Arc fault detection method and device based on voltage time difference variation

Abstract

The present invention discloses an arc fault detection method and device based on voltage time differential variation, belonging to the field of low-voltage distribution network fault protection technology, including: treating the arc fault point as an arc voltage source, using the arc voltage change rate and the arc voltage effective value, establishing a detection feature description of the fault waveform generated by the arc voltage source, and the fault waveform presents a square wave-like waveform; sampling the adjacent cycle voltages of the load-side monitoring point, and performing feature recognition on the terminal voltage difference between adjacent cycles based on the detection feature to identify the first fault waveform; after the first fault waveform occurs, continuously identifying and counting the arc faults, using the difference between the non-fault waveform before the fault occurs and the waveform to be identified to realize the identification of continuous arc faults, and confirm the occurrence of the arc fault. The detection method of the present invention has good reliability, clear physical meaning, high detection sensitivity, is not affected by line parameters, only requires a single monitoring point to achieve detection, and is lower in cost.

Description

Technical Field

[0001] The present invention relates to the technical field of low-voltage distribution network fault protection, and in particular to an arc fault detection method and device based on voltage time differential variation. Background Art

[0002] Arc faults are a major cause of electrical fires in low-voltage distribution lines. Arc faults include series, parallel, and grounding types. Series arc faults are the most difficult to detect because the fault current is similar in magnitude to the load current, and the fault current characteristics are easily affected by the load type. Currently, series arc faults are mainly detected by identifying the waveform characteristics of the fault current. When an arc fault occurs, the main characteristics of the current include increased high-frequency noise, the appearance of a current zero-rest zone, a higher current rise rate, increased current harmonic content, and asymmetry between positive and negative half-cycles. Arc fault detection products on the market all utilize one or more of the above characteristics for detection, and a large number of studies at home and abroad have also used mathematical tools to extract one or more of the above characteristics in the time domain or transform domain for detection.

[0003] Compared to current characteristics, the arc voltage at the fault point is less affected by load type. Therefore, identifying the presence of the fault arc voltage can be used to characterize the occurrence of an arc fault. The arc fault point can be considered an arc voltage source. Because the load impedance is much greater than the line impedance of the loop, the voltage downstream of the fault point theoretically provides nearly complete information about the arc fault.

[0004] However, several existing arc fault voltage detection methods all have different problems. First, they directly or indirectly identify the presence of rising / falling edges of the arc voltage at the fault point for fault detection. When an arc fault occurs, the voltage at the arc fault point approximates a square wave shape, and the rising / falling edges are only local features of the square wave. While the rising / falling edges can cause load-side voltage distortion, in practice, nonlinear voltage drops can also easily cause load-side voltage distortion. Therefore, theoretically, these methods cannot fully characterize the presence of an arc fault. Second, they use the random characteristics of the load-side voltage caused by the arc fault for fault detection. Although these random characteristics exist, their strength cannot be controlled and threshold setting is difficult. Third, arc fault detection is achieved by identifying line voltage drops or the waveform of the line voltage drop and arc fault voltage superimposed to detect arc faults between two monitoring points. This method has high detection sensitivity and reliability, but the dual monitoring point requirements and communication requirements make this method expensive in terms of engineering and economic costs. Therefore, existing methods have failed to find appropriate fault characteristics and judgment thresholds for arc fault detection. Summary of the Invention

[0005] The present invention aims to provide an improved arc fault detection method to solve the problems of fault feature selection and criterion threshold setting in the current current detection method for identifying the existence of series arc faults.

[0006] To achieve the above object, the technical solution of the present invention is:

[0007] An arc fault detection method based on voltage time differential change includes: step S1, equating the arc fault point to an arc voltage source, using the arc voltage change rate and the arc voltage effective value to establish a detection feature description of the fault waveform generated by the arc voltage source, wherein the fault waveform presents a square wave-like waveform; step S2, sampling adjacent cycle voltages of a load-side monitoring point, performing feature recognition on the terminal voltage difference between adjacent cycles based on the detection feature, and identifying the first fault waveform; step S3, after the first fault waveform occurs, continuously identifying and counting the arc faults, using the difference between the non-fault waveform before the fault occurs and the waveform to be identified to realize the recognition of continuous arc faults, and confirming the occurrence of the arc fault.

[0008] In the above arc fault detection method based on voltage time differential change, in step S1, the arc voltage change rate is approximately expressed by the following formula:

[0009]

[0010] Where, Indicates the approximate value of the arc voltage change rate, U m represents the peak voltage, ω represents the power frequency angular frequency, T s Indicates the sampling time interval, U rh Indicates the arc peak voltage at the time of arc starting, U arc Indicates the arc burning voltage, L indicates the number of sampling points in the transition zone;

[0011] Arc voltage RMS The formula is:

[0012]

[0013] Where u arc (n) represents the arc voltage sampling point, N represents the number of sampling points per cycle, M represents the number of sampling points corresponding to the arc voltage, U arc Indicates the arc burning voltage, Indicates the arc starting and arc extinction voltage before the arc enters the arcing stage.

[0014] In the above arc fault detection method based on voltage time differential variation, in step S2, the expression of the terminal voltage differential is:

[0015] Terminal voltage difference between non-fault cycles The expression is:

[0016]

[0017] Where i1(t) is the first current cycle used for differential calculation in the non-fault state, i2(t) is the second current cycle used for differential calculation in the non-fault state, R is the sum of the line resistance before and after the fault point, L is the sum of the system impedance and the line inductance before and after the fault point, and ζ1(t) is the sinusoidal fluctuation in the differential waveform of two adjacent cycles of the terminal voltage in the non-fault state.

