A method for detecting series arc faults in low-voltage distribution lines based on waveform similarity algorithm

Abstract

The present invention proposes a method for detecting arc faults in series on low-voltage distribution lines based on a waveform similarity algorithm. First, arc fault tests and fault feature analysis are performed on common loads, and the voltage sag threshold for starting arc fault detection and the H threshold for identifying arc faults are given. Secondly, based on the voltage sag characteristics at the load end when an arc fault occurs, the load end voltage sag value exceeding the voltage sag threshold is used as the detection start condition. Then, the similarity of the load end voltage waveform is calculated based on the Hausdorff distance algorithm. When the Hausdorff distance H value between the load end voltage and the sine wave exceeds the H threshold, it is determined that an arc fault has occurred. Finally, the recognition effect of the detection method proposed by the present invention is verified through examples. The results show that this method can effectively improve the reliability of arc fault detection.

Description

Technical Field

[0001] The present invention belongs to the field of low-voltage distribution network series arc detection, and mainly relates to a low-voltage distribution line series arc detection method suitable for large-scale nonlinear load access. Background Art

[0002] Fires on distribution lines are a major cause of electrical fires, accounting for over 40% of electrical fires each year. Arc faults caused by loose terminals, insulation damage, and conductor damage are a key contributing factor.

[0003] Arc faults can be divided into series arc faults and parallel arc faults depending on the circumstances under which they occur. In a parallel arc fault, the current is limited only by the arc impedance, and the current magnitude is close to the short-circuit current. Most overcurrent protection devices can treat it as a short-circuit fault and operate. At the same time, sensors can directly detect the arc voltage and current waveforms with common characteristics, making detection relatively easy. In a series arc fault, the current is affected by both the arc impedance and the load impedance. The current is generally smaller than the normal operating current of the load, and the waveform is significantly affected by the load type. Traditional circuit protection devices cannot effectively protect against it. Furthermore, because the arc fault location is uncertain, it is difficult to directly detect the arc voltage. Therefore, it is necessary to design a specialized series arc fault detection method to protect the line.

[0004] Existing arc fault detection methods based on arc current have reliability flaws. With the diversification of user-side load types, the waveform characteristics of low-voltage distribution network line currents have become diverse. The addition of a large number of nonlinear loads causes the current during normal operation to also exhibit fault characteristics similar to arc currents. This makes existing series arc fault detection methods prone to misjudgment and no longer meet the demand for effective distribution line protection. Summary of the Invention

[0005] In order to solve the problem of misjudgment of traditional fault arc detection methods caused by nonlinear load operating current and enhance the reliability of fault arc detection methods under complex load environments, the present invention proposes a low-voltage distribution line fault arc detection method based on a waveform similarity algorithm.

[0006] The technical method adopted in the present invention is as follows:

[0007] A method for detecting arc faults in series on low-voltage distribution lines based on a waveform similarity algorithm comprises the following steps:

[0008] S1: Test the voltage sag when different types of loads are connected to determine the maximum voltage drop u for different types of loads max , and determine the starting threshold U according to the maximum voltage drop set ;

[0009] S2: Calculate the waveform similarity between the load terminal voltage waveform and the standard sine wave when the asynchronous motor load with the maximum possible capacity is connected, and determine the detection threshold H in combination with the fault discrimination reliability coefficient. set ;

[0010] S3: Real-time sampling and calculation of the effective value of the single-phase load terminal voltage, calculate the difference between the effective value of the load terminal voltage in the current cycle and the previous cycle, and obtain the voltage sag value u drop (t), according to the voltage sag value u drop (t) and the starting threshold U set Compare and judge whether the arc fault detection is started;

[0011] S4: If the arc fault detection is started, the Hausdorff distance H between the fault voltage at the load end and the standard sine wave is calculated;

[0012] S5: When the Hausdorff distance H in S4 exceeds the detection threshold H set When the voltage is 0, it is determined that an arc fault occurs in the line; otherwise, it is determined that no arc fault occurs.

