A method and device for early warning of distribution network faults based on traveling wave waveform recognition

Abstract

This application relates to the technical field of distribution network fault detection, and discloses a method and device for early warning of distribution network faults based on traveling wave waveform recognition. The method is applied between two paired distribution network traveling wave recognition devices, and the steps include: obtaining distribution network line data of the lines between the devices based on the early warning prompts triggered by traveling wave waveform recognition issued by the devices; extracting associated support position and support wire data from the data; calculating the distance between support poles from the support position data, matching wire sag data based on the support wire data, and then calculating sag loss data; identifying the wire type and corresponding support, matching the wire crossing node and calculating crossing loss data; combining the two types of loss data to calculate the traveling wave waveform loss data, and issuing an early warning if it is greater than a preset reference value. Through multi-dimensional loss analysis, accurate early warning is provided, improving the reliability of distribution network fault early warning.

Description

Technical Field

[0001] This application relates to the technical field of distribution network fault detection, and in particular to a method and device for early warning of distribution network faults based on traveling wave waveform recognition. Background Technology

[0002] In modern power systems, the distribution network directly serves end users, making its safe and stable operation crucial. Distribution network grounding faults, a common type of fault, can cause power outages, impacting power supply reliability and even threatening personal and equipment safety if not detected promptly and accurately. Existing detection methods often combine steady-state analysis and traveling wave detection technology. While steady-state analysis is widely used, it has limitations in complex distribution network environments. Traveling wave detection technology, on the other hand, carries rich fault information in the traveling wave, making it a key direction for improving detection accuracy.

[0003] Existing methods for detecting traveling waves in distribution networks typically involve first installing voltage or current sensors on the line to collect traveling wave signals generated by faults. When a fault occurs, the traveling wave signal propagates along the line at near the speed of light, and the sensor captures it. The collected signal is then transmitted to a data processing unit, where it undergoes preprocessing such as filtering and amplification to remove noise interference. Afterward, specific algorithms are used to extract traveling wave characteristics, such as wavefront arrival time, polarity, and amplitude. Based on the traveling wave propagation speed and the line topology, the fault location is calculated, thus achieving fault localization.

[0004] However, the actual distribution network environment is complex, with numerous interference parameters. For example, the physical shape of the lines in the distribution network gives them inherent unfavorable characteristics, which can cause attenuation and distortion of traveling waves. These interferences distort the acquired traveling wave signals, increasing the difficulty of feature extraction and consequently reducing the accuracy of fault warning and location results, making it difficult to meet the distribution network's demand for precise detection. Summary of the Invention

[0005] To improve the accuracy of distribution network grounding fault detection and provide early warning, this application provides a distribution network fault early warning method and device based on traveling wave waveform recognition.

[0006] Firstly, this application provides a method for early warning of distribution network faults based on traveling wave waveform recognition, employing the following technical solution:

[0007] A method for early warning of distribution network faults based on traveling wave waveform recognition, applied between two paired distribution network traveling wave recognition devices, includes the following steps:

[0008] Based on the distribution network fault warning prompt issued by the distribution network traveling wave identification device, the distribution network line data of the power line between two paired distribution network traveling wave identification devices is obtained, wherein the distribution network fault warning prompt is triggered according to the traveling wave waveform identification.

[0009] Extract the associated support location data and support cable data from the power distribution network line data;

[0010] The distance data between multiple support rods is calculated based on the support position data;

[0011] Based on the support wire data, wire sag data is matched from a preset wire database;

[0012] The sag loss data is calculated based on the distance data between the support rods and the wire sag data.

[0013] The first type of wire and the second type of wire are identified from the support wire data;

[0014] Based on the association, a first type of support and a second type of support are extracted from the support location data. The first type of support is associated with the first type of cable, and the second type of support is associated with the second type of cable.

[0015] The wire crossing nodes are matched based on the spatial orientation of the first type of support and the second type of support;

[0016] The crossing loss data is calculated based on the wire crossing nodes;

[0017] Calculate the travel wave waveform loss data based on the sag loss data and the crossing loss data;

[0018] If the traveling wave waveform loss data is greater than the preset waveform loss reference data, a waveform identification warning will be issued.

