A series arc fault recognition method suitable for low voltage ac power distribution system

Abstract

The application discloses a series fault arc identification method suitable for a low-voltage alternating current distribution system, which is different from a traditional fault arc detection method, uses a zero sequence current as a detection signal, and determines whether an arc fault occurs or a line is in a normal state, wherein the feature quantity selected by the application accurately describes four stages of arc starting, arc burning, arc extinguishing and zero relaxation, and has a good fault arc detection effect on a low-voltage alternating current distribution system containing complex scenes such as multiple loads, multiple arc generation modes and multiple line topologies.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of electrical fault detection of low-voltage alternating current power distribution systems, and relates to a method for extracting features of zero sequence current from three dimensions of time domain, frequency domain and signal disorder degree based on physical processes of fault arcs, and applying threshold determination method to identify series fault arcs. BACKGROUND

[0002] According to the data published by the Fire Rescue Bureau of the Emergency Management Department, from 2012 to 2021, there were 1.324 million residential fires in the country, of which electrical fires accounted for 42.7%, and one of the main reasons for electrical fires was the occurrence of fault arcs in the line. According to the topology of the line, fault arcs can be divided into series fault arcs and parallel fault arcs. When parallel fault arcs occur, the line current will increase significantly, and traditional protection switches (such as miniature circuit breakers, residual current operated protective switching, fuses, etc.) can cut off the line in time; when series fault arcs occur, the loop current amplitude is lower than the current amplitude in the normal state of the line, and traditional protection switches cannot detect series fault arcs. As early as 1999, the United States has developed the UL1699 standard to guide the development of alternating current fault arc circuit interrupters (Arc Fault Circuit Interrupter, AFCI). In 2013, the International Electrotechnical Commission also developed an international standard (IEC62606-2013) for alternating current fault arc detection devices (Arc Fault Detection Device, AFDD). China also began to implement the national standards GB 14287.4-2014 (Electrical Fire Monitoring System Part 4: Fault Arc Detector) and GB / T31143-2014 (Arc Fault Protection Electrical Appliances (AFDD) General Requirements) in 2015.

[0003] In recent years, many researchers have conducted in-depth research on arc fault detection technology. However, due to the line load, arc condition, circuit topology and other reasons, arc fault has the characteristics of concealment, complexity and uncertainty. It is still difficult to accurately identify series arc fault in actual complex scenarios. Arc fault identification technology can be divided into three categories: arc fault detection methods based on physical phenomena, arc fault detection methods based on time-frequency domain feature analysis, and arc fault detection methods based on artificial intelligence models. When arc fault occurs, it is accompanied by obvious arc light, arc sound, heat radiation and electromagnetic radiation and other physical phenomena, so some experts and scholars use these physical phenomena to detect arc fault. However, this method needs to know the arc fault position in advance, and is easily limited and disturbed by sensor installation position and external environment, etc., and is not suitable for low-voltage alternating current power distribution systems. When arc fault occurs in the line, the line current and voltage signals have time domain and frequency domain characteristics, and such signals are easy to measure, so some experts and scholars extract the time-frequency domain features of voltage and current signals to detect arc fault. This method is simple and effective, and has high engineering realizability, but at the same time, some special loads have arc characteristics when the line is normal, which is easy to misoperate. With the application of artificial intelligence in various fields, some experts and scholars also use machine learning methods to build arc fault identification models, and have achieved good results, but this method requires a large amount of training data and preparation work, and the calculation complexity of some machine learning methods is high. At present, the cost of using this method to make arc fault detection module is high, and it is not suitable for wide application in actual scenarios.

[0004] Chinese patent CN 105425118A discloses a multi-information fusion arc fault detection method and device. This method extracts three time-frequency domain features from the current signal and uses a cerebellar model neural network to establish an arc fault detection model. However, this method is only effective for training set data, and whether it is effective for unknown nonlinear loads remains to be verified. Similarly, Chinese patent CN 110702966A discloses a fault arc detection method based on a probabilistic neural network. This method performs principal component analysis dimensionality reduction processing on the current signal, extracts relevant feature quantities, and uses a probabilistic neural network to build a fault arc detection model. However, this method has not been tested for stability and generalization, and whether it is still effective for arc fault under multiple special loads remains to be verified. In addition, although the neural network method for building arc fault detection model has good results under specific load types, this method has a large amount of calculation and low engineering realizability.