[0013] Furthermore, the step S1 specifically includes:

[0014] Four representative linear and nonlinear loads are selected. The representative loads are electric heater, induction cooker, electric water heater and vacuum cleaner. Among them, the electric heater is a 1000W linear load with a power factor of The induction cooker is 1800W linear load, power factor The electric water heater is a 3000W linear load with a power factor of The vacuum cleaner is a 1300W non-linear load with a power factor of

[0015] The voltage sag value when 4 representative loads are connected is tested. The load terminal voltage when 4 representative loads are connected is sampled and recorded respectively, with a sampling rate of 1kHz. At the end of each power frequency cycle sampling, the effective value of the primary terminal voltage is calculated in real time. At the same time, the difference between the effective value of the load terminal voltage in the current cycle and the previous cycle is calculated and recorded as the voltage sag value u sag (t), the voltage sag value is calculated as follows:

[0016] u sag (t)=u(t)-u(tT) (1)

[0017] Where u sag (t) is the voltage sag value, u(t) is the effective value of the load terminal voltage in the current cycle, and u(tT) is the effective value of the load terminal voltage in the previous cycle;

[0018] Compare the load terminal voltage sag values ​​after the above four different types of loads are connected to obtain the maximum voltage sag u max For the above four representative loads, the voltage sag values ​​generated when they are connected are recorded as u sag1 、u sag2 、u sag3 、u sag4 , and compare the load terminal voltage sag values ​​after the above different loads are connected to obtain the maximum voltage sag u max :

[0019] u max =max[u sag1 ,u sag2 ,u sag3 ,u sag4 ] (2)

[0020] Among them, max means to find the maximum value of the voltage drop among the four types of loads;

[0021] Set the reliability factor for detection startup Setting the starting threshold U for arc fault detection set , ensuring that representative loads are connected reliably without malfunction:

[0022]

[0023] Where: u max The maximum voltage drop when the load is connected; The reliability coefficient for detecting startup can be taken as 1.1 to 1.3.

[0024] Furthermore, the waveform similarity between the load-end voltage waveform and the standard sine wave is the Hausdorff distance between the two waveforms.

[0025] Furthermore, the step S2 specifically includes:

[0026] Record the sampling array u=[u1,u2,…,u n ], sampling rate 10kHz, n is the number of sampling points in one power frequency cycle, n = 200 at 10kHz sampling rate, and record the maximum voltage u in the voltage array mag =max[u1,u2,…,u n ];

[0027] According to the maximum voltage u obtained by sampling mag , generating an amplitude of u mag A standard sine wave;

[0028] The data window length of 0.1T is used as the data input for calculating waveform similarity, that is, Hausdorff distance, where T is the power frequency cycle. Thus, 10 Hausdorff distance calculation values ​​are calculated in sequence within one power frequency cycle to form a Hausdorff distance sequence H = [H1, H2, ..., H 10 ], the calculation formula for each Hausdorff data in the Hausdorff distance sequence is as follows:

[0029]

[0030] Where: |·| is the distance norm between array u and array y; min means for any point a in u i , calculate the distance norm values ​​between it and all data points in y in turn, and compare and get the minimum value of all distance norm values. If u contains n data, the minimum value of n distance norms will be calculated; max means to find the maximum value of the aforementioned n minimum values ​​of distance norms; a i ∈u represents all elements in the voltage array u; b j ∈y represents all elements in the standard sine wave array y;

[0031] Calculate the above waveform similarity Hausdorff distance series H=[H1,H2,…,H 10 ] is the maximum value of the elements in max , detection threshold H set Set to H max Multiply the reliability coefficient

[0032]

[0033] In formula (5): It is generally taken as 1.5 to 3.

