Current protection anti-excitation inrush misoperation method based on current sequence shaping

Abstract

The method for preventing the misoperation of the current protection caused by the magnetizing inrush current based on the current sequence shaping includes the following steps: constructing a standard power frequency sinusoidal sequence based on the 1 / 4 cycle data window extreme value, comparing the Euclidean distance similarity of two sequences to form a magnetizing inrush current fast starting criterion; extracting the second harmonic feature of the "spike-flat-spike" of the current sampling sequence; calculating the mean square deviation sigma (a, b) of the section with sequence number (a, b) to determine the boundary of the non-distortion section of the current signal; constructing a standard sinusoidal curve template, calculating the Pearson correlation coefficient of the non-distortion section current waveform and the standard sinusoidal curve template, finding the corresponding similar section of the two, and repairing the distortion section of the actual current waveform based on the standard curve; and using the current data after the shaping repair to substitute into the traditional protection judgment to realize the current protection function. The method has simple principle, does not need to increase auxiliary equipment, starts quickly, effectively avoids the misoperation of the line protection caused by the magnetizing inrush current, and can be well applied to the distribution network transmission line.

Description

Technical Field

[0001] This invention relates to the field of power distribution network line protection technology, specifically to a current protection method for preventing inrush current maloperation based on current sequence shaping. Background Technology

[0002] Downstream of distribution network feeders are numerous transformers. During no-load closing or reclosing to restore power after fault clearing, these transformers may experience significant inrush currents due to core saturation. As electricity demand increases, the number and capacity of downstream transformers also rise. When this increase reaches a certain level, the superimposed inrush currents on the feeders become substantial, easily causing feeder protection malfunctions. This can lead to transient fault reclosing failures or malfunctions in the fault recovery function of the distribution automation system, resulting in prolonged outages of feeders that could otherwise be quickly restored to power, severely impacting the reliability and quality of the distribution network.

[0003] Currently, there are relatively few research results on the identification and countermeasures for inrush current in distribution networks. In the actual operation of distribution networks, after the failure to close the circuit due to inrush current, the common approach is to "avoid" the inrush current by increasing the current protection setting value or extending the protection action time. Fu Zhenqiang, Zhou Zhenyu, Wang Shu, et al. Analysis of the causes of maloperation of 10kV distribution line protection [J]. Northeast Electric Power Technology, 2017, 38(01):39-42. The technical solutions described in this paper aim to avoid the maloperation of feeder protection caused by inrush current. However, these methods will sacrifice the sensitivity, speed and reliability of the feeder protection.

[0004] Analysis of inrush current flowing through feeder switches in various scenarios reveals the probability distribution law of inrush current on distribution network feeders (Liu Chao, Liu Jian, Zhang Zhihua, et al. Analysis and discussion on inrush current and its probability distribution in distribution network [J]. Automation of Electric Power Systems, 2017, 41(04): 170-175.). The distribution law can be used to optimize the current setting value of instantaneous overcurrent protection. By comparing the differences in waveform characteristics between inrush current and fault current, a distribution network inrush current identification method based on dynamic quadrilateral analysis is proposed (Yan Fuping, Duan Jiandong, Li Xiao, et al. Distribution network inrush current identification method based on dynamic quadrilateral analysis [J]. Power System Technology, 2015, 39(07): 2017-2022.). However, its identification effect is not good when inrush current and line load current coexist. Summary of the Invention

[0005] To address the problem of inrush current caused by unloaded switching of transformers in distribution networks, leading to maloperation of upstream line protection, this invention proposes a current protection method based on current sequence shaping to prevent inrush current maloperation. After traditional protection is activated, an Euclidean distance algorithm is used to construct an inrush current identification criterion. Then, the boundary of the undistorted section is determined through the second harmonic characteristics of the original current signal. Based on the undistorted section signal, the distorted portion of the original current signal is shaped and repaired. Finally, the shaped and repaired current signal is sent to the traditional overcurrent protection for judgment, avoiding line protection maloperation and ensuring the reliability of distribution network power supply. The aim is to reduce the risk of overcurrent protection maloperation caused by inrush current generated by unloaded transformer switching, ensuring the safe and stable operation of the distribution network.

