A power transmission line fault single-ended traveling wave distance measurement method based on waveform dominant feature simulation inference

Abstract

The application relates to a power transmission line fault single-end traveling wave distance measurement method based on waveform dominant feature simulation inference. A traveling wave device is used to collect line single-end current traveling wave signals, a simulation model of the same bus type is established to obtain an offline simulation sample library, fault traveling wave data and corresponding line simulation samples are read; an improved mode distance similarity metric is constructed, fault recording wave data and simulation data in the simulation sample library are compared and selected, and simulation samples closest to the fault recording wave data are obtained; a wave head prompt interval of the determined closest simulation sample and a corresponding distance measurement formula are read, existing wave head calibration technology is used to pre-calibrate measured data, whether the pre-calibrated wave head is in the prompt interval and whether the polarity is consistent with the simulation wave head are judged to determine whether the measured wave head calibration is effective; for the pre-calibrated wave head, the arrival time difference between the wave head and the first wave head is substituted into the distance measurement formula provided by the closest simulation sample to perform calculation, a distance measurement sequence is obtained, and the distance measurement is judged through the standard deviation of the fault distance in the sequence.

Description

Technical Field

[0001] This invention relates to a method for single-end traveling wave fault location in transmission lines based on waveform dominant feature simulation inference, belonging to the field of power system relay protection. Background Technology

[0002] Currently, with the rapid development of power systems, high voltage, long distance, large capacity, AC / DC hybrid interconnection, and regional power grid integration have become a reality. Single-ended current traveling wave recording devices for transmission lines are widely used in 110kV and above voltage levels due to their economic efficiency, high speed, and insensitivity to system oscillations. Therefore, accurate and timely fault location and troubleshooting after a fault occurs play a crucial role in improving power supply reliability.

[0003] The key to traveling wave ranging lies in the accurate identification of the initial wavefront, the reflected wave from the fault point, or the reflected wave from the opposite bus. However, current single-ended traveling wave ranging is affected by factors such as transition resistance, channel attenuation, noise interference from the recording line channel, and zero-mode components, making automatic calibration of the traveling wavefront difficult in practical applications and hindering the automatic determination of whether there are marking errors. Simulated samples, on the other hand, are regular, free from noise interference, have obvious traveling wave dominant characteristics, and have known fault locations, which can enhance the dominant characteristics of the traveling wavefront and suppress noise interference. Therefore, this invention provides a wavefront calibration method that uses similarity measurement to select the most similar simulated sample to the measured data and incorporates the prompts from the simulated sample. This method can increase the number of measurable wavefronts, reduce erroneous wavefront calibrations from measured data, and use multiple identified wavefronts to verify the correctness of the ranging results. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a single-end traveling wave ranging method for transmission line faults based on waveform dominant feature simulation inference. This method can effectively eliminate erroneously calibrated wavefronts in the current measured data for ranging and can detect and verify the correctness of the measured fault distance. This solves the technical problem that wavefront calibration is difficult to automatically verify in previous measured data.

[0005] The technical solution of this invention is: a method for single-end traveling wave ranging of transmission line faults based on waveform dominant feature simulation inference, the specific steps of which are as follows:

[0006] Step 1: Read the traveling wave recording data of the fault phase current and the corresponding offline simulation sample set A of the fault line. m A simulation model with the same topology and length as the local line and two types of transition resistance (high resistance and low resistance) was established. The entire length of the transmission line was traversed with faults occurring at 2km intervals to obtain an offline fault sample library for the line.

[0007] Step 2: Construct an improved pattern distance similarity measurement algorithm, compare fault waveform data and simulation data in the simulation sample library, and obtain the simulation sample that is most similar to the measured fault data.

[0008] Step 3: Read the simulation sample prompt interval R i =[R is R ie The measured wavefront sequence t is obtained by pre-calibrating the measured data using existing wavefront calibration techniques and the corresponding ranging formula. The validity of the pre-calibrated measured wavefront is determined by checking whether the wavefront of the measured data is within the simulation indication range and whether the polarities of the two are consistent.

[0009] Step 4: For each hint interval, if t exists in t... u Satisfy R is <t u <R ie , where Δt i =t u -t0(t u Substituting (≠t0), into the corresponding ranging formula, we obtain the result from the fault distance x. k The range measurement sequence is composed of x in the sequence. k The quantity and sequence standard deviation are used to determine whether the ranging results are valid.

[0010] Step 1 specifically refers to:

[0011] Step 1.1: Collect data using a traveling wave recorder and record the current traveling wave data for more than 2ms before and after the fault.