[0019] By adopting the above technical solution, the presence of sag changes and cable crossings in the line can increase the location error of the grounding fault point. This method provides early warning for the accuracy of distribution network grounding fault detection through loss analysis of multi-dimensional data: Based on the distribution network fault early warning prompt triggered by traveling wave waveform recognition, distribution network line data is obtained, and support position data and support wire data are extracted from it; by calculating the distance data between support poles and matching wire sag data, sag loss data is obtained, the first and second types of wires and their corresponding supports are identified, and the wire crossing nodes are matched, and then the crossing loss data is calculated; the traveling wave waveform loss data is obtained by combining the two types of loss data, and compared with the waveform loss reference data to achieve early warning for detection accuracy; the physical characteristics of the line that affect the detection accuracy are incorporated into the fault early warning analysis, accurately identifying waveform anomalies caused by sag and wire crossing, improving the reliability and practicality of early warning for fault inspection accuracy.

[0020] Optionally, the step of calculating the sag loss data based on the distance data between the support rods and the wire sag data further includes the following sub-steps:

[0021] The distance data between each of the support rods is grouped in a one-to-one correspondence with the sag data of the wire.

[0022] Calculate the inter-rod loss ratio between the support rod distance data and the preset reference rod distance data;

[0023] Calculate the sag loss ratio between the wire sag data and the preset reference sag data;

[0024] Temporary sag loss data is calculated by weighted average of the inter-bar loss ratio and the sag loss ratio;

[0025] By iterating through the distance data between each set of support rods and the sag data of the wire, multiple temporary sag loss data are obtained, and the sag loss data is calculated based on the multiple temporary sag loss data.

[0026] By adopting the above technical solution, the distance data between support rods and the sag data of the wire are grouped one by one, and the ratio of loss between rods and the ratio of sag loss are calculated separately. The temporary sag loss data is generated by weighted averaging, and then the final sag loss data is obtained by integrating multiple sets of data. This approach considers both the coupling effect of rod spacing and sag in a single set of data and eliminates local errors through comprehensive analysis of multiple sets of data.

[0027] Optionally, the step of calculating the travel wave waveform loss data based on the sag loss data and the crossing loss data further includes the following sub-steps:

[0028] Based on the support position data, the sag position corresponding to the sag loss data and the crossing position corresponding to the crossing loss data are identified; the sag position is the position of the sag point closest to the crossing position and whose sag angle is greater than the preset sag reference angle;

[0029] Calculate the positional distance between the sag position and the crossing position;

[0030] Calculate the crossing loss ratio between the crossing loss data and the preset crossing reference data;

[0031] The travel wave waveform loss data is calculated by weighting the sag loss ratio and the crossing loss ratio.

[0032] If the location distance value is greater than the preset reference distance value, then the weight of the sag loss ratio is 0 and the weight of the crossing loss ratio is 1; otherwise, the weight of the crossing loss ratio is adjusted according to the positive correlation of the location distance value.

[0033] By adopting the above technical solution, the traveling wave waveform loss calculation is optimized by accurately associating the spatial positions of sag and crossing: First, the sag position closest to the crossing position and meeting the angle conditions is located, and the distance between the two positions is calculated. The spatial position correlation is incorporated into the loss analysis, making the traveling wave loss calculation more closely reflect the coupling effect of sag and crossing in the actual propagation path. Then, the weighted weights of the sag loss ratio and the crossing loss ratio are dynamically adjusted based on the positional relationship. For long distances, only the crossing loss influence is retained, while for short distances, the crossing weight is adjusted according to the positive correlation of distance. The weighting mechanism can adjust the parameter weights according to the line characteristics, adapting to different distribution network environments. The dynamic weight adjustment enhances the adaptability to lines with different spatial layouts, which is conducive to improving the accuracy of traveling wave waveform loss data.

[0034] Optionally, the method further includes the following steps:

[0035] Within a predetermined first distance range around the sag point, multiple sag points are identified, and first clustered data of the sag distribution is calculated based on the sag positions within the first distance range.

[0036] A first intermediate value is calculated based on the first aggregated data and a preset first reference data, and the weight of the sag loss ratio is adjusted according to the positive correlation of the first intermediate value.

[0037] By employing the above technical solution, multiple sag points are identified around the sag point, and the first clustered data is calculated. Based on this, the first intermediate value is positively correlated with the weight of the sag loss ratio. The weight of the sag influence is enhanced in high-cluster areas, making the loss calculation more consistent with the actual distribution characteristics. This is beneficial for improving the accuracy of downwave waveform loss assessment in complex terrain and strengthening the sensitivity to anomalies in dense sag areas.