[0006] The technical solution adopted in this invention is as follows:

[0007] The current protection method for preventing inrush current maloperation based on current sequence shaping includes the following steps:

[0008] Step 1: Calculate the current surge sequence in real time, and construct a standard power frequency sinusoidal sequence based on the extreme value of the 1 / 4 cycle data window. Compare the Euclidean distance similarity between the two sequences to form a criterion for rapid start-up of excitation inrush current.

[0009] Step 2: Use the FFT algorithm to extract the second harmonic characteristics of the current sampling sequence in the pattern of "peak-flat-peak" through a sliding data window;

[0010] Step 3: For the second harmonic RMS value sequence, calculate the root mean square error σ(a,b) of the segment with sequence number (a,b), and determine the boundary of the undistorted segment in the current signal by threshold comparison;

[0011] Step 4: Construct a standard sine curve template, use the current waveform in the undistorted section to calculate its Pearson correlation coefficient with the standard sine curve template, find the corresponding similar sections, and repair the distorted sections of the actual current waveform based on the standard curve until the shaping end criterion is met and the repair stops.

[0012] Step 5: Substitute the current data after the reshaping and repair into the traditional protection judgment to realize the current protection function.

[0013] In step 1, considering the situation where the inrush current generated by the transformer is superimposed on the line load current, causing load current distortion and potentially affecting the maloperation of overcurrent protection, a fast start-up criterion for inrush current is proposed using the Euclidean distance similarity algorithm after the traditional protection starts. This includes the following steps:

[0014] S1.1: Real-time calculation of the current mutation sequence Δi(n)=(Δi1,Δi2,…,Δi n ), Δi1,Δi2,…,Δi nThese represent the values ​​at each point in the current abrupt change sequence. A standard power frequency sinusoidal sequence y(n) = (y1, y2, ..., yn) is constructed based on the extreme values ​​of the 1 / 4 cycle data window. n ); y1, y2, ..., y n These represent the values ​​at each point in the standard power frequency sine sequence.

[0015] S1.2: Calculate the Euclidean distance similarity B between the current mutation sequence and the standard power frequency sinusoidal sequence, as shown in equation (1):

[0016]

[0017] Where: k is the number of sampling points in the 1 / 4 cycle data window. n represents the sampling data sequence number in the 1 / 4 cycle data window. S1.3: Startup criterion B set B set For the similarity threshold, considering both margin and accuracy, a value of 0.8 to 0.9 is recommended. In step 2, the second harmonic characteristics of the inrush current superimposed on the load current are characterized by: the inrush current having a discontinuity angle, which, when superimposed on the load current signal, will not cause waveform distortion in that section. The protection device uses the FFT algorithm to extract the second harmonic through a sliding data window, obtaining the second harmonic characteristics of "peak-smooth-peak", where the smooth section corresponds to the undistorted waveform section.

[0018] Step 3 includes the following steps:

[0019] S3.1: For the second harmonic RMS value sequence, with the real-time calculated sequence number as b, take ba = N / 10 as the initial segment, where N is the number of sampling points per cycle, and calculate the root mean square error of the (a, b) segment. I 2rms (k) is the effective value of the second harmonic. Let be the mean of the second harmonic effective values ​​within the interval (a,b).

[0020] S3.2: If σ(a,b)≥σ set , σ set The recommended value is the preset threshold. Repeat S3.1.

[0021] S3.3: If σ(a,b) < σ set The segment is determined to be a stable segment. Waiting for the next sampling time, point b is updated, b = b + 1 is set, σ(a,b) is recalculated, and it is determined that σ(a,b) < σ set To check if the condition is met, repeat step S3.3 until σ(a,b)≥σ is found. set The current (a,b) segment is determined to be the boundary of the non-distorted segment in this calculation process.

