amplitude to the current amplitude in the matching pair includes: For any matching pair: The ratio of the amplitude of the voltage surge peak point to the amplitude of the current surge peak point in any matched pair is taken as the ratio of the voltage amplitude to the current amplitude in any matched pair.

[0009] Preferably, the step of obtaining the relative standard deviation of the ratio based on the overall distribution characteristics of the ratio of voltage amplitude to current amplitude in the matched pair includes: The number of valid matching pairs that meet the amplitude validity criteria among the statistical matching pairs; If the number of valid matching pairs is greater than or equal to the preset sample size, then the mean and standard deviation of the ratios corresponding to all valid matching pairs are calculated respectively. By combining the mean and the standard deviation, the relative standard deviation of the ratio is obtained.

[0010] Preferably, obtaining the proportion of isolated current surges based on the proportion of unmatched current surge peaks includes: The number of unmatched current abrupt peak points is recorded as the number of isolated current points. Divide the number of isolated current points by the total number of current surge peaks to obtain the percentage of isolated current surges.

[0011] Preferably, determining the current arc state by combining the relative standard deviation and the proportion of isolated current mutations includes: If the proportion of isolated current abrupt changes exceeds a preset threshold, it is determined to be conducted interference; If the proportion of isolated current mutations is less than or equal to a preset proportion threshold, and the relative standard deviation is less than or equal to a preset deviation threshold, then it is determined to be a suspected arc fault. If the proportion of isolated current mutations is less than or equal to a preset proportion threshold, and the relative standard deviation is greater than a preset deviation threshold, then it is determined to be a random disturbance.

[0012] Preferably, the step of combining the mean and the standard deviation to obtain the relative standard deviation of the ratio includes: The ratio between the standard deviation and the mean is used as the relative standard deviation of the ratio.

[0013] In a second aspect, the present invention provides an intelligent arc fault detection device, the device comprising: The acquisition module is used to acquire electrical data from the monitoring point, including current signals and voltage signals; The matching module is used to filter current surge peak points and voltage surge peak points according to the amplitude distribution of electrical data in the current cycle and historical cycles, respectively; construct a bipartite graph with current surge peak points and voltage surge peak points as vertices, use the time interval between current surge peak points and voltage surge peak points as matching weights, and match current surge peak points and voltage surge peak points to obtain each matching pair; The analysis module is used to obtain the relative standard deviation of the ratio based on the overall distribution characteristics of the ratio of voltage amplitude to current amplitude in the matched pair; and to obtain the proportion of isolated current mutations based on the proportion of current mutation peaks in unmatched pairs. The judgment module is used to determine the current arc state by combining the relative standard deviation and the proportion of isolated current mutations.

[0014] The present invention has at least the following beneficial effects: This invention first filters current and voltage surge peaks based on the amplitude distribution of electrical data from the current and historical periods. By matching current and voltage surges and utilizing the overall distribution characteristics of the voltage-to-current amplitude ratio in the matched pair, the consistency of the response is evaluated. This eliminates interference from non-faulty electrical events such as random disturbances or nonlinear load switching on the detection results, enhancing the robustness of the detection method in complex electrical environments. By introducing the proportion of unmatched current surges, current surges that do not cause a synchronous voltage response can be identified, thus accurately distinguishing real series arc faults. This avoids the misjudgment problems caused by traditional current signal analysis alone, improving the accuracy of arc fault detection results. Attached Figure Description

[0015] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1A flowchart of an intelligent arc fault detection method provided in an embodiment of the present invention; Figure 2 This is a structural block diagram of an intelligent arc fault detection device provided in an embodiment of the present invention. Detailed Implementation

[0017] To further illustrate the technical means and effects adopted by the present invention to achieve the intended purpose, the following detailed description of an intelligent arc fault detection method and detection device proposed according to the present invention is provided in conjunction with the accompanying drawings and preferred embodiments.

[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0019] The following description, in conjunction with the accompanying drawings, details the specific scheme of the intelligent arc fault detection method and detection device provided by the present invention.

[0020] An embodiment of an intelligent arc fault detection method: This embodiment proposes an intelligent detection method for electric arc faults, such as... Figure 1 As shown, an intelligent arc fault detection method according to this embodiment includes the following steps: Step S1: Obtain electrical data from the monitoring point, including current and voltage signals.