[0038] Optionally, the method further includes the following steps:

[0039] Within a predetermined second distance range around the wire crossing node, multiple wire crossing nodes are identified, and second cluster data of the crossing distribution is calculated based on the crossing positions within the second distance range;

[0040] A second intermediate value is calculated based on the second aggregated data and a preset second reference data, and the weight of the crossing loss ratio is adjusted according to the positive correlation of the second intermediate value.

[0041] By employing the above technical solution, multiple wire crossing nodes are identified around the wire crossing node, and a second clustered data is calculated. Based on this, a second intermediate value is derived to positively adjust the crossing loss ratio weight. The loss weight of densely crossed areas is enhanced, which better reflects the actual traveling wave propagation loss characteristics, improves the accuracy of loss assessment under complex line layouts, and helps strengthen the ability to identify anomalies in densely crossed areas.

[0042] Optionally, the method further includes the following steps:

[0043] The distribution network branch is identified based on the distance data between the support rods and the wire sag data;

[0044] Within a preset local area of ​​the distribution network, calculate the number of branches of the identified distribution network branches;

[0045] If the number of branches exceeds the preset reference number of branches, a distribution network branch early warning will be issued.

[0046] Otherwise, calculate the branch ratio of the number of branches to the reference number of branches, and adjust the weight of the crossing loss ratio according to the positive correlation of the branch ratio.

[0047] By adopting the above technical solution, the distribution network branches are identified based on the pole spacing and wire length, the number of branches in a local area of ​​the distribution network is counted, and an early warning is given when the threshold is exceeded. If the threshold is not exceeded, the weight of crossing loss is adjusted by positively adjusting the branch ratio. The branch density is included in the loss assessment, and the crossing influence weight is enhanced in high-density branch areas, which helps to improve the accuracy of waveform loss calculation under complex network structures.

[0048] Optionally, the method further includes the following steps:

[0049] Calculate the weight ratio of the crossing loss ratio to the weight of the sag loss ratio;

[0050] The impact value of crossing branches is calculated based on the weight ratio and the branch ratio;

[0051] If the influence value of the crossing branch is greater than the preset first reference influence value, then the distribution network traveling wave identification device is added at the distribution network branch;

[0052] If the influence value of the crossing branch is greater than the preset second reference influence value, then the distribution network traveling wave identification device is added next to the wire crossing node;

[0053] The first reference influence value is less than the second reference influence value.

[0054] By adopting the above technical solution, the impact value of crossing branches is calculated based on the weight ratio and the branch ratio. When the first threshold is exceeded, a distribution network traveling wave identification device is added at the distribution network branch. When the second threshold is exceeded, a distribution network traveling wave identification device is added next to the wire crossing node. The comprehensive impact of branches and crossings is quantitatively evaluated, the monitoring density in complex areas is increased in a targeted manner, and the detection conditions for graded response are constructed.

[0055] Optionally, the method further includes the following steps:

[0056] A simulated ground fault signal is applied at the first location on the distribution network line, and the corresponding first traveling wave waveform loss data is calculated.

[0057] A simulated ground fault signal is applied at the second location on the distribution network line, and the corresponding second traveling wave waveform loss data is calculated.

[0058] There is no cable crossing between the first position and the second position. Calculate the first simulated difference between the first traveling wave waveform loss data and the second traveling wave waveform loss data.

[0059] The reference distance value is adjusted based on the positive correlation of the first simulated difference.

[0060] By adopting the above technical solution, the reference distance value is optimized by simulating fault signals: a signal is applied at two locations where no cable crosses, the first simulated difference of the traveling wave waveform loss data is calculated, and the reference distance value is adjusted accordingly; this makes the reference distance more consistent with the actual line loss characteristics, improves the adaptability of the location distance to the weight adjustment, and enhances the accuracy of traveling wave loss calculation.

[0061] Optionally, the method further includes the following steps:

[0062] A simulated ground fault signal is applied at the third location on the distribution network line, and the corresponding third traveling wave waveform loss data is calculated.

[0063] A simulated ground fault signal is applied at the fourth location on the distribution network line, and the corresponding fourth traveling wave waveform loss data is calculated.

[0064] A cable crosses between the third position and the fourth position, and the second simulated difference between the third and fourth traveling wave waveform loss data is calculated.

[0065] The positive correlation coefficient between the location distance value and the crossing loss ratio is adjusted based on the second simulated difference positive correlation.