[0022] ​Step 4 includes the following steps:

[0023] S4.1: Determination of the standard sine curve: Construct a standard power frequency sine curve with an amplitude of 1 as a template, and define the continuous 1 / 4 cycle segment within 1 cycle as regions I, II, III, and IV respectively;

[0024] S4.2: In the real-time sampled current (a,b) data window, based on the monotonic trend and the sign of the first point of this part of the curve, locate the starting region in the standard curve. Based on the starting region sequence and the actual current waveform (a,b) segment sequence, as shown in Equation (2), calculate the Pearson correlation coefficient ρ(x,y). Let the standard current sequence x = {x1,x2,…,x} n},x1,x2,…,x n These represent the values ​​at each point in the standard current sequence. The actual current sequence is y = {y1, y2, ..., y...} n},y1,y2,…,y n These represent the values ​​at each point in the actual current sequence. The length of the undistorted segment is n = b - a + 1.

[0025]

[0026] x i This represents the standard current sequence value with index i, x j y represents the value at each point in the calculation of the mean of a standard current sequence. i This represents the actual current sequence value for number i, y j This represents the value at each point in the calculation of the mean of the actual current sequence.

[0027] By setting the threshold ρ set It is recommended to set it to 0.995. This applies when ρ(x,y)≥ρ set Determine the standard curve segment corresponding to the (a,b) segment, and calculate the numerical multiple of the standard curve.

[0028] S4.3: Reshaping and Repairing the Distorted Segment: The repair curve sequence of the distorted segment is determined according to equation (3).

[0029]

[0030] N represents the number of sampling points within one cycle.

[0031] Based on the starting point k of the standard sine segment st Let i(a) = K·y(k) be the boundary starting point a of the original sine segment. st The initial phase of the repair curve sequence is determined by equation (4).

[0032]

[0033] Following this process, the current sequence of each cycle is shaped and repaired, and the shaped and repaired current signal is sent to the traditional overcurrent protection.

[0034] S4.4: Shaping termination criterion: According to the non-distorted segment boundary determination method, the shaping operation ends when the length n of the non-distorted segment satisfies equation (5), where: n set The threshold value is recommended to be 0.75N.

[0035] n>n set (5).

[0036] This invention provides a current protection method for preventing inrush current maloperation based on current sequence shaping, with the following technical advantages:

[0037] 1) This method addresses the problem of maloperation of upstream line protection caused by inrush current due to no-load closing of the distribution network transformer. It proposes to use the second harmonic characteristics of the current signal and waveform shaping technology to identify the inrush current and restore the distorted waveform region, thereby ensuring the reliability of power supply in the distribution network.

[0038] 2) The method of the present invention constructs an excitation inrush current identification and start-up criterion based on the Euclidean distance similarity algorithm, which can quickly and reliably lock out protection and prevent false operation.

[0039] 3) The method of the present invention uses current sequence shaping technology to reshape and repair the distorted part of the current signal, which effectively improves the reliability of protection discrimination.

[0040] 4) This method is simple in principle, does not require any additional auxiliary equipment, starts up quickly, and effectively avoids line protection from malfunctioning due to inrush current. It can be well applied to distribution network transmission lines and is conducive to its promotion and application in engineering practice. Attached Figure Description

[0041] Figure 1 This is a flowchart of a method for preventing inrush current malfunction based on current shaping technology.

[0042] Figure 2 This is a schematic diagram of the second harmonic characteristics when the load current is superimposed with the inrush current.

[0043] Figure 3 A schematic diagram illustrating an example of current signal segmentation for inrush current superimposed with load current.

[0044] Figure 4 This is a schematic diagram illustrating an example of current sequence shaping technology.

[0045] Figure 5 This is a schematic diagram of a 10kV distribution network simulation system.

[0046] Figure 6 This is a schematic diagram illustrating the situation when inrush current is superimposed on load current.

[0047] Figure 7 This is a schematic diagram illustrating the situation during a short-circuit fault.