[0005] Chinese patent CN 113176479A discloses a series arc fault detection method suitable for low-voltage distribution network, which measures the voltage at the beginning and end of the line, analyzes the voltage difference signal in time and frequency domain, extracts relevant characteristic quantities, and uses threshold judgment method to detect arc fault. This method is effective for specific lines and simple in engineering, but does not consider the influence of line load, and the protection only extends to the beginning of the user side, and does not detect arc fault on the user side line.

[0006] In summary, although the current arc fault detection method has good detection effect under specific load type, it does not consider the influence of circuit topology and arc generation mode at the same time, and there are still misjudgment and missed judgment when applied in actual distribution network. Although load type and circuit topology affect the loop current characteristics, they have weak influence on zero sequence current. AC arc fault can be divided into four stages: arc initiation, arc burning, arc extinction and zero rest. The zero sequence current has large current fluctuation in the arc initiation and arc extinction stages. In addition, arc fault has strong instability, and the series into the line will cause the change of line parameters, which will inevitably affect the change of electrical parameters. The instability of arc burning aggravates the circuit noise, and the zero sequence current has obvious ripple, so the sudden change characteristics and ripple characteristics of AC arc fault can be used to detect arc fault. Although some special loads also have arc-like characteristics when the line is normal, the waveform of zero sequence current has certain rules under normal line state, on the contrary, when the line has arc fault, the instability of arc fault will affect the regularity of zero sequence current waveform, so these characteristics can also be used to avoid interference and identify arc fault. According to the change characteristics of zero sequence current waveform when arc fault occurs, the arc fault can be identified, which has wider application range, can effectively detect arc fault in complex scenes such as multiple load types, multiple circuit topologies and multiple arc generation modes, and the implementation process of the method is simple, the identification result is reliable, and the engineering realization is high. SUMMARY

[0007] The purpose of the present application is to provide a series arc fault identification method suitable for low-voltage AC distribution system, which uses zero sequence current as detection signal, and uses pulse factor, adjacent pulse peak interval time, Shannon entropy and harmonic amplitude as fault feature set to represent arc characteristics. Through experiment verification, the method can effectively detect arc fault in complex scenes such as multiple load types, multiple circuit topologies and multiple arc generation modes, and the calculation process is simple and easy to implement in engineering.

[0008] The technical scheme adopted by the present application is as follows:

[0009] Step 1: sample the zero sequence current signals of different branches in the low-voltage AC distribution system or the zero sequence current signals of the upper bus of each branch point by point at a sampling frequency f, and divide the signals into time windows with a length Ts Extract the corresponding detection signal x n (zero sequence current) under the current analysis period, where n = 1, 2, …, N;

[0010] Step 2: Calculate the pulse factor, adjacent pulse peak interval time, Shannon entropy, and harmonic amplitude sum of the zero sequence current in an analysis period;

[0011] Step 3: If the Shannon entropy is less than a preset threshold 1, execute step 4, otherwise execute step 5;

[0012] Step 4: If the pulse factor is less than a preset threshold 2, mark the line as normal, otherwise execute step 6;

[0013] Step 5: If the harmonic amplitude sum is less than a preset threshold 3, mark the line as suspected arc fault, otherwise execute step 7;

[0014] Step 6: If the adjacent pulse peak interval time satisfies the 10ms integer multiple relationship, mark the line as normal, otherwise mark the line as suspected arc fault;

[0015] Step 7: If the pulse factor is less than a preset threshold 2, mark the line as normal, otherwise mark the line as suspected arc fault;

[0016] Step 8: Count the number of periods in which suspected arc faults occur in consecutive q analysis periods, if the number of periods is greater than a preset threshold 4, determine that the line has an arc fault, otherwise determine that the line is in a normal state.

[0017] Preferably, the sampling frequency f and the time window length T s are 1MHz and 20ms respectively.

[0018] Preferably, the calculation method of the pulse factor is as follows:

[0019] S11: Take the absolute value of the zero sequence current in an analysis period to obtain the processing signal {|x1|, |x2|, …, |x N |};

[0020] S12: Determine the maximum value max(|x n |) of the processing signal;

[0021] S13: Determine the average value of the processing signal;

[0022] S14: The pulse factor is the ratio of the maximum value to the average value of the processing signal {|x1|, |x2|, …, |x N |}.