[0066] By adopting the above technical solution, the parameters are optimized through simulated faults involving cable crossings: the working conditions are applied at two locations with crossings, the second simulated difference of traveling wave waveform loss is calculated, and the positive correlation coefficient between the location distance value and the weight of the crossing loss ratio is adjusted accordingly; this coefficient is made to fit the actual crossing loss characteristics, improving the accuracy of weight adjustment and enhancing the adaptability of traveling wave loss calculation.

[0067] Secondly, this application provides a power distribution network fault early warning device based on traveling wave waveform recognition, which adopts the following technical solution:

[0068] A distribution network fault early warning device based on traveling wave waveform recognition includes a processor, wherein the processor executes the steps of the distribution network fault early warning method based on traveling wave waveform recognition as described in any one of the above claims.

[0069] In summary, this application includes at least one of the following beneficial technical effects:

[0070] By extracting support location and wire data, calculating sag and crossing loss, and conducting comprehensive analysis, the physical characteristics of the line are incorporated into the early warning system, accurately identifying waveform anomalies and improving the reliability and practicality of fault early warning.

[0071] The design incorporates grouped calculations of sag loss and dynamic weight adjustment, combined with spatial location, clustering, and branch density to optimize loss assessment. This approach aligns with the actual propagation of traveling waves, enhances adaptability to complex line layouts, and improves the accuracy of loss data.

[0072] Based on the impact value classification, identification devices are added, parameters are optimized by simulating faults, monitoring of complex areas is strengthened in a targeted manner, and line characteristics are dynamically adapted to further improve detection accuracy and adaptability. Attached Figure Description

[0073] Figure 1 This is a flowchart illustrating the steps of a power distribution network fault early warning method based on traveling wave waveform recognition.

[0074] Figure 2 This is a flowchart illustrating the steps for calculating sag loss data based on the distance data between the support rods and the wire sag data.

[0075] Figure 3 This is a flowchart illustrating the steps for calculating the travel wave waveform loss data based on the sag loss data and the crossing loss data. Detailed Implementation

[0076] The embodiments of this application are described in detail below, and examples of the embodiments are shown in the accompanying drawings.

[0077] In the description of this specification, the references to "certain embodiments," "one embodiment," "some embodiments," "illustrative embodiment," "example," "specific example," or "some examples" refer to specific features, structures, materials, or characteristics described in connection with the described embodiment or example, which are included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0078] This application discloses a distribution network fault early warning method based on traveling wave waveform recognition, mainly applied to distribution network grounding fault detection scenarios. By quantifying and analyzing the losses during the propagation of the fault traveling wave, it achieves early warning for improved fault detection accuracy. This method requires the coordinated operation of two paired distribution network traveling wave recognition devices. These devices can integrate voltage / current sensors, data acquisition modules, and communication modules, enabling real-time monitoring of line traveling wave signals and triggering early warning prompts. (Refer to...) Figure 1 It includes the following steps:

[0079] When a potential fault occurs in a distribution network line, the distribution network traveling wave identification device will capture abnormal traveling wave characteristics through a traveling wave waveform recognition algorithm (such as wavelet transform wavefront detection), thereby triggering a distribution network fault early warning. At this time, based on the early warning, the system automatically obtains the distribution network line data of the power lines between the two paired devices. The distribution network line data includes, but is not limited to, basic data such as line topology information, tower location coordinates, wire type parameters, and laying method.

[0080] Extract related support location data and support wire data from the distribution network line data. The support location data specifically includes the latitude and longitude or relative coordinate information of support structures such as poles and supports. The support wire data includes parameters such as wire type, cross-sectional size, and material. The relationship between the two is reflected in "a specific wire laid for a certain support structure", for example, "pole A-1 corresponds to the erection of LGJ-240 type overhead line".

[0081] Based on the coordinate information in the support location data, the straight-line distance between adjacent support structures is calculated using spatial distance calculation formulas, such as the distance formula between two points in a Cartesian coordinate system, resulting in multiple support pole distance data. For example, the distance between tower A-1 and tower A-2 is 500m, thus forming one support pole distance data.

[0082] The pre-set wire database stores sag parameter models for different wires under specific conditions (such as span, tension, and ambient temperature). Based on the wire type and corresponding support pole distance in the support wire data, the theoretical sag value of the wire at the current span is matched from the database and used as the wire sag data. For example, the sag value of the LGJ-240 overhead line at a 500m span is 2.8m.