[0048] Figure 8 This is a schematic diagram illustrating the situation when a short-circuit fault occurs after the inrush current. Detailed Implementation

[0049] The flowchart of the current protection method for preventing inrush current maloperation based on current sequence shaping is as follows: Figure 1 As shown, the specific steps include:

[0050] Step 1: Inrush Current Identification Start Criteria

[0051] To address the issue of inrush current generated by transformer energizers being superimposed on line load current, causing load current distortion and potentially affecting overcurrent protection maloperation, a proposed method is to utilize an Euclidean distance similarity algorithm to construct an inrush current identification and activation criterion after traditional protection activation. The real-time calculation of the current mutation sequence Δi(n)=(Δi1,Δi2,…,Δi n Based on the extreme values ​​of the 1 / 4 cycle data window, a standard power frequency sinusoidal sequence y(n) = (y1, y2, ..., yn) is constructed. n The Euclidean distance similarity B between the current mutation sequence and the standard power frequency sinusoidal sequence is calculated as shown in Equation (1).

[0052]

[0053] Where k is the number of sampling points in a 1 / 4 cycle data window. The start criterion is B. set B set For the similarity threshold, considering both margin and accuracy, a value of 0.8 to 0.9 is recommended. When the activation criterion is met, it is determined that inrush current has occurred and the protection is locked; when B>B set When a short-circuit fault is detected or the circuit is closed under no-load with a fault present, the protection is activated. It is important to note that even in the presence of interference, the maximum similarity value and threshold B are selected. set By making comparisons, the reliability of the judgment results can be ensured.

[0054] Step 2: Second harmonic characteristics of inrush current superimposed on load current:

[0055] The inrush current has a discontinuity angle, and when superimposed on the load current signal, it will not cause waveform distortion in this section. The protection device uses the FFT algorithm to extract the second harmonic through a sliding data window to obtain the second harmonic characteristics of "peak-smooth-peak", where the smooth section corresponds to the waveform non-distortion section. ​

[0056] like Figure 2 As shown, when the transformer is closed under no-load conditions, at t a Inrush current begins to appear at a certain time, superimposed on the load current, causing distortion of the original current signal. FFT calculation is performed on the current signal using a sliding data window. When there is no inrush current in the data window (t... a Before a certain point in time, due to the extremely low second harmonic content, the calculated second harmonic amplitude is close to 0. As the data window moves, the occurrence of inrush current causes the second harmonic amplitude to begin to increase until the inrush current reaches its peak value, after which the second harmonic amplitude shows a decreasing trend, such as... Figure 2 t in a ~t b Time period. Because the inrush current has an intermittent angle characteristic, its superposition on the load current signal will not cause waveform distortion, such as... Figure 2 t in b ~t c During this time period, the sliding data window calculation keeps the second harmonic amplitude near a steady-state value, resulting in a period of flattening. When the inrush current continues to superimpose on the load current, the second harmonic amplitude will increase again, such as... Figure 3 t in c ~t d Time period. Ultimately, the second harmonic characteristic of "peak-flat-peak" is obtained.

[0057] Step 3: For the second harmonic RMS value sequence, calculate the root mean square error σ(a,b) of the segment with sequence number (a,b), and determine the boundary of the undistorted segment in the current signal by threshold comparison:

[0058] (1): For the second harmonic effective value sequence, take the real-time calculated sequence number as b, and take ba = N / 10 as the initial segment, where N is the number of sampling points per cycle, and calculate the root mean square error of segment (a, b). I 2rms (k) is the effective value of the second harmonic. Let be the mean of the second harmonic effective values ​​within the interval (a,b).

[0059] (2): If σ(a,b)≥σ set , σ set The recommended value is the preset threshold. Repeat operation (1).

[0060] (3): If σ(a,b)<σ set The segment is determined to be a stable segment. Waiting for the next sampling time, point b is updated, b = b + 1 is set, σ(a,b) is recalculated, and it is determined that σ(a,b) < σ setIf the condition is met, repeat operation (3) until σ(a,b)≥σ is found. set The current (a,b) segment is determined to be the boundary of the non-distorted segment in this calculation process.