[0023] The calculation method of the pulse factor can be represented by the following formula:

[0024]

[0025] wherein, x n represents the zero sequence current signal in an analysis period, N is the number of zero sequence current sampling points in an analysis period, taking 1MHz sampling rate and 20ms time window as an example, N is equal to 20000.

[0026] Preferably, the calculation steps of the adjacent pulse peak interval time are as follows:

[0027] S21: taking the absolute value of the zero sequence current signal in an analysis period, and splitting the zero sequence current in an analysis period into two parts, i.e. {|x1|,|x2|,…··,|x N / 2 |} and {|x N / 2+1 |,|x N / 2+2 |,···,|x N |};

[0028] S22: calculating the position of the maximum value of the former part and the position of the maximum value of the latter part respectively;

[0029] S23: determining the position difference of the maximum values of the two parts of signals;

[0030] S24: the adjacent pulse peak interval time is the product of the position difference and the interval time between adjacent two sampling points. The interval time of adjacent pulse peaks can be expressed by the following formula:

[0031] IT=(p2-p1)×t i

[0032] wherein, p1 is the position of the maximum value in the sequence {|x1|,|x2|,···,|x N / 2 |}, p2 is the position of the maximum value in the sequence {|x N / 2+1 |,|x N / 2+2 |,···,|x N |}, t i is the time interval between adjacent two sampling points, i.e. 1 / f, taking 1MHz sampling rate as an example, t i =0.001ms.

[0033] Preferably, the calculation method of Shannon entropy is as follows:

[0034]

[0035] Preferably, the calculation steps of the sum of harmonic amplitudes are as follows:

[0036] S31: performing Fourier transform on the zero sequence current signal in an analysis period to obtain the amplitudes of each harmonic, i.e. {y 50 ,y100 ,y 150 ,y 200 ,···,y 500000};

[0037] S32: The sum of harmonic amplitudes is the sum of the amplitudes of all harmonics, i.e.

[0038]

[0039] Preferably, the determination of whether the interval between adjacent pulse spikes satisfies the 10ms integer multiple relationship can be achieved through the following steps:

[0040] S41: Calculate the remainder between the interval time of adjacent pulse spikes and 10ms;

[0041] S42: If the remainder is less than 0.03ms or greater than 9.96ms, it means that the interval between adjacent pulse spikes satisfies the 10ms integer multiple relationship; otherwise, it means that the interval between adjacent pulse spikes does not satisfy the 10ms integer multiple relationship.

[0042] Compared with the prior art, the present invention has the following beneficial technical effects:

[0043] This invention utilizes zero-sequence current as the detection signal, selecting electrical characteristics related to the arc initiation, arc burning, arc extinction, and zero-rest stages of a fault arc. This effectively reduces the influence of background noise on the fault arc's characteristic signals. A fault arc exhibits significant current abrupt changes during the arc initiation and extinction stages, and during the arc burning stage, it easily radiates energy into the external space, leading to a difference between the live wire current and the neutral wire current. This method is based on these unique physical characteristics of fault arcs; these physical processes specific to arc faults manifest as pulse spikes and ripples in the zero-sequence current. Furthermore, the zero-sequence current, through the mutual cancellation of the live wire current and the neutral wire current, physically shields the influence of load type on the arc characteristics. Therefore, this method is less affected by load type and circuit topology, making it applicable to more complex low-voltage AC power distribution systems.

[0044] This invention selects the pulse factor to characterize the pulse spikes generated by the fault arc during the arc initiation and extinction stages. It utilizes the instability characteristics of the fault arc to select the interval between adjacent pulse spikes to avoid interference from arc-like signals. Shannon entropy is selected to characterize the ripple characteristics generated by the fault arc during the arc combustion stage. The regularity of arc-like signals is used to extract harmonic amplitudes and avoid interference signals. These characteristics are closely related to the physical properties of the arc and the circuit characteristics, and the fault arc detection method established based on this has strong interpretability and applicability.