[0021] First, electrical data is collected from monitoring points at the end of the low-voltage distribution network or in long-distance power supply circuits. A hardware comparator detects the zero-crossing point of the voltage at the monitoring point. When a zero-crossing event is detected, a high-precision analog-to-digital converter (ADC) is triggered to initiate synchronous voltage and current sampling. The ADC simultaneously acquires branch current and port voltage signals, and the acquisition process lasts for one complete power frequency cycle, with a fixed number of sampling points. In this way, the system acquires the current data sequence (current signal) and voltage data sequence (voltage signal) for each cycle. Each data point in the sequence is indexed by a time sampling point. Unique identifier ;in, The value can be 256 or 512. In this embodiment, The value is 512.

[0022] The above methods were used to collect current and voltage signals for each cycle at the monitoring point. The current and voltage signals are collectively referred to as electrical data.

[0023] Step S2: Based on the amplitude distribution of electrical data in the current cycle and historical cycles, filter out the peak points of current and voltage sudden changes respectively; construct a bipartite graph with the peak points of current and voltage sudden changes as vertices, use the time interval between the peak points of current and voltage sudden changes as the matching weight, and match the peak points of current and voltage sudden changes to obtain each matching pair.

[0024] When assessing current arc faults, it is necessary to comprehensively analyze the distribution characteristics of electrical data from the current cycle and historical cycles. The historical cycle refers to the cycle preceding the current cycle. It should be noted that the first cycle after system power-on is not analyzed due to a lack of historical data.

[0025] When analyzing the amplitude variations of electrical data at sampling points, existing techniques typically calculate the difference between the current cycle's data value and the historical cycle's data value directly. However, when the load is a frequency converter, its output frequency may fluctuate slightly near the power frequency, causing the waveform of the current cycle to be out of phase with the historical reference. This slight misalignment can generate large-amplitude spurious pulses during direct differential calculation, interfering with detection. To address this issue, this embodiment employs a neighborhood minimum value differential calculation method. For each sampling point, not only are historical points at the same location compared, but also its left and right adjacent historical points are compared, and the minimum difference is taken as the final current differential value.

[0026] Specifically, neighborhood difference calculation is performed on the current signals of corresponding sampling points in the current cycle and historical cycles to obtain the current difference sequence. As a concrete example, the current difference sequence can be obtained as follows: For the k-th sampling point in the current cycle, calculate the absolute value of the difference between the current amplitude of this sampling point and the current amplitude of the same index point and its adjacent sampling points in the historical cycle. The minimum of these three absolute values ​​is taken as the current difference value of this sampling point. The current difference value of the k-th sampling point in the current cycle can be expressed as: in, This represents the current difference value at the k-th sampling point in the current period. This represents the current amplitude at the k-th sampling point in the current period. This represents the current amplitude at the (k-1)th sampling point in the historical period. This represents the current amplitude at the k-th sampling point in the historical period. This represents the current amplitude at the (k+1)th sampling point in the historical period. Indicates the absolute value sign. This represents the function that takes the minimum value.

[0027] It should be noted that when the sampling point index k=1, the last sampling point of the previous period is used as the historical data of k-1; when the index k=N, the first sampling point of the next period is used or the data of the current point is directly copied as the historical data of k+1, where N is the number of sampling points in the current period.

[0028] The above method can be used to obtain the current difference value of each sampling point in the current period. The current difference values ​​of all sampling points in the current period are arranged in time sequence to obtain the current difference sequence.

[0029] The voltage difference sequence is obtained by performing neighborhood difference calculation on the voltage signals of corresponding sampling points in the current period and historical periods. As a specific example, the voltage difference sequence can be obtained as follows: For the k-th sampling point in the current period, calculate the absolute value of the difference between the voltage amplitude of this sampling point and the voltage amplitude of the same index point and its adjacent sampling points in the historical period. The minimum of these three absolute values ​​is taken as the voltage difference value of this sampling point. The voltage difference value of the k-th sampling point in the current period can be expressed as: in, This represents the voltage difference value at the k-th sampling point in the current period. This represents the voltage amplitude at the k-th sampling point in the current period. This represents the voltage amplitude at the (k-1)th sampling point in the historical period. This represents the voltage amplitude at the k-th sampling point in the historical period. This represents the voltage amplitude at the (k+1)th sampling point in the historical period. Indicates the absolute value sign. This represents the function that takes the minimum value.

[0030] Using the above method, the voltage difference value of each sampling point within the current period can be obtained. The voltage difference values ​​of all sampling points within the current period are then arranged in temporal order to obtain the voltage difference sequence.