[0061] (4): To Figure 2 The segmentation results of the inrush current superimposed on the load current signal are as follows: Figure 3 As shown. Figure 3 middle Figure 3 (a) is the original current signal waveform, in which the distorted section and the sinusoidal section alternate repeatedly. Due to the attenuation characteristics of the inrush current, the distortion degree of the distorted section gradually weakens. At the same time, due to the superposition effect of the load current, there is no longer a discontinuity angle in the waveform. Figure 3 (b) is the real-time calculated RMS value of the second harmonic, and the mean square error σ of the calculated RMS value of the second harmonic is as follows: Figure 3 As shown in (c), the stable segments for each cycle are identified using the method described above, as marked by the red solid lines in the figure. The original current signal is then segmented using the boundary points of these stable segments. Figure 3 (d) is the current waveform after retaining the sinusoidal section. It can be seen that the stable section of its second harmonic effective value corresponds to the sinusoidal part of the original current signal.

[0062] Step 4: Construct a standard sine curve template. Calculate the Pearson correlation coefficient between the undistorted current waveform and the standard curve template. Find corresponding similar segments between the two, and repair the distorted segments of the actual current waveform based on the standard curve until the shaping end criterion is met, at which point the repair stops. The implementation process is as follows:

[0063] (1) Determination of standard sine curve: Construct a standard power frequency sine curve with an amplitude of 1 as a template, and define the continuous 1 / 4 cycle segment within 1 cycle as regions I, II, III and IV respectively.

[0064] (2) In the real-time sampled current (a,b) data window, based on the monotonic trend and the sign of the first point of this part of the curve, the starting region is located in the standard curve. Based on the starting region sequence and the actual current waveform (a,b) segment sequence, the Pearson correlation coefficient ρ(x,y) is calculated as shown in equation (2). Let the standard current sequence x = {x1,x2,…,x} n The actual current sequence y = {y1, y2, ..., y} n}, where n = b - a + 1.

[0065]

[0066] ρ(x,y) is between [-1,+1], where +1 indicates that the two signals are perfectly linearly positively correlated; -1 indicates that the two signals are perfectly linearly negatively correlated. The closer the value of ρ(x,y) is to 1, the stronger the correlation between the two signals, and vice versa.

[0067] Similarity calculation using a sliding adaptive data window reveals that the trends of change between the two sequences are extremely similar only when they are close to or overlap with the original sinusoidal segment sequence, in which case the Pearson correlation coefficient ρ(x,y)≈1; otherwise, the Pearson correlation coefficient ρ(x,y)<1. Therefore, a threshold ρ can be set... set , to exceed the threshold ρ set The first data window is used to identify standard sinusoidal segments that exhibit trends highly similar to the original sinusoidal segment sequence. Considering accuracy and computational error, ρ... set Set to 0.995. When ρ(x,y)≥ρ set At that time, determine the standard curve segment corresponding to the (a,b) segment, and calculate the numerical multiple of the standard curve.

[0068] (3) Shaping and Repairing Distorted Sections: In the data window of each cycle, the boundary endpoint b of the original sine segment is taken as the endpoint of the data window. Since the original sine segment has n sampling points, the remaining (Nn) sampling points are the distorted segments that need to be repaired. The repair curve sequence of the distorted segments is determined according to equation (3).

[0069]

[0070] Based on the starting point k of the standard sine segment st Let i(a) = K·y(k) be the boundary starting point a of the original sine segment. st The initial phase of the repair curve sequence is determined by equation (4).

[0071]

[0072] Following this process, the current sequence of each cycle is shaped and repaired, and the shaped and repaired current signal is sent to the traditional overcurrent protection.

[0073] (4) Criterion for ending the shaping process: Due to the attenuation characteristics of the inrush current, the effective value of the second harmonic also decreases. When the inrush current attenuates to a certain extent, it will cause the stable section of the effective value to become longer, which has little impact on the line protection. Therefore, a certain threshold can be set to characterize that the inrush current content is very low, thereby ending the shaping and repair of the original current signal. According to the non-distortion section boundary judgment method, the shaping operation ends when the length n of the adaptive data window satisfies equation (5), where nset For adaptive data window threshold, a value of 0.75N is recommended.

[0074] n>n set (5).