[0083] The data is de-unitized using a proportional calculation method: the ratio of the actual distance between support poles to the reference span (e.g., 1000m) is used as the pole-to-pole coefficient, and the ratio of the actual sag to the reference sag (e.g., 5m) is used as the sag coefficient. Then, the two coefficients are weighted using preset weights (e.g., 0.4 for the pole-to-pole coefficient and 0.6 for the sag coefficient) to obtain the sag loss data. This data quantitatively reflects the impact of sag changes on the attenuation of traveling waves; a larger value indicates a more significant impact.

[0084] The support wire data distinguishes between Category I and Category II wires. Category I wires are overhead wires (such as steel-cored aluminum stranded wire), while Category II wires are cables (such as cross-linked polyethylene insulated cables) or other types of transmission lines (such as aluminum alloy conductors). The differences in physical properties between the two types of wires result in different traveling wave propagation characteristics, and their intersection areas are prone to traveling wave reflection and distortion.

[0085] Based on the above correlation, the first type of support (such as overhead line towers) corresponding to the first type of wire material and the second type of support (such as cable terminal poles) corresponding to the second type of wire material are extracted from the support location data. For example, the overhead line LGJ-240 corresponds to towers A-1 to A-5, and the cable YJV-10 corresponds to terminal pole B-1. That is, towers A series are the first type of support, and terminal pole B-1 is the second type of support.

[0086] Spatial orientation analysis identifies the spatial intersection area between the first and second types of supports. When the coordinate positions of the two types of supports meet a preset distance threshold (e.g., within 5m), and the corresponding cable laying paths have intersecting or parallel sections, it is determined to be a cable crossing node. For example, if tower A-5 and terminal pole B-1 are 3m apart, and the overhead line and cable cross at this point, a cable crossing node is formed.

[0087] The method of de-normalization is adopted by using a proportional calculation: taking the wire crossing angle (e.g., 90° as the reference) and parallel length (e.g., 10m as the reference) at the crossing node as references, the ratio of the actual crossing angle to the reference angle and the ratio of the actual parallel length to the reference length are calculated, and the maximum value of the two is taken as the crossing loss data. This data quantitatively reflects the distortion effect of wire crossing on traveling waves; the larger the value, the stronger the interference.

[0088] The weighted average method is used to combine sag loss data and crossing loss data: based on the actual operating conditions of the line, the weights are preset (e.g., sag loss weight 0.3, crossing loss weight 0.7), and the two types of data are accumulated according to their weights to obtain traveling wave waveform loss data; this data comprehensively reflects the overall impact of the line's physical characteristics on the traveling wave signal.

[0089] The preset waveform loss reference data is a critical value (e.g., 0.6) calibrated using historical fault data. If the calculated traveling wave waveform loss data is greater than this critical value, the system determines that the current line conditions may cause the fault location error to exceed the allowable range, and immediately issues a waveform identification warning, prompting maintenance personnel to focus on checking this area.

[0090] By quantifying the impact of sag and crossing on traveling wave losses, the physical characteristics of the line are transformed into calculable loss parameters, enabling early warning for more accurate fault detection. Compared to traditional methods, this method eliminates the dimensional differences of different line parameters through de-normalization and dynamic weighting, making loss analysis more universal. Simultaneously, it clearly distinguishes between overhead lines and cables, specifically targeting key interference points such as crossing nodes, significantly improving the accuracy of early warning.

[0091] Reference Figure 2 The step of calculating sag loss data based on the distance data between support rods and the wire sag data also includes the following sub-steps:

[0092] The distance data between each support pole is grouped one-to-one with the wire sag data; each group of data corresponds to the physical characteristics of the same section of the line; that is, "the distance between two adjacent support poles" and "the sag value of the wire in that section of the line" form a related group. For example:

[0093] Group 1: Support rods A1-A2 spacing 500m + corresponding wire sag 2.8m;

[0094] Group 2: Support rods A2-A3 spacing 450m + corresponding wire sag 2.5m;

[0095] Group n: Support rod An-An+1 spacing 520m + corresponding wire sag 3.0m;

[0096] By grouping, it is ensured that subsequent calculations can accurately correlate the pole spacing and sag characteristics of the same line segment, avoiding cross-interference of parameters from different line segments.