[0031] If the waveform of the current signal or voltage signal in the current cycle only undergoes a slight left or right shift in phase, the minimum values ​​of the above three will approach zero and thus be suppressed; however, due to the fundamental change in shape, the waveform distortion of a real arc fault means that the minimum values ​​of the above three remain relatively large and are thus preserved.

[0032] After obtaining the current difference sequence and voltage difference sequence corresponding to the current cycle, local peak detection is performed on the current difference sequence and voltage difference sequence respectively to screen out the current change peak point and voltage change peak point.

[0033] Specifically, for the current difference sequence, the standard deviation of all elements in the current difference sequence is calculated, and the product of this standard deviation and the confidence coefficient is used as the preset current amplitude threshold. In this embodiment, the confidence coefficient is set to 5. Next, local peak detection is performed on the current difference sequence, and local maxima points with current amplitudes greater than the preset current amplitude threshold are used as current mutation candidate points. The method for detecting local maxima points is existing technology and will not be elaborated further here. All current mutation candidate points are sorted in descending order of amplitude, and the first preset number of current mutation candidate points in the sorted sequence are used as current mutation peak points. In this embodiment, the preset first number is set to 8. In specific applications, the implementer can set it according to the specific situation. It should be noted that if the total number of all current mutation candidate points is less than the preset first number, all current mutation candidate points are directly used as current mutation peak points. It should be noted that if the total number of current mutation peak points in the current cycle is 0, the current arc state is directly determined to be normal, and subsequent steps are not executed.

[0034] For a voltage differential sequence, the standard deviation of all elements in the sequence is calculated, and the product of this standard deviation and the confidence coefficient is used as a preset voltage amplitude threshold. Local peak detection is performed on the voltage differential sequence, and local maxima points with voltage amplitudes greater than the preset voltage amplitude threshold are designated as voltage abrupt change candidate points. All voltage abrupt change candidate points are sorted in descending order of amplitude, and the first preset number of candidate points are designated as voltage abrupt change peak points. Similarly, if the total number of all voltage abrupt change candidate points is less than the preset number, all candidate points are directly designated as voltage abrupt change peak points.

[0035] The above methods enable the screening of peak points for current and voltage mutations, compressing the originally complex high-dimensional data stream into two tiny feature point sets.

[0036] Considering the complex industrial environment where multiple inverter switching noise pulses and real arc pulses may coexist within a sampling window, simple time window overlap judgment is prone to mismatches, such as mismatching the current of pulse A with the voltage of pulse B. This embodiment will then utilize graph theory algorithms to globally search for the voltage-current pairing combination that best conforms to the time synchronization principle.

[0037] Specifically, a bipartite graph is constructed with the peak current change point as the left vertex and the peak voltage change point as the right vertex. To quantify the correlation between any two peak current and voltage changes, a time deviation weight is calculated. It should be noted that when the number of peak current and voltage changes is unequal, virtual vertices are added to the side with fewer vertices until the number of vertices on both sides is equal. The edge weights between the virtual vertices and all vertices on the other side are uniformly set to preset values.

[0038] The time interval between each current surge peak and each voltage surge peak is calculated. The smaller the time interval, the closer the current surge and voltage drop occur, and the greater the likelihood that they belong to the same electrical physical process; the larger the time interval, the lower the correlation between them. Based on these characteristics, if the time interval between the current surge peak and the voltage surge peak is greater than a preset maximum synchronization tolerance threshold, the edge weight between the corresponding current surge peak and the corresponding voltage surge peak is set to a preset value. This preset value needs to be set to a large value, indicating that the two points are electrically unrelated; in this embodiment, the preset value is [value missing]. In specific applications, implementers can set the parameters according to their specific circumstances. If the time interval between the peak current fluctuation point and the peak voltage fluctuation point is less than or equal to the preset maximum synchronization tolerance threshold, then the time interval is used as the edge weight between the corresponding peak current fluctuation point and the corresponding peak voltage fluctuation point. In this embodiment, the preset maximum synchronization tolerance threshold is the duration corresponding to 5 sampling points. In specific applications, implementers can set the parameters according to their specific circumstances.

[0039] The edge weights between all current and voltage surge peaks form a weight matrix. The number of rows in the weight matrix is ​​equal to the number of current surge peaks, and the number of columns in the weight matrix is ​​equal to the number of voltage surge peaks. The element values ​​in the weight matrix are the edge weights between current and voltage surge peaks.