[0075] (5) Figure 3 Taking the first stable section after the inrush current occurs as an example, the corresponding sinusoidal section contains 35 sampling points. Therefore, the length of the adaptive data window is taken as n = 35. Since the first point data i(k) in the data window... a Since the value is less than 0, and the first three data points show an increasing trend, the standard sine curve is determined as follows: like Figure 4 As shown in (b). The results of similarity calculation using the Pearson correlation coefficient are as follows. Figure 4 As shown in (c), at 0 ms, ρ = 0.9975 > 0.995, satisfying the similarity condition. Record the value of k at this time. st =0, and the values ​​of each point within the adaptive data window, the average value multiple K is calculated to be 0.248. When performing reshaping and repair on the distorted segment, using a = 45, y(k st = -1, and the initial phase of the repair curve sequence is calculated according to equation (4). Finally, the repair curve sequence of the distorted segment was obtained as follows: The current sequence after distortion section repair is as follows Figure 4 As shown in (d).

[0076] Step 5: Use the current data after the reshaping and repair to make a judgment using traditional protection, so as to avoid false tripping of current protection due to the presence of inrush current and improve the reliability of current protection.

[0077] Step Six: Use PSCAD software to build such a structure. Figure 5 The 10kV distribution network simulation system shown aims to verify the effectiveness of the proposed method. In this system, all distribution transformers use the Dyn11 connection group with a transformation ratio of 10 / 0.4kV and a transformer capacity of 6MVA; the active load per phase is 1MW, and the reactive load per phase is 0.15MVar; the line length between adjacent nodes is set to vary from 3 to 10km; the system frequency is 50Hz, and the sampling frequency is set to 4kHz, i.e., 80 sampling points per cycle.

[0078] (1) Simulation test of inrush current superimposed on load current. For example... Figure 6 As shown, Figure 6 (a) To protect the original measured current signal, the maximum similarity value B = 0.518 was calculated using the Euclidean distance algorithm. set ​This satisfies the inrush current identification criteria and blocks the current protection. Simultaneously, current sequence shaping technology is immediately activated to calculate the stable sections of each cycle based on the effective value of the second harmonic of the original current signal. Figure 6 (b) The red solid line portion. The original current signal is then segmented using boundary sampling points in the stable region, retaining only the current waveform of the sinusoidal region, as shown in [example image]. Figure 6 As shown in (c), the retained current signal is shaped and repaired using current sequence shaping technology, and the standard sine curve is determined as follows. Within a one-cycle data window, the similarity calculation using the Pearson correlation coefficient shows that at 4.8 ms, ρ = 0.9964 > 0.995, and the average numerical multiple K = 0.362. When performing reshaping and repair on distorted segments, using a = 45, y(k... st = -0.998, and the initial phase of the repair curve sequence is calculated according to equation (4). Finally, the repair curve sequence of the distorted segment was obtained as follows: The current sequence after distortion section repair is as follows Figure 6 As shown in (d). Subsequent cycles are also repaired using current sequence shaping technology to correct the distorted sections, and finally the shaped and repaired current waveform is sent to the conventional protection system for judgment.

[0079] (2) Simulation testing of short-circuit faults. For example... Figure 7 As shown, Figure 7 (a) To protect the original current signal measured, Figure 7 (b) represents the current abrupt change and the effective value of the fundamental frequency. Figure 7 (b) It can be seen that the sudden change in current rapidly exceeds the protection setting I in the early stage of a phase-to-phase fault. set =0.8, and the maximum similarity value B = 1 > B is obtained by calculating using the Euclidean distance algorithm. set If a short circuit fault or fault-prone closing occurs at this time, the current protection is directly activated to ensure the fast operation of the current protection.

[0080] (3) Simulation test of short-circuit fault occurring after inrush current. For example... Figure 8 As shown, Figure 8 (a) To protect the original measured current signal, the Euclidean distance algorithm calculates the maximum similarity value B = 0.563 at this point. set The inrush current identification criterion is met, and the current protection is locked. The lockout is released because the fundamental effective value begins to increase at 0.224 ms, but the sudden current change once again rapidly exceeds the protection setting I. set =0.8, and then the maximum similarity value B = 0.865 is obtained by calculating using the Euclidean distance algorithm. set ​The system determines that a short-circuit fault has occurred or that the circuit is being closed with a fault, and therefore directly activates the current protection. During the current protection blocking period, current sequence shaping technology is initiated to calculate the stable sections of each cycle based on the effective value of the second harmonic of the original current signal. Figure 8 The red solid line portion in (b) is then used to segment the original current signal using boundary sampling points in the stable region, retaining only the current waveform of the sinusoidal region of the signal, as shown in [example image]. Figure 8 As shown in (c). After reshaping and repair, the final current sequence is as follows. Figure 8 As shown in (d), during the current protection blocking period, the shaped and repaired current signal is sent to the conventional overcurrent protection for judgment to ensure that the conventional overcurrent protection does not malfunction.