[0045] This invention determines the threshold of each characteristic quantity through data statistics, thereby realizing threshold judgment. The method is simple to calculate and easy to implement in engineering, and can provide a reference for the development of arc fault protection switches. Attached Figure Description

[0046] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0047] Figure 1 This is the fault arc testing experimental platform used in this invention;

[0048] Figure 2 This is the experimental circuit topology used in this invention;

[0049] Figure 3 These are the zero-sequence current signal waveforms under different line conditions described in this invention;

[0050] Figure 4 These are the zero-sequence current signal waveforms under different load types as described in this invention;

[0051] Figure 5 This is a flowchart of the fault arc detection algorithm described in this invention;

[0052] Figure 6 This is a confusion matrix representing the accuracy of fault arc detection under different circuit topologies described in this invention. Detailed Implementation

[0053] The following description, in conjunction with the accompanying drawings, illustrates exemplary embodiments of this application, including various details to aid understanding. These should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this application. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0054] To enable those skilled in the art to better understand the present invention, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are merely some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort should fall within the scope of protection of the present invention.

[0055] like Figure 1As shown, the zero sequence current signal of the miniature circuit breaker (MCB) installation position is collected at a 1MHz sampling rate and a 20ms time window, and the collected zero sequence current waveform is as shown in Figure 2 As shown.

[0056] It should be noted that the present application carries out series arc fault experiments under different experimental loads, different fault positions and different arc generation modes. Among them, the experimental loads include: resistors (46, 23, 15.3, 11.5Ω), electric kettles (1400W), ovens (1500W), fluorescent lamps (180W), induction cookers (1700W), microwave ovens (1300W), electric drills (200W), switching power supplies (400W), halogen lamps (400W) and dimmable lamps (1000W, 30, 60, 90, 120 degrees). The fault positions include the main circuit fault, the load branch 1 fault and the load branch 2 normal, the load branch 1 normal and the load branch 2 fault, which can be popularly equivalent to Figure 3 As shown, the four circuit topologies. The arc generation modes include arc generators and carbonized cables, which respectively simulate the fault arcs generated by poor contact and insulation aging.

[0057] Calculation of pulse factor: the pulse factor in an analysis period is calculated by the following formula, taking the zero sequence current waveform in Figure 2 As an example, the pulse factors under normal and fault conditions are 5.9 and 172 respectively.

[0058]

[0059] Where, x n represents the zero sequence current signal in an analysis period, and N is the number of zero sequence current sampling points in an analysis period. Taking a sampling rate of 1MHz and a time window of 20ms as an example, N is equal to 20000.

[0060] Calculation of adjacent pulse peak interval time: as shown in Figure 4 As shown, under the arc-like load, the zero sequence current waveform under normal condition also has pulse peaks, and at this time the corresponding pulse factor is larger. However, unlike the irregular pulse peaks generated by the fault arc, the pulse peaks in the normal operation of the circuit have regularity. Therefore, the adjacent pulse peak interval time is used to avoid the interference of arc-like signals, and its calculation method is shown in the following formula.

[0061] IT=(p2-p1)×t i

[0062] Where, p1 is the position of the maximum value in the sequence {|x1|,|x2|,···,|x N / 2 |}, and p2 is the position of the maximum value in the sequence {|x N / 2+1 |,|xN / 2+2 |,···,|x N The position of the maximum value in |}, t i It is the time interval between two adjacent sampling points, i.e., 1 / f, and its specific value here is 0.001ms.

[0063] Depend on Figure 4 It is known that when the dimmer lamp load is working normally, the interval between adjacent pulse spikes is an integer multiple of 10ms, while the pulse spikes generated by the fault arc do not meet this characteristic in terms of interval time. Therefore, the interval between adjacent pulse spikes can be used to avoid interference from arc-like signals.

[0064] Calculation of Shannon entropy: Fault arcs are unstable, causing irregular ripple characteristics in the zero-sequence current. Therefore, Shannon entropy is used to characterize the degree of disorder in the zero-sequence current signal. Its calculation method is as follows:

[0065]

[0066] Calculation of Harmonic Amplitude Sum: In practical low-voltage AC power distribution systems, some nonlinear loads generate harmonic currents, causing the zero-sequence current to exhibit regular ripple characteristics, which in turn increases Shannon entropy and leads to aliasing problems. Therefore, the harmonic amplitude sum is used to analyze the regular ripple contained in the zero-sequence current. Its calculation steps are as follows:

[0067] Perform a Fourier transform on the zero-sequence current to obtain the amplitude of each harmonic frequency, i.e., {y 50 ,y 100 ,y 150 ,y 200 ,···,y 500000};

[0068] The sum of the amplitudes of each harmonic frequency is obtained by summing the amplitudes of the harmonics.