[0012] Step 1.2: In PSCAD / EMTDC, build a simulation model with the following topology: adjacent busbars at both ends have the same category, the faulted line length is consistent, and the transition resistance is either high or low. The fault angle, the number of outgoing circuits of the busbar line, and the fault type do not change the polarity and abrupt change time of the traveling wave, and can be set to constant values.

[0013] Step 1.3: The simulation model traverses the entire length of the faulty line at 2km intervals to perform batch simulations, resulting in two sets of offline simulation sample libraries numbered in order of fault distance.

[0014] Step 2 specifically includes:

[0015] Step 2.1: Improved Pattern Distance. By calculating the slopes of two time series, the time series are divided into m segments according to the sign and magnitude of the slopes. Each segment has a pattern set of {sharp increase, sharp decrease}, and its corresponding pattern is represented as c = {1, -1}. The segmented linearized time series is as follows:

[0016] S={(t1,y1s ,y 1l ),…(t i ,y is ,y ie ),…(t m ,y ms ,y me )}(1)

[0017] In the formula, y is y ie Let t represent the starting and ending values ​​of the i-th segment of the sequence, respectively. i Let be the initial time of the i-th segment of the sequence.

[0018] Step 2.2: The initial point of the first mutation (t1, c1), followed by multiple pattern combinations based on the corresponding pattern (sharp rise or sharp fall) and the corresponding time of each segment in chronological order:

[0019] S={(t1,c1),(t2,c2),…(t i ,c i ),…(t m ,c m )}(2)

[0020] In the formula, c i ∈c, representing the pattern corresponding to the nth segment. We can obtain the pattern and time corresponding to each sequence segment, and take the i-th segment (t) corresponding to two sequences. i c i We will conduct an analysis and comparison.

[0021] Step 2.3: Based on the measured data, only the first three most significant time series segments in the simulation and measured data exhibit significant macroscopic morphological characteristics shortly after the fault occurs. The first wave head arrives first, is the strongest and most significant, and has no multipath interference, so it is used as the benchmark. After aligning with the first wave head, the polarity and time difference corresponding to the two remaining most significant time series segments are compared. The improved mode distance between the two time series is calculated using equation (3), and the mathematical expression is:

[0022]

[0023] In the formula, S s ,S a Let f(c) be the time series of measured and simulated data. i ), f(t) i α represents the polar distance and temporal distance corresponding to each segment of the pattern sequence, respectively. i ,β i For the weights. Where f(c) i )=|c ai -c bi|, c={1,-1} characterizes the polarity of the traveling wave. When the polarities of two time series are the same, the polarity distance is 0. When they are opposite, the polarity distance is 2. The time distance and polarity distance are set to the same order of magnitude, and the time difference is linearized within the range [0,2] with a saturation cutoff, i.e.:

[0024]

[0025] In the formula, Δt is the time difference between the end of the segments corresponding to the two time series, and Δt is less than t. l At that time, the time distance is 0. (The time is greater than t.) h At that time, the time distance is 2. Between t... l and t h At that time, the time distance is determined by a linear function generated from the time threshold and the distance threshold.

[0026] Step 2.4: Sort the improved model distances and select the smallest D. m D m Less than the pattern distance threshold D th When this is the case, the m-th sample can be determined as the nearest neighbor simulation sample. When D m Greater than the improved mode distance threshold D th If there is no most similar sample in the sample library, the reported data is considered special data.

[0027] Step 3 specifically refers to:

[0028] Step 3.1: Read the prompt interval R of the nearest neighbor simulation sample i And the corresponding distance measurement formula.

[0029] Step 3.2: Pre-calibrate the fault current t to be measured u and polarity p u This yields N traveling wave front sequences t = [t0, t1…t2]. N ], t u ∈t. Step 3.3: Determine t u Is it within the indicated interval R? i The discriminant is:

[0030]

[0031] That is, the polarity of the measured data and the simulated sample are consistent, and the measured data calibration wavefront is within the R range indicated by the simulated sample. i =[t i -ε,t i The value of +ε] indicates that the measured calibration wavefront is valid.

[0032] Step 4 specifically refers to:

[0033] Step 4.1: Calibrate the fault wavefront t based on the simulation sample feedback. u Based on the distance measurement formula provided by the simulation sample, the single-ended distance measurement principle is used to measure the distance and obtain x. i .