Claims

1. A current protection method for preventing inrush current maloperation based on current sequence shaping, characterized in that... Includes the following steps: Step 1: Calculate the current surge sequence in real time, and construct a standard power frequency sinusoidal sequence based on the extreme value of the 1 / 4 cycle data window. Compare the Euclidean distance similarity between the two sequences to form a criterion for rapid start-up of excitation inrush current. Step 2: Use the FFT algorithm to extract the second harmonic characteristics of the current sampling sequence in a "peak-flattening-peak" pattern through a sliding data window; Step 3: For the second harmonic RMS value sequence, calculate the root mean square error of the segment with sequence number (a, b). By comparing thresholds, the boundaries of the undistorted segments in the current signal are determined. Step 4: Construct a standard sine curve template, use the current waveform in the undistorted section to calculate its Pearson correlation coefficient with the standard sine curve template, find the corresponding similar sections, and repair the distorted sections of the actual current waveform based on the standard curve until the shaping end criterion is met and the repair stops. Step 5: Substitute the current data after the reshaping and repair into the traditional protection judgment to realize the current protection function.

2. The current protection method for preventing inrush current maloperation based on current sequence shaping according to claim 1, characterized in that: In step 1, the rapid start-up criterion for inrush current includes the following steps: S1.1: Real-time calculation of current mutation sequence , These represent the values ​​at each point in the current surge sequence; and a standard power frequency sinusoidal sequence is constructed based on the extreme values ​​of the 1 / 4 cycle data window. ; These represent the values ​​at each point in the standard power frequency sinusoidal sequence; S1.2: Calculate the Euclidean distance similarity between the current abrupt change sequence and the standard power frequency sinusoidal sequence. B As in equation (1): （1）； in: This represents the number of sampling points for a 1 / 4 cycle data window; This indicates the sequence number of the sampled data in the 1 / 4 cycle data window; S1.3: Startup Criterion , This is the similarity threshold.

3. The current protection method for preventing inrush current maloperation based on current sequence shaping according to claim 1, characterized in that: In step 2, the inrush current is superimposed on the second harmonic characteristics of the load current. The inrush current has a discontinuity angle, and its superposition on the load current signal will not cause waveform distortion in this section. The protection device uses the FFT algorithm to extract the second harmonic through a sliding data window to obtain the second harmonic characteristics of "peak-smooth-peak". The smooth section corresponds to the non-distorted waveform section.

4. The current protection method for preventing inrush current maloperation based on current sequence shaping according to claim 1, characterized in that: Step 3 includes the following steps: S3.1: For the second harmonic RMS value sequence, take the sequence number calculated in real time as b, and take... ba=N / 10 is the initial segment, where: N Given the number of sampling points per cycle, calculate the root mean square error of segment (a, b). , This is the effective value of the second harmonic. The mean effective value of the second harmonic within the interval (a,b); S3.2: If it occurs , The preset threshold value is 0.

02. Repeat S3.1; S3.3: If The segment is determined to be a stable segment. Waiting for the next sampling time, point b is updated by setting b = b + 1 and recalculating. And determine If the condition is met, repeat step S3.3 until it occurs. The current (a,b) segment is determined to be the boundary of the non-distorted segment in this calculation process.