[0018] The terminal voltage difference between the non-fault cycle and the fault cycle The expression is:

[0019]

[0020] Where i arc (t) is the first fault current cycle after the fault occurs, i(t) is the last non-fault current cycle before the fault occurs, u arc (t) is the arc voltage at the fault point, ζ2(t) is the differential waveform of the terminal voltage excluding the arc voltage at the fault point after the non-fault state is converted to the fault state, that is, the sinusoidal fluctuation in the differential waveform;

[0021] Terminal voltage difference between fault cycles The expression is:

[0022]

[0023] Where i arc1 (t) is the previous current cycle used for differential operation under fault condition, i arc2 (t) is the current cycle used for differential operation under fault conditions, u arc1 (t) is the arc voltage at the fault point corresponding to the previous current cycle, u arc2 (t) is the arc voltage at the fault point corresponding to the subsequent current cycle, and ζ3(t) is the sinusoidal fluctuation in the differential waveform of the two adjacent cycles of the terminal voltage under the fault state.

[0024] In the above-mentioned arc fault detection method based on voltage time differential change, before the arc fault occurs, ζ1(t) shows a small fluctuation; when the arc fault occurs, a square wave waveform appears, and the first fault waveform is identified from ζ2(t); after the fault occurs, the differences between adjacent fault cycles show different characteristics due to the randomness of arc burning, and ζ3(t) will show different characteristics.

[0025] In the above-mentioned arc fault detection method based on voltage time differential change, step S3, "using the difference between the non-fault waveform before the fault occurs and the waveform to be identified to realize the identification of continuous arc faults", specifically includes: before the first fault arc occurs, using the adjacent wave differential inspection mode for inspection, taking k consecutive waveforms as a group, and performing time differentiation with the next adjacent group of k waveforms, with half-wave as the unit, to obtain 2k voltage differential △u waveforms; if the characteristics of the 2k waveforms are within the limit, it indicates that the waveform is a non-fault cycle, and then the next group of k waveforms is used to continue identification; if any of the voltage differential △u waveform characteristics exceeds the limit, an arc fault is suspected to have occurred. After the first fault cycle is identified, the first waveform of the waveform group before the suspected fault occurs is used as a reference for the detection mode, and the suspected fault waveforms are differentially identified and counted one by one using this waveform; when m consecutive full-cycle non-fault waveforms are identified, it indicates that the fault has disappeared, and the adjacent wave differential inspection mode is restored at this time.

[0026] In the above arc fault detection method based on voltage time differential variation, in the adjacent wave differential inspection mode, an arc fault is suspected to have occurred when one of the following detection conditions is met:

[0027]

[0028] Where, represents the voltage change rate of the i-th voltage difference △u, represents the maximum value of the voltage change rate in the i-th voltage difference △u, It represents the effective value of the i-th voltage difference △u, and α1 and β1 are the set detection thresholds.

[0029] In the above arc fault detection method based on voltage time differential variation, after entering the reference waveform differential detection mode, the non-fault waveform before the fault occurs is first used as the reference waveform u ref , reference waveform u ref and the waveform u to be identified x Perform phase synchronization and then perform differential identification. The criteria for fault identification are as follows:

[0030]

[0031] Where, α up and α down They represent the upper and lower detection thresholds of the parameter wave differential detection mode, which are used to characterize the arc starting / extinction process; β2 represents the half-wave effective value setting threshold, which is used to characterize the existence of the arc voltage drop; Represents the reference waveform and the waveform to be identified u x The rate of change of voltage of the voltage differential, Indicates half-wave effective value.

[0032] In the above arc fault detection method based on voltage time differential variation, the parameter wave differential detection mode also includes:

[0033] For reference waveform u ref and the waveform u to be identified x , perform N-point fast Fourier transform respectively to obtain the frequency domain phasor of each power frequency component:

[0034] a+bj={FFT[u ref (n)]} 工频

[0035] c+dj={FFT[u x (n)]} 工频

[0036] Where a and b represent the reference waveform u ref The real and imaginary parts of the power frequency phasor component, c and d represent the waveform u to be identified. x Real and imaginary parts of power frequency phasor components;

[0037] The reference waveform u is obtained based on the frequency domain phasor of the power frequency component ref and the waveform u to be identified x Phase difference △θ:

[0038] Δθ=[arctan(b / a)-arctan(d / c)]

[0039] The reference waveform u ref To the waveform u to be identified x The formula for the number of aligned data shifts △N is:

[0040] ΔN=[Δθ / 2π·N]

[0041] Where N represents the reference waveform u ref and the waveform u to be identified x The number of sampling points per cycle, "[……]" means rounding the calculation result;

[0042] The reference waveform u ref Circularly shift right by △N bits to eliminate the phase difference:

[0043] u′ ref (n) = u ref ((n-ΔN)) N R N (n)

[0044] Where u ref ((n)) N Indicates u ref (n) Periodic extension sequence with N as period, R N(n) represents a rectangular window sequence of length N.