[0097] For each data set, the distance between support poles is calculated and compared to the preset reference distance (after de-normalization) to obtain the pole loss ratio. The preset reference distance is a baseline value determined according to the distribution network design specifications, such as 1000m. This value can be dynamically adjusted according to the line voltage level. The pole loss ratio = actual support pole distance ÷ reference distance (the result is rounded to two decimal places). For example, in group 1, 500m ÷ 1000m = 0.50, meaning the pole loss ratio is 0.50. This ratio reflects the degree of deviation between the actual span and the reference span; the larger the ratio, the closer the line span is to or exceeds the reference value, and the higher the risk of attenuation of traveling waves due to increased propagation distance.

[0098] For each set of data, the ratio of the wire sag data to the preset reference sag data is calculated (after de-normalization) to obtain the sag loss ratio. Wherein: the preset reference sag data is the standard sag value of this type of wire at the reference pole distance (e.g., 5.0m, calibrated in the technical manual provided by the wire manufacturer); sag loss ratio = actual wire sag value ÷ reference sag data (the result is rounded to two decimal places). For example: in group 1, 2.8m ÷ 5.0m = 0.56, that is, the sag loss ratio is 0.56. This ratio reflects the degree of deviation between the actual sag and the standard sag; the larger the ratio, the closer the sag is to the critical value, and the higher the risk of propagation path distortion of the traveling wave due to conductor sag.

[0099] A weighted average algorithm is used to integrate the inter-pole loss ratio and sag loss ratio within the same group to obtain temporary sag loss data for that group. The weighting weights are preset based on line characteristics and can be optimized using historical fault data, for example:

[0100] The weight of the inter-pole loss ratio is 0.4, and the weight of the impact of the span on the traveling wave attenuation is 0.4.

[0101] Sag loss ratio weight = 0.6, which is the weight of the effect of sag on traveling wave distortion;

[0102] Temporary sag loss data = inter-bar loss ratio × 0.4 + sag loss ratio × 0.6;

[0103] Taking Group 1 as an example: 0.50×0.4+0.56×0.6=0.20+0.34=0.54, meaning the temporary data for this group is 0.54. This step, through weighted fusion, quantifies the "span-sag" coupling effect of the same line segment, avoiding the limitations of single-parameter analysis.

[0104] Iterate through all data groups, repeating the above sub-steps to obtain the temporary sag loss data for each group, such as group 1 = 0.54, group 2 = 0.48, group 3 = 0.59, and so on. Calculate the arithmetic mean of all temporary data to obtain the sag loss data covering the entire line between the two distribution network traveling wave identification devices: Sag loss data = (temporary data from group 1 + temporary data from group 2 + ... + temporary data from group n) ÷ n; For example, if the three temporary data groups are 0.54, 0.48, and 0.59 respectively, their average value = (0.54 + 0.48 + 0.59) ÷ 3 ≈ 0.54, meaning the sag loss data for this line is 0.54.

[0105] Integrating multiple sets of data can effectively eliminate the impact of local line parameter anomalies (such as deviations caused by measurement errors in a single set of data) on the overall results, ensuring that the sag loss data can truly reflect the comprehensive impact of the sag characteristics of the entire line on the traveling wave.

[0106] Reference Figure 3The step of calculating the travel wave waveform loss data based on sag loss data and crossing loss data also includes the following sub-steps:

[0107] Based on support location data, including spatial information such as support rod coordinates and cable laying paths, the sag positions corresponding to sag loss data and crossing positions corresponding to crossing loss data are identified respectively. Crossing positions are the specific coordinates of the cable crossing nodes, such as latitude and longitude: N30°12′, E120°08′, corresponding to the physical locations where two types of cables intersect or run parallel. The sag position is the location of the sag point closest to the crossing position with a sag angle greater than a preset sag reference angle. Specifically, the coordinates of the sag point "closest to the crossing position with a sag angle greater than the preset sag reference angle" are selected from all sag points on the line. The sag angle is the angle between the lowest point of the cable and the lines connecting the two end support points. The preset sag reference angle can be set to 5°; sag with too small an angle has negligible impact on traveling waves. For example, if the crossing location is P0 (N30°12′, E120°08′), and the sag angles of its three surrounding sag points are 3°, 7°, and 6° respectively, then the sag point P1 (N30°12′02″, E120°08′01″) with the closest sag angle of 7° is selected as the sag location. By using angle selection and distance priority principles, the sag point with the most significant impact on the propagation of traveling waves in the crossing area is accurately identified, avoiding interference from irrelevant sag points.