[0040] After constructing the weight matrix, the next step is to match the peak points of current and voltage sudden changes. This involves establishing a one-to-one connection between the peak points of current and voltage sudden changes, minimizing the sum of the weights on all connections (i.e., the total time deviation), which solves the problem of minimum weight matching in a bipartite graph.

[0041] Specifically, this embodiment uses the Kuhn-Munkres algorithm to match the peak points of current and voltage surges based on the weight matrix. The MCU can complete the following iterative calculations in a very short time: (1) Initialize the vertex label: Assign a vertex label value to each vertex, which initially satisfies specific constraints.

[0042] (2) Finding augmenting paths: Starting from each unmatched current point, try to find an alternating path in the equal subgraph that connects to an unmatched voltage point.

[0043] (3) Adjusting the top label: If no augmenting path can be found under the current top label, adjust the top label value according to the minimum relaxation amount to introduce new edges into the equal subgraph and try to find it again.

[0044] (4) Augmenting Match: Once an augmenting path is found, a matching pair is added by flipping the matching state on the path.

[0045] (5) Iteration termination: Repeat the above process until no new matching pairs can be added, or all points have been matched.

[0046] The matching algorithm described above is existing technology and will not be elaborated further here. Ultimately, the optimal matching result is obtained, which contains multiple matching pairs. Each matching pair consists of a current surge peak and a voltage surge peak, representing highly synchronized electrical events in the microscopic time domain. The above matching method automatically weighs all possibilities, avoiding a deterioration in the overall matching effect due to greedy local selection (i.e., only considering the nearest point). For example, the algorithm might abandon a local match at a distance of 1 so that another point at a distance of 2 can also find a match, thus preventing the latter from being forced to match a distant point at a distance of 10, ultimately minimizing the global total deviation.

[0047] Step S3: Based on the overall distribution characteristics of the ratio of voltage amplitude to current amplitude in the matched pair, obtain the relative standard deviation of the ratio; based on the proportion of the number of unmatched current mutation peak points, obtain the proportion of isolated current mutations.

[0048] In this embodiment, multiple matching pairs are obtained in step S2. Next, these matching pairs will be analyzed to evaluate the ratio of voltage amplitude to current amplitude in the matching pairs. By utilizing the physical characteristic that the arc impedance is relatively stable, random disturbances will be further eliminated.

[0049] Before evaluating the ratio of voltage amplitude to current amplitude in a matched pair, it is necessary to eliminate numerical instability caused by small signals. If the amplitude of the voltage or current surge point in the matched pair is extremely small (close to the quantization noise floor of the ADC), the ratio obtained by dividing them will be a meaningless random number, interfering with subsequent statistical results.

[0050] The number of valid matched pairs that meet the amplitude validity criteria is counted from all matched pairs. These valid matched pairs are selected based on the following criteria: the amplitude of the voltage surge peak in the matched pair is greater than a preset effective voltage signal noise floor threshold, and the amplitude of the current surge peak in the matched pair is greater than a preset effective current signal noise floor threshold. The preset effective current signal noise floor threshold is typically set to 1%-2% of the current channel range, and the preset effective voltage signal noise floor threshold is typically set to 0.5%-1% of the voltage channel range. In this embodiment, the preset effective current signal noise floor threshold is 0.5A, and the preset effective voltage signal noise floor threshold is 2V. In specific applications, the implementer can set these values ​​according to the specific circumstances.

[0051] The following embodiment uses a valid matching pair as an example for explanation. Other valid matching pairs can be processed using the method provided in this embodiment.

[0052] Specifically, for any valid matching pair: the ratio of the amplitude of the voltage surge peak point to the amplitude of the current surge peak point in the valid matching pair is taken as the ratio of the voltage amplitude to the current amplitude in the valid matching pair. The ratio of voltage amplitude to current amplitude numerically reflects the equivalent impedance characteristics of the source that caused the surge.

[0053] Using the above methods, the ratio of voltage amplitude to current amplitude in each effective matched pair can be obtained.

[0054] For real series arc faults, since the physical state of the arc plasma is relatively stable during a single breakdown, its equivalent impedance should remain within a relatively fixed range for a very short time (milliseconds). Therefore, the calculated ratios of multiple voltage amplitudes to current amplitudes should be relatively close. For composite processes such as random disturbances in the power grid or simultaneous switching of multiple loads, due to the lack of a single physical constraint, the ratios of voltage and current changes are usually chaotic, and the calculated ratios of voltage amplitude to current amplitude will fluctuate drastically.