[0045] In the above-mentioned arc fault detection method based on voltage time differential variation, step S2 further includes, before performing "characteristic identification of terminal voltage differences between adjacent cycles based on detection characteristics", performing envelope filtering on the differential voltage;

[0046] The envelope filtering of the differential voltage specifically includes: performing low-pass filtering on the differential voltage to filter out high-frequency glitch noise, using NP sampling points as local envelope identification windows, calculating the amplitude and position of the maximum and minimum points in each local envelope identification window, using a data interpolation algorithm to generate the upper and lower envelopes of the waveform, and taking the average of the upper and lower envelopes as the envelope filtering result.

[0047] The present invention also relates to an arc fault detection device, which runs the above-mentioned arc fault detection method based on voltage time differential change.

[0048] Beneficial effects: The arc fault detection method of the present invention has better reliability, clearer physical meaning, high detection sensitivity, and is not affected by line parameters. Compared with the existing technology, the method of the present invention only requires a single monitoring point to achieve detection, and its cost is lower.

[0049] In order to make the above features and advantages of the present invention more clearly understood, embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 Schematic diagram of arc current / voltage waveform under non-switching and switching power supply loads.

[0051] Figure 2 Schematic diagram of the flow of the arc fault detection method based on voltage time differential variation of the present invention.

[0052] Figure 3 Schematic diagram of arc voltage waveform and voltage change rate.

[0053] Figure 4 This is the equivalent circuit diagram of a series arc fault.

[0054] Figure 5 This is the principle diagram of dual-mode time difference fault detection.

[0055] Figure 6 This is a logic diagram of time waveform difference.

[0056] Figure 7 This is the envelope filtering result of the arc voltage taking a computer load as an example.

[0057] Figure 8The waveform diagram shows the arc fault feature identification results based on different loads.

[0058] Figure 9 The waveform diagram of the line current and terminal voltage when three loads, namely electric stove, vacuum cleaner and microwave oven, are switched on and off arbitrarily.

[0059] Figure 10 This is the fault feature recognition result diagram of the recorded waveform.

[0060] Figure 11 Figure 2 is the actual arc and fault feature identification result of resistive load.

[0061] Figure 12 This is the fault identification result diagram when the three loads of electric stove, vacuum cleaner and microwave oven are running at the same time.

[0062] Figure 13 This is a diagram showing the actual differential pattern of the waveform recorded using the method of the present invention and the final recognition result. DETAILED DESCRIPTION

[0063] To make the purpose and technical solutions of the embodiments of the present invention more clear, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the described embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0064] According to the mechanism of low-voltage arc generation, the voltage at the arc fault point is approximately in the shape of a square wave. For the sharing of the square wave shape, please refer to the existing document 1 "Wang Wei, Xu Bingyin, Sun Zhongyu, Liang Dong. Voltage differential detection method for arc fault in low-voltage distribution network [J]. Proceedings of the CSEE. 2023, 43(22): 8674-8688.", which will not be repeated in this invention. As described in the existing document 2 "Wang Wei, Xu Bingyin, Zou Guofeng, Sun Zhongyu, Liang Dong. Low-voltage AC series arc fault detection method based on voltage characteristic energy [J]. Power System Protection and Control, 2023, 51(19): 81-93.", the current characteristics of series arc faults are greatly affected by the load type. If the load type is classified based on the arc voltage waveform alone, it can be divided into two categories: non-switching power supply load and switching power supply load.

[0065] Figure 1 The following is a schematic diagram of arc current / voltage waveforms under non-switching and switching power supply loads. Part (a) takes a vacuum cleaner, which is a non-switching power supply load, as an example, and part (b) takes a computer, which is a switching power supply load, as an example, to show the arc current / voltage waveforms under arc faults. Figure 1It can be seen that for current, the characteristics of arc fault are greatly affected by the load type. However, for voltage characteristics, regardless of the load type, the voltage waveform at the fault point is approximately square wave. Figure 1 The arc voltage under computer load in part (b) is a square wave with high-frequency oscillation characteristics, and its overall low-frequency profile is still a square wave shape.

[0066] In the above-mentioned low-voltage series arc fault, the voltage waveform at the fault point is a square wave-like shape. Therefore, based on this waveform feature, the arc fault detection method based on voltage time differential change of the present invention can be constructed. Figure 2 FIG. 1 is a flow chart of an arc fault detection method based on voltage time differential variation according to the present invention, as shown in FIG. Figure 2 As shown, the arc fault detection method of the present invention includes the following steps.

[0067] In step S1, the arc fault point is equivalent to an arc voltage source, and the arc voltage change rate and the arc voltage effective value are used to establish a detection feature description of the fault waveform generated by the arc voltage source, wherein the fault waveform presents a square wave-like waveform.

[0068] The arc voltage change rate is defined as:

[0069] Δu rate (n)=|u(n+1)-u(n)| 0≤n≤N-1 Formula (1)

[0070] In the above formula (1), Δu rate (n) represents the voltage change rate of adjacent voltage sampling points, u(n) represents the voltage at the nth sampling point, and N represents the number of sampling points per cycle.

[0071] The formula for setting the power supply voltage is as follows:

[0072] u s (n) = U m ·sin(ω·nT s ) Formula (2)

[0073] In the above formula (2), u s (n) represents the power supply voltage, U m represents the voltage peak, ω represents the power frequency angular frequency, T s Indicates the sampling time interval.

[0074] The arc voltage waveform consists of the arc starting zone, transition zone, arc burning zone, and arc extinction zone. According to the above formula (1), the arc voltage change rate is approximated by the following formula:

[0075]

[0076] In the above formula (3), Indicates the approximate value of the arc voltage change rate, U rh Indicates the arc peak voltage at the time of arc starting, U arc Indicates the arc burning voltage, and L indicates the number of sampling points in the transition zone.