[0069]

[0070] The harmonic amplitudes of three types of loads—resistors (excluding power frequency ripple and pulse spikes), dimmer lamps (including pulse spikes), and induction cookers (including power frequency ripple)—were calculated under different operating conditions. The results are shown in Table 1. It can be seen that the harmonic amplitudes of loads containing power frequency ripple are larger. Therefore, this characteristic can effectively distinguish loads containing power frequency ripple.

[0071] Table 1 Harmonic amplitudes under different experimental conditions and

[0072] Resistance Dimmer lamp IH cooker Normal <0.03 <0.03 >0.29 Arc fault <0.05 <0.06 <0.28

[0073] After determining the four characteristic quantities, it can be done through Figure 5 The algorithm shown detects fault arcs, and the specific steps are as follows:

[0074] Step 1: If the Shannon entropy is less than a preset threshold 1, execute S2, otherwise execute Step 3;

[0075] Step 2: If the pulse factor is less than a preset threshold 2, mark the line as normal, otherwise execute Step 4;

[0076] Step 3: If the harmonic amplitude sum is less than a preset threshold 3, mark the line as suspected arc fault, otherwise execute Step 5;

[0077] Step 4: If the adjacent pulse peak interval time satisfies the 10ms integer multiple relationship, mark the line as normal, otherwise mark the line as suspected arc fault;

[0078] Step 5: If the pulse factor is less than a preset threshold 2, mark the line as normal, otherwise mark the line as suspected arc fault;

[0079] Step 6: Count the number of periods in which suspected arc faults occur in the consecutive q analysis periods, if the number of periods is greater than a preset threshold 4, determine that the line has an arc fault, otherwise determine that the line is in a normal state.

[0080] When determining whether the adjacent pulse peak interval time satisfies the 10ms integer multiple relationship, first, the remainder between the interval time and 10ms needs to be calculated, if the remainder is less than 0.03ms or greater than 9.97ms, it means that the adjacent pulse peak interval time satisfies the 10ms integer multiple relationship, otherwise it means that the adjacent pulse peak interval time does not satisfy the 10ms integer multiple relationship.

[0081] Through analyzing 119000 groups of data under different experimental conditions (including 80000 groups of normal data and 39000 groups of fault data), the four thresholds in the formula (1) are determined as shown in Table 2. Figure 5

[0082] Table 2 Threshold values in the fault arc detection algorithm

[0083] Pre-set threshold 1 Pre-set threshold 2 Pre-set threshold 3 Pre-set threshold 4 Value reference 5.88 47.38 0.14 4

[0084] The international standard IEC 62606 and the national standard GB / T 31143 both require the fault arc protection switch to break the arc fault under 63A current within 120ms, so the value of q is 6.

[0085] The above algorithm is used to detect arc faults in complex scenes containing 16 kinds of loads, 4 kinds of circuit topologies and 2 kinds of arc generation modes, and the results are shown in Table 3 and Table 4. Figure 6 Figure 6 ​​It can be seen that the detection accuracy of the method according to the application is 99.6% (7654 / 7680), 99.4% (4135 / 4160), 99.6% (2072 / 2080) and 99.3% (3178 / 3200) under the circuit topologies A, B, C and D respectively. Therefore, the method according to the application can provide a reference for the development of arc fault protection switches.

[0086] Table 3 Arc detection accuracy under circuit topology A

[0087] Load type Normal (%) Arc generator (%) Carbonized cable (%) Resistance (5, 10, 15, 20 A) 100(640 / 640) 99.7(638 / 640) 99.5(637 / 640) IH cooker 100(160 / 160) 96.9(155 / 160) 100(160 / 160) Dimmer lamp (30, 60, 90, 120 degrees) 100(640 / 640) 99.8(639 / 640) 99.5(637 / 640) Switching power supply 100(160 / 160) 100(160 / 160) 100(160 / 160) Oven 100(160 / 160) 99.4(159 / 160) 100(160 / 160) Halogen lamp 100(160 / 160) 97.5(156 / 160) 100(160 / 160) Hot water kettle 100(160 / 160) 97.5(156 / 160) 100(160 / 160) Drill 100(160 / 160) 100(160 / 160) 100(160 / 160) Microwave oven 100(160 / 160) 100(160 / 160) 100(160 / 160) Fluorescent lamp 100(160 / 160) 99.4(159 / 160) 98.8(158 / 160)