[0034] Furthermore, the maximum voltage u obtained by sampling mag , generating an amplitude of u mag The specific steps are as follows:

[0035] Step 1: Determine the sampling frequency f, and calculate the sampling period Ts = 1 / f accordingly;

[0036] Step 2: Generate a discrete time series t with a length of T = 0.02s, where the initial value of t is 0 and the step length is the sampling period Ts. The generated time series t = [0, Ts, 2*Ts, ..., (N-1)*Ts] starting from 0 and with a total length of N, where N = T / Ts is the length of the time series t;

[0037] Step 3: For each time point t in the time seriesi , calculate the value y of the standard sine wave at that time point i =u mag sin(2*π*50*t i ), thus obtaining a discrete standard sine wave numerical sequence y=[y1,y2,…,y n ], n is the number of sampling points per power frequency cycle at a sampling rate of 10kHz, n = 200.

[0038] Furthermore, the step S3 specifically includes:

[0039] The single-phase load terminal voltage is sampled in real time with a sampling rate of 1kHz. At the end of each power frequency cycle sampling, the effective value of the primary terminal voltage is calculated in real time. At the same time, the difference between the effective value of the load terminal voltage in the current cycle and the previous cycle is calculated to obtain the voltage sag value u drop (t);

[0040] The voltage sag value exceeding the limit is used as the starting condition for detection. drop (t) is greater than the starting threshold U set When the arc fault detection starts, the voltage sag value u drop (t) is less than the starting threshold U set When the arc fault detection is not started, the startup strategy is as follows:

[0041]

[0042] After the arc fault detection is started, the load voltage is sampled at a sampling rate of 10kHz. The sampling data storage window length is the next power frequency cycle after the detection is started. The load voltage sampling array u is obtained. arc =[u1,u2,…,u n ], n is the number of sampling points in one power frequency cycle, n = 200 at a sampling rate of 10kHz, and the maximum voltage u in the voltage array is recorded peak =max[u1,u2,…,u n ].

[0043] Compared with the prior art, the technical effects of the present invention are as follows:

[0044] The voltage starting threshold proposed by the present invention can be used to distinguish between nonlinear loads and arc faults, thereby avoiding the problem of false start-up caused by the high similarity between the operating current characteristics of the nonlinear load and the fault arc current characteristics; the present invention uses waveform similarity to calculate the degree of distortion of the load-end voltage waveform, which can effectively identify the start of asynchronous motors while achieving reliable detection of fault arcs, thus making up for the shortcomings of similar fault arc detection methods. In summary, the method proposed by the present invention can effectively improve the accuracy of series arc detection in low-voltage distribution lines, and the proposed criteria are simple and require low terminal computing power. The invention takes into account both effectiveness and practicality and has strong engineering practical value. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 This is a flow chart of a method for detecting series arc faults in low-voltage distribution lines based on a waveform similarity algorithm according to an embodiment of the present invention;

[0046] Figure 2 This is a diagram of a power distribution structure for a residential building according to an embodiment of the present invention;

[0047] Figure 3 is the voltage sag magnitude when different loads are connected in the embodiment of the present invention;

[0048] Figure 4 is the Hausdorff distance of the load terminal voltage when the three-phase asynchronous motor according to the embodiment of the present invention is started;

[0049] Figure 5 It is the Hausdorff distance of the load terminal voltage when a series arc fault occurs in the embodiment of the present invention. DETAILED DESCRIPTION

[0050] The present invention will be further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0051] Residential buildings are selected as research objects, and their distribution network is a three-level distribution network structure, such as Figure 2 As shown. Matlab / Simulink simulation software was used to simulate the series arc fault detection method based on voltage sag characteristics and Hausdorff distance algorithm proposed in this paper, and the arc fault simulation circuit was built based on the Cassie arc model. The specific parameters of the model are as follows: the distribution network side is equivalent to a 220V constant voltage source, and the 3rd, 5th, and 7th harmonic sources with effective values ​​of 5V, 3V, and 2V are added. The initial phases of each harmonic are π / 6, π / 3, and π / 2 respectively. The line length r is 100m and the inner diameter is 16mm 2 , line resistance R = 0.1Ω, line inductance L = 80μH, line s1 length is 100m, inner diameter is 10mm 2, line resistance R = 0.16Ω, line inductance L = 95μH, line l1 length is 100m, inner diameter is 4mm 2 , line resistance R = 0.5Ω, line inductance L = 100μH. The three-phase asynchronous motor is a squirrel-cage asynchronous motor with an input power of 7.5kW, a rated voltage of 400V, a rated frequency of 50Hz, and a rated speed of 1430rpm. The loads tested for arc faults include electric heaters, electric water heaters, induction cookers, and vacuum cleaners. The following analyzes the adaptability of the new series arc fault detection method based on specific scenarios.