[0077] Arc burning voltage U arc The formula is as follows:

[0078] Uarc=Uc+Ua+Earc×larc Formula (4)

[0079] In the above formula (4), U c and U a Represents cathode voltage drop and anode voltage drop, E arc represents the arc column electric field strength, l arc In low-voltage power distribution systems, since the arc gap of millimeters is sufficient to extinguish the arc, the arc column voltage drop can be ignored and only the cathode and anode voltage drops are considered. Therefore, U arc ≈U c +U a , that is, the arc voltage drop during arcing is considered to be approximately constant.

[0080] Figure 3 The diagram of arc voltage waveform and voltage change rate is shown in Figure 1. Part (a) shows the arc voltage and change rate with arcing / arc extinction peaks, and part (b) shows the arc voltage and change rate without arcing / arc extinction peaks. Figure 3 In formula (3), the voltage change rate in the arc starting area / arc extinction area is equal to the numerical differential of the power supply voltage.

[0081] Taking resistive load as an example, when the sampling frequency is 10kHz, that is, the sampling time interval T s is 0.1ms, if the arc starting time is 3ms, U m The arc voltage is 310V and the arc burning voltage is 15V. The arc voltage change rate in the arc starting zone / arc extinction zone ranges from 5.7 to 9.7V. The voltage change rate in the arc burning zone is approximately zero. The voltage change rate in the transition zone depends on the transition zone time and the number of sampling points in the transition zone. The transition zone is approximately regarded as a linear change. If the transition time is 300us, the arc voltage change rate in the transition zone reaches 60V.

[0082] Another indication of the existence of arc voltage is the effective value of arc voltage. The formula is:

[0083]

[0084] In the above formula (5), u arc (n) represents the arc voltage sampling point, N represents the number of sampling points per cycle, M represents the number of sampling points corresponding to the arc voltage, Uarc Indicates the arc burning voltage, Indicates the arc starting and arc extinction voltage before the arc enters the arcing stage.

[0085] To simplify the analysis, only the arc burning voltage is considered, and the arc burning voltage U arc Assuming the voltage is 15V and the arc lasts only 2ms per half cycle, the effective value of the arc voltage of half wave exceeds 6V.

[0086] Step S2: sampling adjacent cycle voltages of the load-side monitoring point, performing feature recognition on the terminal voltage difference between adjacent cycles based on the detection characteristics, and identifying the first fault waveform.

[0087] Figure 4 This is the equivalent circuit diagram of a series arc fault. Figure 4 in, u s (t) is the AC power supply, u arc (t) is the arc voltage source equivalent to the arc fault point, u M (t) is the voltage of the monitoring point upstream of the fault point, u0(t) is the terminal voltage, R L is the resistance part of the load, L L is the inductance part of the load, L s is the system impedance, R1 and R2 are the line resistances before and after the fault point, L1 and L2 are the line inductances before and after the fault point, R is the total line resistance, R=R1+R2, L is the total line inductance, L=L s +L1+L2.

[0088] Reference Figure 4 According to Kirchhoff's voltage theorem, when a series arc fault occurs, the formula for the terminal voltage u0(t) is:

[0089]

[0090] In the above formula (6), u s (t) represents the AC power supply voltage, i arc (t) represents the fault current, u arc (t) represents the voltage of the equivalent arc voltage source at the fault point.

[0091] When no arc fault occurs, the formula for the terminal voltage u0(t) is:

[0092]

[0093] In the above formula (7), i(t) is the current at the monitoring point on the load side.

[0094] According to the above equations (6) and (7), three terminal voltage differential expressions can be obtained:

[0095] Terminal voltage difference between non-fault cycles The expression is:

[0096]

[0097] In the above formula (8), i1(t) is the front current cycle used for differential operation in the non-fault state, i2(t) is the rear current cycle used for differential operation in the non-fault state, and ζ1(t) is the sinusoidal fluctuation in the differential waveform of two adjacent cycles of the terminal voltage in the non-fault state.

[0098] The terminal voltage difference between the non-fault cycle and the fault cycle The expression is:

[0099]

[0100] In the above formula (9), i arc (t) is the first fault current cycle after the fault occurs, i(t) is the last non-fault current cycle before the fault occurs, u arc (t) is the arc voltage at the fault point, ζ2(t) is the differential waveform of the terminal voltage excluding the arc voltage at the fault point after the non-fault state is converted to the fault state, that is, the sinusoidal fluctuation in the differential waveform.

[0101] Terminal voltage difference between fault cycles The expression is:

[0102]

[0103] In the above formula (10), i arc1 (t) is the previous current cycle used for differential operation under fault condition, i arc2 (t) is the current cycle used for differential operation under fault conditions, u arc1 (t) is the arc voltage at the fault point corresponding to the previous current cycle, u arc2 (t) is the arc voltage at the fault point corresponding to the subsequent current cycle, and ζ3(t) is the sinusoidal fluctuation in the differential waveform of the two adjacent cycles of the terminal voltage under the fault state.

[0104] The above ζ1(t), ζ2(t), and ζ3(t) reflect the effect of current changes on the time differential voltage. Because the current is mainly composed of power frequency components, ζ1(t), ζ2(t), and ζ3(t) appear as sinusoidal fluctuations in the differential signal.