[0034] Step 4.2: The measured fault distance sequence x can be verified by calculating the standard deviation of multiple fault distances. i The accuracy of the wavefront ranging is as follows. When i is greater than 2, the standard deviation σ is calculated according to formula (6). The threshold of σ is set to η. When σ < η, the difference between the elements of the ranging sequence is small, so the ranging is correct. The empirical value of η can be 3. Standard deviation of wavefront ranging:

[0035]

[0036] In the formula, x i The distances to the i faults for the (i+1)th wavefronts identified are given by u = x. i The average value, N is x i Number.

[0037] Step 4.3: Automatically verify the fault data based on the number of wavefronts and the standard deviation σ. When i ≥ 2 and the standard deviation σ < ε, the fault data passes the verification, the ranging is correct, and the ranging result x1 is output. When the standard deviation σ < ε or i < 2, the fault data has too few identified wavefronts or the fault sequence elements differ significantly, so the ranging fails the verification, the ranging result x1 is output, and the data is transferred to a human for further verification of the ranging accuracy.

[0038] The beneficial effects of this invention are: it is simple and efficient; once the offline simulation sample library is established, it can be continuously applied until the faulty line is expanded or rebuilt; this method can detect and verify the correctness of the measured fault; by incorporating simulation sample prompts, the number of wavefronts available for ranging can be increased; multiple ranging distances can be obtained from multiple wavefronts obtained through prompts, thereby verifying the accuracy of the ranging; simultaneously, the nearest neighbor simulation sample can serve as an auxiliary prompt for manual verification. This method is of great significance in solving the current problems of difficulty in determining the correctness of the calibrated wavefront and the low reliability of the calibrated wavefront in single-ended traveling wave ranging. Attached Figure Description

[0039] Figure 1 This is a schematic diagram of the 220kV simulation model construction of the present invention;

[0040] Figure 2 These are the specific implementation steps included in step 1 of the present invention;

[0041] Figure 3 These are the specific implementation steps included in step 2 of the present invention;

[0042] Figure 4This refers to the improved mode distance of the offline simulation sample library in step 2 of embodiment 2 of the present invention;

[0043] Figure 5 It refers to the test data and the nearest neighbor simulation sample in step 2 of embodiment 2 of the present invention;

[0044] Figure 6 These are the specific implementation steps included in step 3 of the present invention;

[0045] Figure 7 This refers to the prompting range of simulation data in step 3 of embodiment 1 of the present invention;

[0046] Figure 8 This refers to the pre-calibration of the test data in step 3 of embodiment 1 of the present invention;

[0047] Figure 9 This is the correct calibration result after incorporating simulation prompts in step 3 of embodiment 1 of the present invention;

[0048] Figure 10 This refers to the specific implementation steps included in step 4 of the present invention. Detailed Implementation

[0049] The present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0050] Example 1: In PSCAD / EMTDC, a simulation model is built using a topology that ensures adjacent busbars at both ends have the same category and the faulty line has the same length. The simulation model is as follows: Figure 1 As shown. A traveling wave recorder was used to synchronously acquire data at a sampling rate of 1MHz, recording traveling wave data for at least 2ms before and after the fault. Fault recording data for a 220kV transmission line in a certain area was obtained from the recorder as measured data. The specific implementation steps are as follows:

[0051] Step 1: Establish a simulation model based on the topology of a 220kV transmission line in a certain area, change the parameters to conduct batch simulations to obtain an offline simulation sample library. The specific implementation steps are as follows: Figure 2 As shown.

[0052] Step 1.1: Collect data using a traveling wave recorder and record the current traveling wave data for more than 2ms before and after the fault.

[0053] Step 1.2: Establish a simulation model with the same bus type as the transmission line, consistent line length, and transition resistance divided into high resistance and low resistance. Since the fault angle, number of outgoing circuits of the bus line and fault type will not change the polarity of the traveling wave and the time of sudden change, typical values ​​can be arbitrarily selected.

[0054] Step 1.3: Establish offline simulation samples for this line. The line is 93km long. Two sets of simulation samples are generated in batches with transition resistance of 200Ω and 50Ω respectively and fault location traversal step size of 2km. Each set contains 46 samples.

[0055] Step 2: Calculate the mode distance between each sample and the measured fault data in the offline simulation sample library for the same faulty line, and select the nearest neighbor simulation sample. The specific implementation steps are as follows: Figure 3 As shown.

[0056] Step 2.1: Calculate the slope of both the measured and simulated data. Divide the time series into m segments based on the sign and magnitude of the slope. This yields the piecewise linearized time series S. s S a .