[0052] Example

[0053] The following examples are provided based on two scenarios: different loads connected during normal operation and an arc fault occurring during operation of an electric heater load.

[0054] Example 1: Before the load is connected, the line has no load. At 0.06s, four different types of loads are connected to the line. Figure 1 As shown:

[0055] S1: Test the voltage sag when different common loads (electric heaters, induction cookers, electric water heaters, vacuum cleaners) are connected to determine the maximum voltage drop u when different types of loads are connected max , and determine the starting threshold U according to the maximum voltage drop set .

[0056] The specific implementation process of step S1 is as follows:

[0057] Step 1: Perform a load access test on the load types listed in Table 1:

[0058] Table 1 Main parameters of test load

[0059]

[0060] Step 2: Test the load voltage sag test results. Figure 3 , load A temporary drop value u drop1 =4.22V, load B temporary drop value u drop2 =4.68V, load C temporary drop value u drop3 =8.61V, load D sag value u drop4 =6.13V.

[0061] Step 3: Among the four loads, load C has the largest starting voltage sag, u max =max[u drop1 ,u drop2 ,u drop3 ,u drop4 ]=8.61V.

[0062] Step 4: Calculate the start threshold according to formula (3).r 1 el Take 1.2 and set the starting threshold U set =10.33V.

[0063] S2: Calculate the waveform similarity between the load terminal voltage waveform and the standard sine wave when the asynchronous motor load with the maximum possible capacity is connected, and determine the detection threshold H in combination with the fault discrimination reliability coefficient. set The specific implementation process is as follows:

[0064] Step 1: Record the sampling array u=[u1,u2,…,u n ], n = 200, and record the maximum voltage u in the voltage array peak =max[u1,u2,…,u n ]=292.9V.

[0065] Step 2: Based on the maximum voltage u obtained by sampling mag , generating an amplitude of u mag The steps for a standard sine wave are as follows:

[0066] Step 1: Determine the sampling frequency f = 10kHz, and calculate the sampling period Ts = 1 / f = 0.1ms based on this.

[0067] Step 2: Generate a discrete time series t of length T = 0.02s, where the initial value of t is 0 and the step size is the sampling period of 0.1ms. The generated time series t = [0, Ts, 2*Ts, ..., (N-1)*Ts], starting from 0 and with a total length of N, is obtained, where N = 200.

[0068] Step 3: For each time point t in the time series i , calculate the value y of the standard sine wave at that time point i =292.9sin(2*π*50*t i ), thus obtaining a discrete standard sine wave numerical sequence y=[y1,y2,…,y n ], n=200.

[0069] Step 3: Use a data window length of 0.1T as the data input for calculating the waveform similarity Hausdorff distance, where T is the power frequency period. Thus, within one power frequency period, 10 Hausdorff distance calculation values ​​can be obtained, forming a Hausdorff distance sequence H = [H1, H2, ..., H 10 ].

[0070] Step 4: Calculate the above Hausdorff distance series H = [H1, H2, ..., H 10 ]The maximum value of the elements in H max =4.81, such as Figure 4 As shown, Take 2, the detection threshold

[0071] Example 2: Before the fault, the distribution line with the electric heater load was operating normally; at 0.06s, a series arc fault occurred on line l1.