[0105] Before a fault occurs, the differential waveform of the terminal voltages of adjacent cycles exhibits small fluctuations, i.e., ζ1(t). When a fault occurs, a square-wave-like waveform appears, indicating an arc fault. The first fault arc voltage waveform can be obtained by differentiating the non-fault waveform from the first fault waveform, i.e., the first fault waveform can be identified from ζ2(t). After a fault occurs, the differential waveforms of adjacent fault cycles exhibit different characteristics due to the randomness of arcing, i.e., ζ3(t).

[0106] It can be seen that the terminal voltage difference between the non-fault cycle and the fault cycle can identify the presence of the upstream fault arc voltage.

[0107] Step S3: After the first fault waveform occurs, the arc fault is continuously identified and counted, and the continuous arc fault is identified by using the difference between the non-fault waveform before the fault occurs and the waveform to be identified to confirm the occurrence of the arc fault.

[0108] Based on the analysis in the above steps, the first fault waveform can be identified by using the adjacent wave voltage difference. That is, when the terminal voltage difference between the first non-fault cycle and the fault cycle meets the square wave characteristics, it is identified as the first fault waveform. Series arc faults generally have a continuous characteristic when they occur. The arc protector UL1699 certification and the national standard GB14287.4 also use the number of half-wave arcs that appear within 1s as the basis for alarm or action. Therefore, it is necessary to continuously identify and count arc faults after the arc fault occurs. The difference between the non-fault waveform before the fault occurs and the waveform to be identified can be used to identify continuous arc faults. In actual systems, since the load power factor and grid frequency fluctuations will cause the terminal voltage phase to change, when the waveform interval used for differentiation is long, the phase shift of the voltage needs to be considered.

[0109] For the above step S3, "using the difference between the non-fault waveform before the fault occurs and the waveform to be identified to realize the identification of the continuous arc fault", the fault detection principle of dual-mode time difference can be used for inspection.

[0110] Figure 5 This is the principle diagram of the fault detection of dual-mode time difference. Figure 5 As shown in the figure, before the first fault arc occurs, the adjacent wave differential inspection mode is used for inspection; after the first fault cycle is identified, the non-fault waveform before the fault occurs is used as the reference waveform, and then the reference wave differential detection mode is switched to confirm and count the faults; when multiple fault-free waveforms are continuously identified, it indicates that the fault has disappeared, and the identification mode is switched back to the adjacent wave differential inspection mode.

[0111] Considering the possibility of a gradual weak characteristic fault before the arc fault occurs, in order to ensure the reliable startup of the reference waveform detection algorithm, the time difference method of the adjacent wave differential waveform group is used in the inspection mode. Figure 6 This is a time waveform differential logic diagram. Figure 6 As shown, k consecutive waveforms are grouped together and time-differentiated with the next group of k waveforms, using half-wave units to generate 2k voltage differential Δu waveforms. If the characteristics of any of the 2k waveforms do not exceed the threshold, the waveform is considered a non-fault cycle, and identification continues with the next group of k waveforms. If any of the voltage differential Δu waveform characteristics exceed the threshold, an arc fault is suspected, and the system enters the parametric differential detection mode. The first waveform of the waveform group before the suspected fault occurs is used as the detection reference. This waveform is used to differentially identify and count the suspected fault waveforms one by one, and the number of such waveforms is recorded as the basis for alarm or action. When m consecutive full-cycle non-fault waveforms are identified, the fault has disappeared, and the system returns to the adjacent-wave differential inspection mode.

[0112] To ensure detection sensitivity, in the adjacent wave differential inspection mode, a suspected fault is considered to have occurred when one of the following detection conditions is met.

[0113]

[0114] In the above formula (11), represents the voltage change rate of the i-th voltage difference △u, represents the maximum value of the voltage change rate in the i-th voltage difference △u, It represents the effective value of the i-th voltage difference △u, and α1 and β1 are the set detection thresholds.

[0115] For the above-mentioned parameter wave differential detection mode, after entering the parameter wave differential detection mode, the non-fault waveform before the fault occurs is first used as the reference waveform u ref , reference waveform u ref and the waveform u to be identified x Perform phase synchronization and then perform differential identification. The criteria for fault identification are as follows:

[0116]

[0117] In the above formula (12), α up and α down They represent the upper and lower detection thresholds of the parameter wave differential detection mode, which are used to characterize the arc starting / extinction process; β2 represents the half-wave effective value setting threshold, which is used to characterize the existence of the arc voltage drop; Represents the reference waveform and the waveform to be identified u x The rate of change of voltage of the voltage differential, Represents the half-wave effective value. The criterion is based on arc voltage with obvious arcing peak; When , it is considered that the waveform has no fault; For arc voltage detection without obvious arcing spike, this criterion is consistent with the half-wave effective value. When all the above conditions are met at the same time, the half-wave is considered to be a fault waveform.

[0118] For the terminal voltage u0(t), although the upstream series arc fault can cause terminal voltage distortion, the terminal voltage component is still mainly power frequency. Therefore, identifying the power frequency phase of the terminal voltage can obtain its phase difference with the reference voltage waveform.

[0119] For reference waveform u ref and the waveform u to be identified x , perform N-point fast Fourier transform respectively to obtain the frequency domain phasor of each power frequency component:

[0120] a+bj={FFT[uref(n)]} 工频 Formula (13)

[0121] c + dj = {FFT[ux(n)]} 工频 Formula (14)

[0122] In the above equations (13) and (14), a and b represent the reference waveform u respectively. ref The real and imaginary parts of the power frequency phasor component, c and d represent the waveform u to be identified. x The real and imaginary parts of the power frequency phasor components.