5. The current protection method for preventing inrush current maloperation based on current sequence shaping according to claim 1, characterized in that: Step 4 includes the following steps: S4.1: Construct a standard power frequency sine curve with an amplitude of 1 as a template, and define the continuous 1 / 4 cycle segments within 1 cycle as regions I, II, III, and IV respectively; S4.2: In the (a,b) segment of the real-time sampled current, based on the monotonic trend and the sign of the first point of this part of the curve, locate the starting region in the standard curve. Based on the sequence of the starting region and the sequence of the (a,b) segment of the real-time sampled current, as shown in Equation (2), calculate the Pearson correlation coefficient. Assume a standard current sequence , These represent the numerical values ​​at each point in the standard current sequence; the actual current sequence. , These represent the numerical values ​​at each point in the actual current sequence; where the length of the undistorted segment is... n = b - a + 1 ; （2） Indicates the sequence number is i Standard current sequence values, This represents the value at each point in the calculation of the mean of a standard current sequence. Indicates the sequence number is i The actual current sequence values, This represents the value at each point in the calculation of the mean of the actual current sequence; By setting a threshold Let it be 0.995; when it satisfies Determine the standard curve segment corresponding to the (a,b) segment, and calculate the numerical multiple of the standard curve. ; S4.3: The repair curve sequence of the distorted segment is determined according to equation (3). （3）； This represents the number of sampling points within one cycle; Based on the starting point of the standard sine segment Boundary starting point of the original sine segment a ,make The initial phase of the repair curve sequence is determined by equation (4). ; （4） Following this process, the current sequence of each cycle is shaped and repaired, and the shaped and repaired current signal is sent to the traditional overcurrent protection. S4.4: According to the method for determining the boundary of the non-distorted segment, when the length of the non-distorted segment... n The shaping operation ends when equation (5) is satisfied, where: For threshold; （5）。 6. A method for determining the boundary of the undistorted section of a current signal, characterized in that... Includes the following steps: A1: For the second harmonic RMS value sequence, taking the sequence number calculated in real time as b, take... ba=N / 10 is the initial segment, where: N Given the number of sampling points per cycle, calculate the root mean square error of segment (a, b). , This is the effective value of the second harmonic. The mean effective value of the second harmonic within the interval (a,b); A2: If it appears , The preset threshold value is 0.

02. Repeat A1; A3: If The segment is determined to be a stable segment. Waiting for the next sampling time, point b is updated by setting b = b + 1 and recalculating. And determine To check if the condition is met, repeat step A3 until a result is found. The current (a,b) segment is determined to be the boundary of the non-distorted segment in this calculation process.

7. A current sequence shaping method based on Pearson correlation coefficient, characterized in that... Includes the following steps: a1: Construct a standard power frequency sine curve with an amplitude of 1 as a template, and define the continuous 1 / 4 cycle segment within 1 cycle as regions I, II, III, and IV respectively; a2: In the real-time sampled current (a,b) data window, based on the monotonic trend and the sign of the first point of this part of the curve, the starting region is located in the standard curve. Based on the sequence of the starting region and the sequence of the actual current waveform (a,b) segments, as shown in Equation (2), the Pearson correlation coefficient is calculated. Assume a standard current sequence , These represent the numerical values ​​at each point in the standard current sequence; the actual current sequence. , These represent the numerical values ​​at each point in the actual current sequence; where the length of the undistorted segment is... n = b - a + 1 ; （2） Indicates the sequence number is i Standard current sequence values, This represents the value at each point in the calculation of the mean of a standard current sequence. Indicates the sequence number is i The actual current sequence values, This represents the value at each point in the calculation of the mean of the actual current sequence; By setting a threshold Let it be 0.995; when it satisfies Determine the standard curve segment corresponding to the (a,b) segment, and calculate the numerical multiple of the standard curve. ; a3: The repair curve sequence of the distorted segment is determined according to equation (3). （3）； This represents the number of sampling points within one cycle; Based on the starting point of the standard sine segment Boundary starting point of the original sine segment a ,make The initial phase of the repair curve sequence is determined by equation (4). ; （4） Following this process, the current sequence of each cycle is shaped and repaired, and the shaped and repaired current signal is sent to the traditional overcurrent protection. a4: According to the method for determining the boundary of the non-distorted segment, when the length of the non-distorted segment... n The shaping operation ends when equation (5) is satisfied, where: For threshold; （5）。

Images