[0072] The implementation of S1 and S2 is the same as that of Example 1, thereby determining the start threshold U set =10.33V, detection threshold

[0073] S3: Real-time sampling and calculation of the effective value of the single-phase load terminal voltage, calculate the difference between the effective value of the load terminal voltage in the current cycle and the previous cycle, and obtain the voltage sag value u drop (t), according to the voltage sag value u drop (t) and the starting threshold U set Compare and determine whether the arc fault detection is started. The specific implementation process is as follows:

[0074] Step 1: Real-time sampling of the single-phase load terminal voltage, with a sampling rate of 1kHz, and at the end of each power frequency cycle sampling, real-time calculation of the effective value of the primary terminal voltage, and at the same time, the difference between the effective value of the load terminal voltage of the current cycle and the previous cycle is calculated in a similar way to S1 above to obtain the voltage sag value u drop (t)=23.8V.

[0075] Step 2: u drop (t)>U set , detection starts.

[0076] Step 3: After the arc fault detection is started, the load voltage is sampled at a sampling rate of 10kHz. The sampling data storage window length is the next power frequency cycle after the detection is started, thereby obtaining the load voltage sampling array u arc =[u1,u2,…,u n ], n = 200, and record the maximum voltage u in the voltage array peak =max[u1,u2,…,u n ]=276.63V.

[0077] S4: The maximum voltage u obtained by sampling peak =276.63V, follow the similar steps of generating the standard sine wave as above, and generate an amplitude of u peakThe Hausdorff distance H between the fault voltage at the load end and the standard sine wave is calculated using a similar method to the above Hausdorff distance calculation, as shown in Figure 5 shown.

[0078] S5: According to the calculation results of S4, H max =56.35>H set , it is judged that an arc fault has occurred, which is consistent with the actual situation that a series arc fault has occurred in line l1, and the detection result is accurate.

[0079] In summary, this embodiment verifies the correctness and superiority of the present invention. The present invention achieves accurate identification of series arc faults in asynchronous motors and nonlinear loads, solves the problem of misjudgment in traditional detection methods, and improves the reliability of fault detection results.

Claims

1. A method for detecting series arc faults in low-voltage distribution lines based on a waveform similarity algorithm, characterized in that: The steps include: S1: Test the voltage sag when different types of loads are connected to determine the maximum voltage drop when different types of loads are connected , and determine the starting threshold based on the maximum voltage drop ; S2: Calculate the waveform similarity between the load terminal voltage waveform and the standard sine wave when the asynchronous motor load with the maximum possible capacity is connected. The waveform similarity is the Hausdorff distance, and the detection threshold is determined in combination with the fault discrimination reliability coefficient. ; S3: Real-time sampling and calculation of the effective value of the single-phase load terminal voltage, and the difference between the effective value of the load terminal voltage in the current cycle and the previous cycle to obtain the voltage sag value , according to the voltage sag value and start threshold Compare and judge whether the arc fault detection is started; S4: If the arc fault detection is started, the Hausdorff distance H between the fault voltage at the load end and the standard sine wave is calculated; S5: When the Hausdorff distance H in S4 exceeds the detection threshold When the voltage is 0, it is determined that an arc fault occurs in the line; otherwise, it is determined that no arc fault occurs.

2. The method for detecting series arc faults in low-voltage distribution lines based on a waveform similarity algorithm according to claim 1, wherein: The step S1 specifically includes: Four representative linear and nonlinear loads are selected. The representative loads are electric heater, induction cooker, electric water heater and vacuum cleaner. Among them, the electric heater is a 1000W linear load with a power factor of =1, the induction cooker is 1800W linear load, power factor =0.95, the electric water heater is a 3000W linear load, the power factor =1. The vacuum cleaner is a 1300W nonlinear load, and the power factor is =0.87; The voltage sag value when 4 representative loads are connected is tested. The load terminal voltage when 4 representative loads are connected is sampled and recorded respectively. The sampling rate is 1kHz. At the end of each power frequency cycle sampling, the effective value of the primary terminal voltage is calculated in real time. At the same time, the difference between the effective value of the load terminal voltage in the current cycle and the previous cycle is calculated and recorded as the voltage sag value. , the voltage sag value is calculated as follows: （1）； In formula (1), is the voltage sag value, is the effective value of the load terminal voltage in the current cycle, is the effective value of the load terminal voltage in the previous cycle; Compare the load terminal voltage sag values ​​after the above four different types of loads are connected to obtain the maximum voltage sag For the above four representative loads, the voltage sag values ​​generated when they are connected are recorded as 、 、 、 , and compare the load terminal voltage sag values ​​after the above different loads are connected to obtain the maximum voltage sag : （2）； Among them, max means to find the maximum value of the voltage drop among the four types of loads; Set the reliability factor for detection startup , set the starting threshold of fault arc detection , ensuring that representative loads are connected reliably without malfunction: （3）； Where: The maximum voltage drop when the load is connected; The reliability coefficient for detecting startup can be taken as 1.1~1.