[0108] Based on the coordinates of the sag position P1 and the crossing position P0, the straight-line distance between them is calculated using a spatial distance formula (such as the two-point distance algorithm in the WGS84 coordinate system), yielding the positional distance value (unit: meters). For example, after converting the latitude and longitude difference between P1 and P0 into a planar distance, the calculated positional distance value is 8 meters. This value reflects the degree of spatial connection between the sag and the crossing.

[0109] Based on the crossing loss data, the ratio of the actual crossing loss to the preset crossing reference data is calculated (after de-normalization), resulting in the crossing loss ratio. Wherein: the preset crossing reference data is the loss value under baseline crossing conditions, such as the standard loss when two types of wires cross perpendicularly, which can be set to 1.0; the crossing loss ratio = actual crossing loss data ÷ crossing reference data (the result is rounded to two decimal places). For example: if the crossing loss data for a certain crossing node is 0.6, then the crossing loss ratio = 0.6 ÷ 1.0 = 0.60, reflecting the proportional relationship between the actual distortion of the traveling wave at that crossing point and the baseline value. In this paper, the crossing point refers to the wire crossing node.

[0110] Based on the sag loss ratio and the crossing loss ratio, the traveling wave waveform loss data is calculated using a weighted average. The weights of the two are dynamically adjusted according to the location distance value, as follows:

[0111] Preset reference distance value: set according to line operation and maintenance experience, such as 20 meters. The impact of sag on traveling waves in the crossing area can be ignored if the distance exceeds this.

[0112] If the location distance value is greater than the reference distance value, such as 25 meters > 20 meters: the weight of the sag loss ratio = 0, the weight of the crossing loss ratio = 1, that is, the traveling wave waveform loss data = the crossing loss ratio;

[0113] If the location distance value is less than or equal to the reference distance value, such as 8 meters less than or equal to 20 meters: the weight of the crossing loss ratio is positively correlated with the location distance value, such as weight = location distance value ÷ reference distance value, and the weight of the sag loss ratio = 1 - weight of the crossing loss ratio.

[0114] For example: Location distance = 8 meters, reference distance = 20 meters, sag loss ratio = 0.54, crossing loss ratio = 0.60, then: crossing loss ratio weight = 8 ÷ 20 = 0.4, sag loss ratio weight = 1 - 0.4 = 0.6; traveling wave waveform loss data = 0.54 × 0.6 + 0.60 × 0.4 = 0.324 + 0.24 = 0.564.

[0115] To adapt to different line environments, the weight adjustment has been further refined:

[0116] For mountainous areas (where sag has a significant impact): the reference distance value can be reduced, such as to 15 meters, to increase the weighting of close-range sag;

[0117] For densely populated urban cable areas (where crossings have a significant impact): the reference distance value can be increased, such as 30 meters, to expand the dominant range of crossing losses.

[0118] By using dynamic weights, the loss calculation can reflect both the "coupling effect of sag and crossing at close range," such as when the two are 5 meters apart, the conductor deformation caused by sag will aggravate the traveling wave reflection at the crossing point, and the "ineffective interference of sag at long distances," such as when the two are 50 meters apart, only the influence of crossing itself needs to be considered.

[0119] This method retains only the sag points that significantly affect the crossing, reducing redundant data interference; it dynamically allocates weights according to distance, making the loss data more consistent with the actual propagation path of the traveling wave, with both factors influencing the loss at close range and a single factor dominating at long range; the reference distance value can be adjusted as needed to adapt to different power distribution network scenarios such as mountainous areas and cities, ultimately improving the accuracy of traveling wave waveform loss data.

[0120] The method also includes the following steps:

[0121] For the sag position determined in the above steps, i.e., the target sag point, a preset first distance range is defined centered on this position, such as 50 meters. This range can be adjusted according to the line voltage level and terrain complexity: 80 meters for mountainous lines and 30 meters for plain lines. Within this range, support position data and sag-related parameters are scanned to identify all sag points, including the target sag point itself, and the following information is collected:

[0122] The total number of sag points within the first distance range is denoted as N. For example, a total of 6 sag points were identified within a 50-meter range.

[0123] The distances between each sag point and the target sag point are denoted as d1, d2, ..., d. n ;

[0124] The sag angle of each sag point must be greater than the preset sag reference angle, i.e., the angle condition, and invalid sag points are eliminated.