[0055] If the number of valid matching pairs is small, it indicates that too few valid synchronization events are captured in the current period, and the sample size is insufficient to support reliable variance calculation. In this case, the system directly abandons the current statistics, marks the current state as uncertain or maintains the state of the previous period, and does not output valid statistical features, effectively preventing misjudgments caused by single-point sporadic noise (variance of 0).

[0056] If the number of valid matching pairs is greater than or equal to the preset sample size, the mean and standard deviation of the ratio of voltage amplitude to current amplitude in all valid matching pairs are calculated respectively. In this embodiment, the preset sample size is 3. In specific applications, the implementer can set it according to the specific situation. If the number of valid matching pairs is less than the preset sample size, resulting in the inability to obtain the relative standard deviation of the ratio, the result of maintaining the arc state of the previous cycle is directly output.

[0057] Next, by combining the mean and standard deviation of the ratio of voltage amplitude to current amplitude for all valid matching, the relative standard deviation of the ratio is obtained. The relative standard deviation of the ratio can be determined as follows: the ratio between the calculated standard deviation and the mean is used as the relative standard deviation of the ratio. It should be noted that if the denominator of the formula for calculating the relative standard deviation of the ratio is 0, the sum of the denominator and the preset zero-prevention parameter is calculated, and this sum is used as the new denominator for calculation.

[0058] For unmatched current surge peaks, these points represent current surges that occur but for which no corresponding voltage drop can be found within the allowable time tolerance. These typically correspond to conducted interference noise injected by the frequency converter. This type of interference is characterized by injecting current pulses into the grid without causing a significant voltage drop. The number of unmatched current surge peaks is recorded as the number of isolated current points; dividing the number of isolated current points by the total number of current surge peaks yields the percentage of isolated current surges.

[0059] The relative standard deviation of the ratio and the proportion of isolated current mutations were determined using the above methods.

[0060] Step S4: Combine the relative standard deviation and the proportion of isolated current mutations to determine the current arc state.

[0061] In this embodiment, the relative standard deviation and the proportion of isolated current mutations were obtained in step S3. Next, these two will be combined to determine the current arc state.

[0062] Specifically, if the proportion of isolated current surges exceeds a preset threshold, it indicates that more than half of the current surges cannot find a corresponding time-synchronized voltage drop. This strongly suggests that the current signal is mainly generated by current-source loads such as frequency converters, and is therefore judged as conducted interference, not a fault.

[0063] If the proportion of isolated current surges is less than or equal to a preset proportion threshold, it indicates that most current surges have corresponding voltage responses, exhibiting preliminary fault characteristics, and further verification is required. The next level of verification is as follows: when the relative standard deviation is less than or equal to a preset deviation threshold, it indicates not only time synchronization but also highly consistent electrical response proportions, conforming to the impedance constraint characteristics of a series arc, and is thus judged as a suspected arc fault; when the relative standard deviation is greater than the preset deviation threshold, it indicates that although there is a voltage response, the response proportion fluctuates wildly, lacking physical regularity, and is thus judged as a random disturbance. In this embodiment, the preset proportion threshold is 0.5, and the preset deviation threshold is 0.3. In specific applications, implementers can set these values ​​according to specific circumstances.

[0064] After assessing the current arc state as described above, to prevent circuit breakers from tripping erroneously due to single transient power grid fluctuations (such as lightning surges), a time-window cumulative confirmation mechanism is introduced. A fault accumulation counter is maintained, with an initial value of 0. The counter is updated based on the arc state assessment results. If a suspected arc fault is identified, the counter is incremented by 1; if conducted interference, random disturbance, or insufficient sample size is identified, the counter is decayed. To avoid premature failure due to excessively rapid counter reset, a soft decay strategy is adopted, i.e., the counter is decremented by 1, but only to a minimum of 0, without negative values.

[0065] When the counter value exceeds the preset trip trigger threshold, indicating that multiple fault cycles have been continuously or densely confirmed, the system determines that the fault has indeed occurred and persists. At this point, the system executes a protection action: it outputs a high-level signal through the I / O port to drive a relay or circuit breaker to cut off the power supply circuit, locks the current state, and writes key data at the time of the fault occurrence (such as relative standard deviation, percentage of isolated current surges, and fault occurrence time) into a non-volatile memory (such as Flash or EEPROM) for subsequent fault cause analysis. In this embodiment, the preset trip trigger threshold is 3. In specific applications, the implementer can set it according to the specific circumstances.