[0057] Step 2.2: The first mutation moment starts at the point (t1, c1), and then multiple pattern combinations are formed by the corresponding pattern (sharp rise or sharp fall) and the corresponding time of each segment in chronological order.

[0058] Step 2.3: After aligning the simulated and measured waveforms according to their first wave heads, compare the polarity and time difference corresponding to the two most significant remaining sequence segments. Calculate the distance between the two time series improved modes of the 46 simulated samples and the measured data sequentially using equation (3), where D... tl The value is 1, and in equation (4), the weights are α2 = 0.25, α3 = 0.25, β2 = 0.27, and β3 = 0.23. The offline simulation sample library mode distance is as follows: Figure 4 As shown.

[0059] Step 2.4: Sort the two sets of improved mode distances. It can be seen that the smallest improved mode distance in the sample library is for sample number 24 when the transition resistance is 50Ω. The polarity distance in the improved mode distance is 0, and the time distance is 3.86 * 10. -5 The test data and the nearest neighbor simulation samples are as follows Figure 5 As shown.

[0060] Step 3: Read the nearest neighbor sample wavefront indication interval R i And the corresponding ranging formula. The N traveling wavefronts of the fault current to be measured are pre-calibrated as t = [t0, t1…t]. n ], t u ∈t, determine t by equation (5) u Is it within the sample hint interval R? i In the middle, and t u With R i Whether the polarity is consistent is used to determine whether the pre-calibrated measured wavefront is effectively calibrated. The specific implementation steps are as follows: Figure 6 As shown.

[0061] Step 3.1: Read the nearest neighbor sample wavefront indication interval R3 as shown in Table 3, and the corresponding ranging formula (7). The nearest neighbor simulation sample interval boundary is taken as the traveling wave transmission time corresponding to 10% of the total line length, i.e., 9.3 km, as the tolerance, i.e., 31 μs. The nearest neighbor simulation sample indication interval is calibrated as follows: Figure 7 As shown.

[0062] Table 3 Simulation prompt range

[0063]

[0064]

[0065] In the formula, v is the traveling wave velocity of the fault current, which is taken as the speed of light. Δt can be obtained from the interval indicated by the simulation sample. zi =t zi -t z0 , Δt fi =t fi -t z0 , t zi t fi Let represent the average values ​​of the i-th interval with positive and negative polarities, respectively.

[0066] Step 3.2: The measured data are calibrated using existing wavefront calibration methods, resulting in five measured calibration wavefronts, as shown in Table 4. The pre-calibration of the data to be measured is as follows: Figure 8 As shown.

[0067] Table 4. Polarity and timing obtained from wavehead pre-calibration of the data to be tested.

[0068]

[0069]

[0070] Step 3.3: The test data is evaluated one by one using equation (5). If the polarity of the measured data is consistent with that of the simulated sample, and the calibrated wavefront of the measured data falls within the indicated range of the simulated sample, then the pre-calibration of the test data is valid. The second measured wavefront t can then be obtained. f1 If the data is outside the simulation feedback range, it indicates an incorrect calibration of the measured data using the current calibration method. Wavefront calibration of measured data incorporated into the simulation samples is as follows: Figure 9 As shown.

[0071] Step 4: For the effectively calibrated wavefront, the ranging formula corresponding to the interval indicated by the simulation sample is used for ranging. The validity of the ranging can be verified through the ranging sequence. The specific implementation steps are as follows: Figure 10 As shown.

[0072] Step 4.1: For the measured wavefront sequence t = [t z0, t f2 , t z3 , t z4 ], there exists t u ∈t(t u ≠t z0 ), calculate t respectively using equation (7) u With t z0 The fault distance can be used to obtain the ranging sequence x. i =[47.16km, 48.13km, 47.11km].

[0073] Step 4.2: The measured fault distance sequence x can be verified by calculating the standard deviation of multiple fault distances. i The accuracy of the standard deviation threshold η is empirically set to 3. Equation (6) is used to evaluate the ranging sequence x. i Calculating the standard deviation, we get σ = 0.469.

[0074] Step 4.3: The fault is automatically verified based on the number of wavefronts and the standard deviation σ. The measured data identified 4 wavefronts, and the verification result σ = 0.469 < η. The automatic verification is successful, and the distance measurement is correct. The output distance measurement result is 47.16 km.

[0075] The specific embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited to the above embodiments. Within the scope of knowledge possessed by those skilled in the art, various changes can be made without departing from the spirit of the present invention.