[0123] The reference waveform u is obtained based on the frequency domain phasor of the power frequency component ref and the waveform u to be identified x Phase difference △θ:

[0124] Δθ=[arctan(b / a)-arctan(d / c)] Formula (15)

[0125] The reference waveform u ref To the waveform u to be identified x The formula for the number of aligned data shifts △N is:

[0126] ΔN=[Δθ / 2π·N]Equation (16)

[0127] In the above formula (16), N represents the reference waveform u ref and the waveform u to be identified x The number of sampling points per cycle, “[……]” indicates rounding the calculation result.

[0128] The reference waveform u ref Circularly shift right by △N bits to eliminate the phase difference:

[0129] u'ref(n)=uref((n-ΔN))NRN(n) equation (17)

[0130] In the above formula (17), u ref ((n)) N Indicates u ref (n) Periodic extension sequence with N as period, R N (n) represents a rectangular window sequence of length N.

[0131] After the arc enters the arc burning stage, its arc burning voltage is determined by the free arc burning characteristics of the AC arc. The nonlinear characteristics of some nonlinear load currents themselves will cause the arc burning voltage to appear unstable, thereby affecting the square wave characteristics of the arc voltage. Therefore, before executing the above step S2 of "characteristically identifying the terminal voltage difference between adjacent cycles based on the detection characteristics", the differential voltage needs to be envelope filtered first.

[0132] The envelope filtering of the differential voltage specifically includes: performing low-pass filtering on the differential voltage to filter out high-frequency glitch noise, using NP sampling points as local envelope identification windows, calculating the amplitude and position of the maximum and minimum points in each local envelope identification window, using a data interpolation algorithm to generate the upper and lower envelopes of the waveform, and taking the average of the upper and lower envelopes as the envelope filtering result.

[0133] The interpolation algorithm includes nearest neighbor interpolation, linear interpolation, cubic spline function interpolation, piecewise cubic Hermite interpolation and other methods, and preferably the cubic spline function interpolation algorithm is used.

[0134] Figure 7 The following is the envelope filtering result of the arc voltage using a computer load as an example. It can be observed that after envelope filtering, the high-frequency fluctuations during the unstable arc burning phase are effectively filtered out. At the same time, the transition information of the positive and negative half-cycle arc is well preserved, and the square wave characteristics of the arc are effectively restored.

[0135] The arc fault feature identification method of the present invention is verified using different loads. Figure 8 The waveform diagram shows the arc fault feature identification results based on different loads. Figure 8 (a) to (e) in the figure respectively use electric heater, vacuum cleaner, microwave oven, induction cooker, and computer as examples to identify arc fault characteristics. Figure 8 The fault feature recognition result is obtained by using 4 consecutive waveforms as a waveform group for adjacent wave inspection. The detection threshold α1 is set to 3V, and the lower limit detection threshold α of the reference wave differential detection mode is set to 3V. down The upper detection threshold α of the differential detection mode is 3V. up The detection threshold β1 is set to 10V, the half-wave effective value threshold β2 is set to 6V, and after detecting the first fault cycle, the parameter wave differential detection mode is started to confirm the fault.

[0136] Observe the recognition results. Figure 8 In part (a), although the first fault cycle is a weak fault caused by arc sparks and does not have an obvious arcing process, its maximum voltage change rate characteristic can still trigger the algorithm to enter the parameter wave differential detection mode. After entering the parameter wave differential detection mode, each fault half-wave can be identified by the criterion of the above formula (12). Figure 8 In part (b), the first fault cycle has no obvious arcing mutation characteristics and is a half-wave weak fault, but it can be triggered into the parameter wave differential detection mode through the half-wave arcing effective value characteristics. Figure 8 In parts (c) to (e), the first fault cycle can be triggered by any fault feature to enter the parameter wave differential detection mode.

[0137] Figure 9 The waveform diagram of the line current and terminal voltage when three loads, namely electric stove, vacuum cleaner and microwave oven, are switched on and off arbitrarily. Figure 9 The recording length is 5000 cycles, during which six sets of arc faults are randomly generated. The occurrence of the arc faults is reflected in the arc voltage waveform. When the first, second, fourth, fifth, and sixth arc faults occurred, the vacuum cleaner was in operation. Since the vacuum cleaner is a resistive and inductive load, the arc onset and extinction characteristics are not obvious. When the third arc fault occurred, only the resistive load of the electric furnace was in operation. Therefore, the arc voltage has a clear arc onset and extinction phase angle, and its arc voltage amplitude is significantly greater than that of the other waveforms. The load changes are reflected in the current waveform, and the five current mutation glitches are caused by the vacuum cleaner switching on and off. Both the arc fault and the load switching are reflected at the voltage monitoring point.

[0138] Figure 10 The figure is the fault feature recognition result of the recorded waveform. Figure 10 As shown in Figure 1, by observing the fault characteristics as a whole, all six arc faults can be effectively identified. In addition to the actual fault, the vacuum cleaner switching can also cause the fault characteristics to exceed the limit (such as Figure 10 (indicated by the arrow in the middle).