[0066] Thus, the method described in this embodiment has enabled intelligent detection of arc faults.

[0067] This embodiment first filters out peak points of current and voltage mutations based on the amplitude distribution of electrical data from the current and historical periods. By matching current and voltage mutations and utilizing the overall distribution characteristics of the voltage-to-current amplitude ratio in the matched pair, the consistency of the response is evaluated. This eliminates interference from non-faulty electrical events such as random disturbances or nonlinear load switching on the detection results, enhancing the robustness of the detection method in complex electrical environments. By introducing the proportion of unmatched current mutations, current mutations that do not cause a synchronous voltage response can be identified, thereby accurately distinguishing real series arc faults. This avoids the misjudgment problems caused by traditional analysis based solely on current signals, improving the accuracy of arc fault detection results.

[0068] An embodiment of an intelligent arc fault detection device: See Figure 2 The diagram illustrates a structural block diagram of an intelligent arc fault detection device according to an embodiment of the present invention. The device may include an acquisition module, a matching module, an analysis module, and a judgment module.

[0069] The acquisition module is used to acquire electrical data from the monitoring point, including current signals and voltage signals. The matching module is used to filter current surge peak points and voltage surge peak points according to the amplitude distribution of electrical data in the current cycle and historical cycles, respectively; construct a bipartite graph with current surge peak points and voltage surge peak points as vertices, use the time interval between current surge peak points and voltage surge peak points as matching weights, and match current surge peak points and voltage surge peak points to obtain each matching pair; The analysis module is used to obtain the relative standard deviation of the ratio based on the overall distribution characteristics of the ratio of voltage amplitude to current amplitude in the matched pair; and to obtain the proportion of isolated current mutations based on the proportion of current mutation peaks in unmatched pairs. The judgment module is used to determine the current arc state by combining the relative standard deviation and the proportion of isolated current mutations.

[0070] It should be understood that Figure 2The structural block diagram and modules of the intelligent arc fault detection device shown can be implemented in various ways. For example, in some embodiments, the device and its modules can be implemented by hardware, software, or a combination of both. The hardware portion can be implemented using dedicated logic; the software portion can be stored in memory and executed by appropriate instructions, such as a microprocessor or dedicated hardware. Those skilled in the art will understand that the above-described methods and devices can be implemented using computer-executable instructions and / or included in processor control code, for example, such code provided on a carrier medium such as a disk, CD, or DVD-ROM, a programmable memory such as read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. The devices and modules of this specification can be implemented not only by hardware circuits such as very large-scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, or programmable hardware devices such as field-programmable gate arrays, programmable logic devices, etc., but also by software executed by various types of processors, or by a combination of the above-described hardware circuits and software (e.g., firmware).

[0071] For more details about the above modules, please refer to other parts of this manual; they will not be repeated here.

[0072] It should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An intelligent method for detecting electric arc faults, characterized in that, The method includes the following steps: Acquire electrical data from monitoring points, including current and voltage signals; Based on the amplitude distribution of electrical data in the current and historical periods, peak points of current and voltage sudden changes are selected respectively. A bipartite graph is constructed with the peak points of current and voltage sudden changes as vertices. The time interval between the peak points of current and voltage sudden changes is used as the matching weight to match the peak points of current and voltage sudden changes, and each matching pair is obtained. Based on the overall distribution characteristics of the ratio of voltage amplitude to current amplitude in the matching pair, the relative standard deviation of the ratio is obtained; based on the proportion of current surge peaks in unmatched pairs, the proportion of isolated current surges is obtained. The current arc state is determined by combining the relative standard deviation and the proportion of isolated current mutations.

2. The intelligent arc fault detection method according to claim 1, characterized in that, Screening for peak points of sudden current changes, including: The current difference sequence is obtained by performing neighborhood difference calculation on the current signals of corresponding sampling points in the current period and the historical period; Local peak detection is performed on the current differential sequence, and local maxima points whose amplitudes exceed a preset current amplitude threshold are selected as candidate points for current abrupt changes. All current mutation candidate points are sorted in descending order of amplitude, and the first preset number of current mutation candidate points are taken as the current mutation peak points.