3.

3. The method for detecting series arc faults in low-voltage distribution lines based on a waveform similarity algorithm according to claim 1, wherein: The waveform similarity between the load terminal voltage waveform and the standard sine wave is the Hausdorff distance between the two waveforms.

4. The method for detecting series arc faults in low-voltage distribution lines based on a waveform similarity algorithm according to claim 1, wherein: The step S2 specifically includes: Record the sampling array of the load terminal voltage when the asynchronous motor with the maximum possible capacity is connected , sampling rate 10kHz, n is the number of sampling points in one power frequency cycle, n=200 at 10kHz sampling rate, and record the maximum voltage in the voltage array ; The maximum voltage obtained by sampling , generating an amplitude of A standard sine wave; A data window length of 0.1T is used as the data input for calculating waveform similarity, where T is the power frequency period. Thus, 10 Hausdorff distance calculation values ​​are calculated in sequence within one power frequency period to form a Hausdorff distance sequence. , the calculation formula for each Hausdorff data in the Hausdorff distance sequence is as follows: （4）； Where: Is the distance norm between array u and array y; min means for any point in u , calculate the distance norm values ​​between it and all data points in y in turn, and compare and get the minimum value of all distance norm values. If u contains n data, the minimum value of n distance norms will be calculated; max means to find the maximum value of the aforementioned n minimum values ​​of distance norms; Represents all elements in the voltage array u; Represents all elements of the standard sine wave array y; Calculate the Hausdorff distance series of the above waveform similarity The maximum value of the inner element is recorded as , detection threshold Set to Multiply the reliability coefficient : （5）； In formula (5): Take it as 1.5~3.

5. The method for detecting series arc faults in low-voltage distribution lines based on a waveform similarity algorithm according to claim 4, characterized in that: The maximum voltage value obtained by sampling , generating an amplitude of The specific steps are as follows: Step 1: Determine the sampling frequency f, and calculate the sampling period Ts = 1 / f accordingly; Step 2: Generate a discrete time series t with a length of T=0.02s, where the initial value of t is 0 and the step length is the sampling period Ts, and generate a time series starting from 0 and with a total length of N , where N = T / Ts is the length of the time series t; Step 3: For each time point in the time series , calculate the value of the standard sine wave at that time point , thus obtaining a discrete standard sine wave numerical sequence , n is the number of sampling points per power frequency cycle at a sampling rate of 10kHz, n=200.

6. The method for detecting series arc faults in low-voltage distribution lines based on a waveform similarity algorithm according to claim 1, wherein: The step S3 specifically includes: The single-phase load terminal voltage is sampled in real time with a sampling rate of 1kHz. At the end of each power frequency cycle sampling, the effective value of the primary terminal voltage is calculated in real time. At the same time, the difference between the effective value of the load terminal voltage in the current cycle and the previous cycle is calculated to obtain the voltage sag value. ; The voltage sag value exceeding the limit is used as the starting condition for detection. Greater than the starting threshold When the arc fault detection starts; the voltage sag value Less than the starting threshold When , the arc fault detection is not started, and the starting strategy is as follows (6): （6）； After the arc fault detection is started, the load voltage is sampled at a sampling rate of 10kHz. The sampling data storage window length is the next power frequency cycle after the detection is started, thereby obtaining the load voltage sampling array , n is the number of sampling points in one power frequency cycle, n=200 at 10kHz sampling rate, and record the maximum voltage in the voltage array .

Images