[0139] Figure 11 Figure 2 shows the actual arc and fault feature identification results for a resistive load. Resistive loads have large arcing and extinction phase angles, resulting in a large arc voltage rate of change. Using only the voltage rate of change feature can identify most fault half-waves. For a few fault half-waves, composite fault features can also be used for identification. Furthermore, it can be observed that the arc voltage of resistive loads is relatively stable before the arc fault disappears, resulting in the highest fault identification accuracy for this type of load.

[0140] Figure 12 This is the fault identification result diagram when the three loads of electric furnace, vacuum cleaner and microwave oven are running at the same time. The existence of resistive, inductive and nonlinear loads makes the arc voltage arcing and arcing phase angle significantly smaller than Figure 11The fault characteristics of the load shown are mainly identified by composite characteristics. In addition, the arc voltage intermittently shows temporary arc disappearance and incomplete arcing before the arc is extinguished. For the intermittent non-fault half-wave, this method can correctly identify it.

[0141] Figure 13 The figure shows the actual differential mode of the waveform recorded by the method of the present invention and the final recognition result. It can be observed that when 6 groups of arc faults occur, the method of the present invention can accurately realize the reliable conversion from the adjacent wave differential inspection mode to the parametric wave differential detection mode; in addition, the 5 switching on and off of the vacuum cleaner caused 4 false starts of the parametric wave differential detection mode. After 5 cycles (mode switching threshold) of the parametric wave differential detection mode in the method of the present invention, it is judged that the fault disappears and it can automatically switch back to the adjacent wave differential inspection mode. From the recognition results, the fault identification misjudgment caused by the false start of the parametric wave differential detection mode lasts for a maximum of 1 cycle. For the false start and misjudgment caused by the start of such high-power equipment, the number of fault arcs per unit time can be used for lockout protection.

[0142] The above detection method can be used in an arc fault detection device, which is commonly abbreviated as AFDD. The AFDD can be a standalone device or integrated into an arc fault circuit breaker.

[0143] For the method of identifying the existence of local characteristics of arc voltage to characterize its existence, the arc fault detection method of the present invention uses the dual characteristics of differential voltage change rate and effective value to perform a global waveform description of the differential waveform. The fault arc voltage amplitude is low and belongs to a weak characteristic signal. The time differential change method uses differential operation to effectively highlight the arc fault characteristic quantity, while effectively offsetting the inherent distortion component of the terminal voltage, thereby realizing the direct separation of the fault arc voltage signal. In the adjacent wave differential inspection mode, the time differential method of the waveform group is used to ensure reliability by increasing the number of identification times of the first wave of the fault; after the first wave of the fault is detected, the parametric wave detection mode is used to identify the fault, avoiding the blind spot of the adjacent wave differential identification method. In the parametric wave differential detection mode, the phase synchronization of the differential voltage signal is performed to eliminate the influence of grid frequency fluctuations and load current changes on the extraction of arc voltage waveforms.

[0144] In summary, the arc fault detection method of the present invention has better reliability, clearer physical meaning, high detection sensitivity, and is not affected by line parameters. Compared with the existing technology, the method of the present invention only requires a single monitoring point to achieve detection, and its cost is lower.

[0145] Although the present invention has been disclosed above with reference to the embodiments, they are not intended to limit the present invention. Anyone with ordinary skill in the art may make slight changes and modifications without departing from the spirit and scope of the present invention. Therefore, the scope of protection of the present invention shall be determined by the scope of the appended patent applications.

Claims

1. A method for detecting arc faults based on voltage time differential variation, characterized in that: include, Step S1, equating the arc fault point to an arc voltage source, and using the arc voltage change rate and the arc voltage effective value to establish a detection feature description of the fault waveform generated by the arc voltage source, wherein the fault waveform presents a square wave-like waveform; Step S2: sampling adjacent cycle voltages at the monitoring point on the load side, performing feature recognition on the terminal voltage difference between adjacent cycles based on the detection characteristics, and identifying the first fault waveform; Step S3: After the first fault waveform occurs, the arc fault is continuously identified and counted, and the continuous arc fault is identified by using the difference between the non-fault waveform before the fault occurs and the waveform to be identified to confirm the occurrence of the arc fault.

2. The arc fault detection method based on voltage time differential variation according to claim 1, characterized in that: In the above step S1, the arc voltage change rate is approximately expressed by the following formula: Where, represents the approximate value of the arc voltage change rate, U m Indicates the peak voltage, Indicates the power frequency angular frequency, T s represents the sampling time interval, U rh Indicates the arc peak voltage at the time of arc starting, U arc Indicates the arc burning voltage, L Indicates the number of sampling points in the transition zone; Arc voltage RMS The formula is: Where, u arc ( n ) represents the arc voltage sampling point, N Indicates the number of sampling points per cycle, M Indicates the number of sampling points corresponding to the arcing voltage, U arc Indicates the arc burning voltage, Indicates the arc starting and arc extinction voltage before the arc enters the arcing stage.