3. The intelligent arc fault detection method according to claim 2, characterized in that, The step of performing neighborhood difference calculation on the current signals of corresponding sampling points in the current period and historical periods to obtain a current difference sequence includes: For any sampling point in the current cycle, calculate the absolute value of the difference between the current amplitude of the sampling point and the same index point in the historical cycle and its adjacent sampling points before and after it, and take the minimum value of the three absolute values ​​as the current difference value of the sampling point. The current difference values ​​of all sampling points within the current period constitute the current difference sequence.

4. The intelligent arc fault detection method according to claim 1, characterized in that, The process involves constructing a bipartite graph with current and voltage surge peaks as vertices, using the time interval between the current and voltage surge peaks as matching weights, and matching the current and voltage surge peaks to obtain matching pairs, including: Construct a bipartite graph with the peak point of current change as the left vertex and the peak point of voltage change as the right vertex; If the time interval between the current surge peak point and the voltage surge peak point is greater than the preset maximum synchronization tolerance threshold, then the edge weight between the corresponding current surge peak point and the corresponding voltage surge peak point is set to a preset value; if the time interval between the current surge peak point and the voltage surge peak point is less than or equal to the preset maximum synchronization tolerance threshold, then the time interval is used as the edge weight between the corresponding current surge peak point and the corresponding voltage surge peak point. The edge weights between all current surge peaks and voltage surge peaks constitute a weight matrix; Based on the weight matrix, the peak points of current sudden changes and the peak points of voltage sudden changes are matched to obtain several matching pairs.

5. The intelligent arc fault detection method according to claim 1, characterized in that, Obtaining the ratio of voltage amplitude to current amplitude in the matching pair includes: For any matching pair: The ratio of the amplitude of the voltage surge peak point to the amplitude of the current surge peak point in any matched pair is taken as the ratio of the voltage amplitude to the current amplitude in any matched pair.

6. The intelligent arc fault detection method according to claim 1, characterized in that, The relative standard deviation of the ratio is obtained based on the overall distribution characteristics of the ratio of voltage amplitude to current amplitude in the matching pair, including: The number of valid matching pairs that meet the amplitude validity criteria among the statistical matching pairs; If the number of valid matching pairs is greater than or equal to the preset sample size, then the mean and standard deviation of the ratios corresponding to all valid matching pairs are calculated respectively. By combining the mean and the standard deviation, the relative standard deviation of the ratio is obtained.

7. The intelligent arc fault detection method according to claim 1, characterized in that, The process of obtaining the proportion of isolated current surges based on the proportion of unmatched current surge peaks includes: The number of unmatched current abrupt peak points is recorded as the number of isolated current points. Divide the number of isolated current points by the total number of current surge peaks to obtain the percentage of isolated current surges.

8. The intelligent arc fault detection method according to claim 1, characterized in that, The determination of the current arc state by combining the relative standard deviation and the proportion of isolated current mutations includes: If the proportion of isolated current abrupt changes exceeds a preset threshold, it is determined to be conducted interference; If the proportion of isolated current mutations is less than or equal to a preset proportion threshold, and the relative standard deviation is less than or equal to a preset deviation threshold, then it is determined to be a suspected arc fault. If the proportion of isolated current mutations is less than or equal to a preset proportion threshold, and the relative standard deviation is greater than a preset deviation threshold, then it is determined to be a random disturbance.

9. The intelligent arc fault detection method according to claim 6, characterized in that, The process of combining the mean and the standard deviation to obtain the relative standard deviation of the ratio includes: The ratio between the standard deviation and the mean is used as the relative standard deviation of the ratio.

10. An intelligent arc fault detection device, the device being used to implement the method of claim 1, characterized in that, The device includes: The acquisition module is used to acquire electrical data from the monitoring point, including current signals and voltage signals; The matching module is used to filter current surge peak points and voltage surge peak points according to the amplitude distribution of electrical data in the current cycle and historical cycles, respectively; construct a bipartite graph with current surge peak points and voltage surge peak points as vertices, use the time interval between current surge peak points and voltage surge peak points as matching weights, and match current surge peak points and voltage surge peak points to obtain each matching pair; The analysis module is used to obtain the relative standard deviation of the ratio based on the overall distribution characteristics of the ratio of voltage amplitude to current amplitude in the matched pair; and to obtain the proportion of isolated current mutations based on the proportion of current mutation peaks in unmatched pairs. The judgment module is used to determine the current arc state by combining the relative standard deviation and the proportion of isolated current mutations.

Images