[0088] The above detailed description does not constitute a limitation on the protection scope of the present application. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent replacement and improvement made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for identifying series fault arcs in low-voltage AC power distribution systems, characterized in that: Step 1: In a low-voltage AC power distribution system, sample the zero-sequence current signal of different branches point by point at frequency f, or sample the zero-sequence current signal of the upstream bus of each branch point by point, and sample according to the time window length T. s Extract the detection signal x corresponding to the current analysis period respectively. n Let n be the zero-sequence current, where n = 1, 2, ..., N; Step 2: Calculate the pulse factor of the zero-sequence current, the interval between adjacent pulse spikes, the Shannon entropy, and the harmonic amplitude within one analysis cycle; The calculation steps for the pulse factor are as follows: S11: Take the absolute value of the zero-sequence current within one analysis cycle to obtain the processed signal {|x1|,|x2|,…,|xN|}; S12: Determine the maximum value of the processed signal, max(|x n |); S13: Determine the average value of the processed signal; S14: The impulse factor is used to process the signal {|x1|,|x2|,…,|x...} N The ratio of the maximum value to the average value of |}; The impulse factor can be expressed by the following formula: Where N is the number of zero-sequence current sampling points in one analysis period; The calculation steps for the interval between adjacent pulse spikes are as follows: S21: Take the absolute value of the zero-sequence current signal within one analysis period, and split the zero-sequence current within one analysis period into two parts, namely {|x1|,|x2|,…,|x N / 2 |} and {|x N / 2+1 |,|x N / 2+2 |,…,|x N |}; S22: Calculate the positions of the maximum values ​​in the first part and the second part respectively; S23: Determine the positional difference between the maximum values ​​of the two parts of the signal; S24: The interval between adjacent pulse spikes is the product of the position difference and the interval between two adjacent sampling points; The interval between adjacent pulse spikes can be expressed by the following formula: IT=(p2-p1)×t i Where p1 is the sequence {|x1|,|x2|,…,|x...} N / 2 The position of the maximum value in |}, p2 is the sequence {|x}. N / 2+1 |,|x N / 2+2 |,…,|x N The position of the maximum value in |}, t i It is the time interval between two adjacent sampling points, i.e., 1 / f; Step 3: If the Shannon entropy is less than the preset threshold 1, proceed to step 4; otherwise, proceed to step 5. Step 4: If the pulse factor is less than the preset threshold 2, mark the line as normal; otherwise, proceed to step 6. Step 5: If the sum of harmonic amplitudes is less than the preset threshold 3, mark the line as having a suspected arc fault; otherwise, proceed to step 7. Step 6: If the interval between adjacent pulse spikes is an integer multiple of 10ms, the line is marked as normal; otherwise, the line is marked as having a suspected arc fault. The determination of whether the interval between adjacent pulse spikes satisfies the 10ms integer multiple relationship can be achieved through the following steps: S41: Calculate the remainder between the interval time of adjacent pulse spikes and 10ms; S42: If the remainder is less than 0.03ms or greater than 9.97ms, it means that the interval between adjacent pulse spikes satisfies the 10ms integer multiple relationship; otherwise, it means that the interval between adjacent pulse spikes does not satisfy the 10ms integer multiple relationship. Step 7: If the pulse factor is less than the preset threshold 2, the line is marked as normal; otherwise, the line is marked as having a suspected arc fault. Step 8: Count the number of cycles in which a suspected arc fault occurs within q consecutive analysis cycles. If the number of cycles is greater than the preset threshold of 4, the line is determined to have an arc fault; otherwise, the line is determined to be in normal condition.

2. The method for identifying series fault arcs in low-voltage AC power distribution systems according to claim 1, characterized in that, Signal sampling frequency f and time window length T s The frequencies are 1MHz and 20ms, respectively.

3. The method for identifying series fault arcs in low-voltage AC power distribution systems according to claim 1, characterized in that, The formula for calculating the Shannon entropy is as follows:

4. The method for identifying series fault arcs in low-voltage AC power distribution systems according to claim 1, characterized in that, The calculation steps for the sum of harmonic amplitudes are as follows: S31: Perform a Fourier transform on the zero-sequence current signal within one analysis period to obtain the amplitude of each harmonic, i.e., {y 50 ,y 100 ,y 150 ,y 200 ,…,y 500000 }; S32: The sum of harmonic amplitudes is the sum of the amplitudes of all harmonics; The sum of harmonic amplitudes can be expressed by the following formula:

Images