Claims

1. A method for single-end traveling wave fault location in transmission lines based on waveform dominant feature simulation inference, characterized in that: Step 1: Read the traveling wave recording data of the fault phase current and the corresponding offline simulation sample set A of the fault line. m A simulation model with the same line topology and length as the local line and two types of transition resistance, namely high resistance and low resistance, was established. The entire length of the transmission line was traversed with faults occurring at intervals of 2km to obtain an offline sample library of faults for the line. Step 2: Construct an improved pattern distance similarity measurement algorithm, compare fault waveform data and simulation data in the simulation sample library, and obtain the simulation sample most similar to the measured fault data. Step 3: Read the simulation sample prompt interval R i With the corresponding ranging formula, the measured data are pre-calibrated using existing wavefront calibration technology to obtain the measured wavefront sequence t. It is then determined whether the wavefront of the data to be measured is within the simulation prompt range and whether the polarities of the two are consistent. This is used to determine whether the pre-calibrated measured wavefront is valid. Step 4: For wavefronts pre-calibrated in the simulation prompt range, substitute them into the corresponding ranging formula to calculate and obtain a ranging sequence composed of fault distances. Determine whether the ranging result is valid by the number of fault distances in the sequence and the standard deviation of the sequence. Step 2 specifically includes: Step 2.1: Calculate the slope of both the measured and simulated data. Divide the time series into m segments according to the sign and magnitude of the slope, thus obtaining the piecewise linearized time series S. s S a ; Step 2.2: Align the first mutation moment and polarity of the two sets of traveling wave data as the starting point t1, c1, and then form multiple pattern combinations according to the corresponding pattern and time of each segment in chronological order. Step 2.3: After aligning with the first wave head time, compare the polarity and time difference corresponding to the two most significant remaining sequence segments, and calculate the distance between the two time series improved modes of the simulation sample and the measured data; Step 2.4: Sort the improved mode distances, select the smallest mode distance, and use the improved mode distance to search for the nearest neighbor simulation sample in the simulation sample library for this line. Calculate the improved mode distance D between each simulation sample and the fault recording data. m The smallest D is obtained by comparison. m D m Less than the minimum threshold D of the pattern distance tl When the m-th sample is determined, it can be identified as the nearest neighbor simulation sample. Step 3 specifically refers to: Step 3.1: Read the prompt interval R of the most recently connected simulation samples i and the corresponding distance measurement formula; Step 3.2: Pre-calibrate the fault current t to be measured using the existing wavefront calibration algorithm. u and polarity p u This yields N traveling wave front sequences t=[t0,t1…t2] N ], t u ∈t; Step 3.3: Determine t u Is it within the hint interval R provided by the nearest neighbor simulation sample? i In addition, the polarity of the two is consistent, which is used to determine whether the pre-calibrated measured wavefront is calibrated effectively.

2. The single-end traveling wave ranging method for transmission line faults based on waveform dominant feature simulation inference according to claim 1, characterized in that, Step 1 specifically refers to: Step 1.1: Collect data using a traveling wave recorder and record the current traveling wave data for more than 2ms before and after the fault; Step 1.2: In PSCAD / EMTDC, build a simulation model with the following topology: the adjacent busbars at both ends have the same category, the fault line length is consistent, and the transition resistance is either high or low. The fault angle, the number of outgoing circuits of the busbar line, and the fault type will not change the polarity and abrupt change time of the traveling wave, and can be set to constant values. Step 1.3: The simulation model traverses the entire length of the faulty line at 2km intervals to perform batch simulations, resulting in two sets of offline simulation sample libraries with different transition resistances (high resistance and low resistance) numbered in order of fault distance.

3. The method for single-end traveling wave analysis and ranging of transmission line faults based on waveform dominant feature simulation inference according to claim 2, characterized in that, Step 2.2 is specifically defined by the pattern corresponding to each segment as either a sharp rise or a sharp fall.

4. The method for single-end traveling wave analysis and ranging of transmission line faults based on waveform dominant feature simulation inference according to claim 1, characterized in that, Step 4 specifically refers to: Step 4.1: Calibrate the fault wavefront t based on the simulation sample feedback. u Substitute the values ​​into the corresponding single-end distance measurement formula to measure the distance, and obtain x. i ; Step 4.2: Calculate the standard deviation of multiple fault distances; Step 4.3: Automatically verify the fault data based on the number of wavefronts and the standard deviation σ, and output the ranging result obtained by calculating the time difference between the arrival of the first two wavefronts.

Images