3. The arc fault detection method based on voltage time differential variation according to claim 1, characterized in that: In the above step S2, the expression of the terminal voltage difference is: Terminal voltage difference between non-fault cycles The expression is: Where, i 1( t ) is the previous current cycle used for differential operation in the non-fault state, i 2( t ) is the last current cycle used for differential operation in the non-fault state, R is the sum of the line resistances before and after the fault point, L is the sum of the system impedance and the line inductance before and after the fault point, ζ 1( t ) is the sinusoidal fluctuation in the differential waveform of two adjacent cycles of the terminal voltage in the non-fault state; The terminal voltage difference between the non-fault cycle and the fault cycle The expression is: Where, i arc ( t ) is the first fault current cycle after the fault occurs, i ( t ) is the last non-fault current cycle before the fault occurs, u arc ( t ) is the arc voltage at the fault point, ζ 2( t ) is the differential waveform of the terminal voltage excluding the arc voltage at the fault point after the non-fault state is converted to the fault state, that is, the sinusoidal fluctuation in the differential waveform; Terminal voltage difference between fault cycles The expression is: Where, i arc1 ( t ) is the previous current cycle used for differential operation under fault conditions, i arc2 ( t ) is the last current cycle used for differential operation under fault conditions, u arc1 ( t ) is the arc voltage at the fault point corresponding to the previous current cycle, u arc2 ( t ) is the arc voltage at the fault point corresponding to the subsequent current cycle, ζ 3( t ) is the sinusoidal fluctuation in the differential waveform of two adjacent cycles of the terminal voltage under fault conditions.

4. The arc fault detection method based on voltage time differential variation according to claim 3, characterized in that: Before an arc fault occurs, ζ 1( t ) is manifested as a small fluctuation; when an arc fault occurs, a square wave waveform appears. ζ 2( t ) Identify the first fault waveform; After a fault occurs, the frequency differences between adjacent faults show different characteristics due to the randomness of arc burning. ζ 3( t ) will show different characteristics.

5. The arc fault detection method based on voltage time differential variation according to claim 1, characterized in that: The above step S3, "using the difference between the non-fault waveform before the fault occurs and the waveform to be identified to realize the identification of continuous arc faults", specifically includes: before the first fault arc occurs, using the adjacent wave differential inspection mode to inspect, taking k consecutive waveforms as a group, and performing time differentiation with the next group of k adjacent waveforms, in units of half waves, to obtain 2k voltage differentials △ u Waveform, if the 2k voltage differential △ u The arc voltage change rate and arc voltage effective value of the waveform are within the limit, indicating that the waveform is a non-fault cycle. Then, the next group of k waveforms will be used to continue the identification. If any voltage difference △ u If the arc voltage change rate and arc voltage effective value of the waveform exceed the limit, an arc fault is suspected. After the first fault cycle is identified, the first waveform of the waveform group before the suspected fault occurs is used as a reference for the detection mode. This waveform is used to perform differential identification and count of the suspected fault waveforms one by one. When m consecutive full-cycle non-fault waveforms are identified, it indicates that the fault has disappeared, and the system returns to the adjacent wave differential inspection mode.

6. The arc fault detection method based on voltage time differential variation according to claim 5, characterized in that: In the adjacent wave differential inspection mode, an arc fault is suspected when one of the following detection conditions is met: Where, Indicates the i Voltage differential △ u The voltage change rate of the waveform, Indicates the i Voltage differential △ u The maximum value of the voltage change rate in the waveform, Indicates the i Voltage differential △ u The effective value of the waveform, α 1. β 1 is the set detection threshold.

7. The arc fault detection method based on voltage time differential variation according to claim 6, characterized in that: After entering the reference waveform differential detection mode, the non-fault waveform before the fault occurs is first used as the reference waveform. u ref , reference waveform u ref The waveform to be identified u x Perform phase synchronization and then perform differential identification. The criteria for fault identification are as follows: Where, α up and α down They represent the upper and lower detection thresholds of the parameter wave differential detection mode, respectively, and are used to characterize the arc starting / extinction process; β 2 represents the half-wave effective value setting threshold, which is used to characterize the existence of arc voltage drop; Indicates the reference waveform and the waveform to be identified u x The rate of change of voltage of the voltage differential, Indicates half-wave effective value.

8. The arc fault detection method based on voltage time differential variation according to claim 7, characterized in that: The parameter wave differential detection mode also includes: Reference waveform u ref and the waveform to be identified u x , respectively N Perform a point-wise fast Fourier transform to obtain the frequency domain phasors of the respective power frequency components: Where, a 、 b Respectively represent the reference waveform u ref The real and imaginary parts of the power frequency phasor components, c 、 d Represent the waveform to be identified u x Real and imaginary parts of power frequency phasor components; The reference waveform is obtained based on the frequency domain phasor of the power frequency component u ref and the waveform to be identified u x Phase difference △ θ : The reference waveform u ref To the waveform to be identified u x Number of aligned data shifts△ N The formula is: Where, N Indicates the reference waveform u ref and the waveform to be identified u x The number of sampling points per cycle, "[……]" means rounding the calculation result; The reference waveform u ref Circular right shift△ N bits to eliminate the phase difference: Where, express by N is the periodic extension sequence of the period, R N ( n ) indicates the length is N A sequence of rectangular windows.

9. The arc fault detection method based on voltage time differential variation according to claim 1, characterized in that: The above step S2 further includes, before performing "characteristic recognition of the terminal voltage difference between adjacent cycles based on the detection characteristics", performing envelope filtering on the differential voltage; The envelope filtering of the differential voltage specifically includes low-pass filtering of the differential voltage to remove high-frequency glitch noise. NP The sampling points are used as the local envelope recognition window, and the amplitude and position of the maximum and minimum points are calculated in each local envelope recognition window. The upper and lower envelopes of the waveform are generated by the data interpolation algorithm, and the average of the upper and lower envelopes is used as the envelope filtering result.

10. An arc fault detection device, characterized in that: The arc fault detection device implements the arc fault detection method based on voltage time differential variation according to any one of claims 1 to 9.

Images