[0125] Based on the above information, the first cluster data is calculated using the following formula: First cluster data = Total number of sag points N ÷ Area of ​​the first distance range; Note: The area of ​​the first distance range is calculated based on the area of ​​a circle, i.e., π × first distance², and the result is rounded to two decimal places.

[0126] For example, if the first distance range is 50 meters (area ≈ 7850 square meters), and 6 valid sag points are identified within the range, then the first cluster data = 6 ÷ 7850 ≈ 0.00076, which quantitatively reflects the density of the distribution of sag points in the area.

[0127] The preset first reference data is the "standard cluster density" calibrated through historical operation and maintenance data, such as 0.0005 / square meter, corresponding to the sag distribution density commonly found on plain lines. Based on the first cluster data and the first reference data, the first intermediate value is calculated: First intermediate value = First cluster data ÷ First reference data; this value intuitively reflects the deviation between the actual clustering degree and the standard value: if the first intermediate value > 1, it indicates that the sag point distribution is denser than the standard state; if < 1, it indicates that the distribution is sparser.

[0128] The weights of the sag loss ratio (denoted as W2) are positively adjusted based on the first median value. The adjustment logic must be combined with the base weights.

[0129] Basic weight: The initial weight of the sag loss ratio determined based on the location distance value, such as 0.6;

[0130] Adjusted weight = base weight × first intermediate value; the first intermediate value shall not exceed 1.0 to avoid weight overflow.

[0131] For example: if the base weight is 0.6, the first median is 1.5, and the clustering degree is higher than the standard, then the adjusted weight = 0.6 × 1.5 = 0.9; if the first median is 0.8, and the clustering degree is lower than the standard, then the adjusted weight = 0.6 × 0.8 = 0.48.

[0132] In distribution networks with complex terrain, sag points often exhibit regional clustering characteristics (e.g., densely distributed sag points in valley sections due to frequent span changes). The overall impact of sag on traveling wave signals in such areas is far greater than that of a single sag point. This step analyzes the spatial clustering of sag points and dynamically adjusts the weighting of the sag loss ratio to make the loss calculation more closely reflect the actual line characteristics.

[0133] In complex power distribution network layouts, such as areas where underground and overhead cables alternate, cable crossing points often exhibit a dense distribution, with three or more crossing points within a 50-meter radius. In such areas, traveling wave signals experience superimposed distortion due to multiple crossings, with cumulative effects such as reflection and refraction, resulting in losses far exceeding those of a single crossing point. This step analyzes the spatial clustering of crossing points and dynamically adjusts the weighting of the crossing loss ratio to make the loss assessment more closely reflect actual propagation characteristics. The specific steps are as follows:

[0134] For the cable crossing node, i.e., the target crossing point, a preset second distance range is defined centered on the node location, such as 80 meters. This range can be adjusted according to the intensity of crossing interference. For parallel sections of cables and overhead lines, the range can be set to 100 meters, and for perpendicular crossing sections, it can be set to 50 meters. Within this range, support location data and cable crossing information are scanned to identify all valid crossing points. Valid crossing points must meet the requirement that "the physical crossing or parallel length of the two types of cables is ≥3 meters," excluding invalid crossings such as temporary wiring. The following information is then collected:

[0135] The total number of valid crossing points within the second distance range is denoted as M. For example, 4 crossing points were identified within a range of 80 meters.

[0136] The distances between each crossing point and the target crossing point are denoted as s1, s2, ... s m ;

[0137] For each crossing point, the combination of wire types, such as "overhead line-cable" or "cable-cable", only crossing points with different media are retained, and connection points with the same type of wire are excluded.

[0138] Based on the above information, the second cluster data is calculated using the following formula: Second cluster data = Total number of valid crossing points M ÷ Area of ​​the second distance range; where the area of ​​the second distance range is calculated as the area of ​​a circle, i.e., π × second distance², and the result is rounded to three decimal places.

[0139] For example, if the second distance range is 80 meters (area ≈ 20096 square meters), and 4 "overhead line-cable" crossing points are identified within the range, then the second aggregated data = 4 ÷ 20096 ≈ 0.00020, which quantitatively reflects the density of the distribution of crossing points in the area.

[0140] The preset second reference data is the "standard crossing cluster density" calibrated through typical line tests, such as 0.00010 / square meter, corresponding to the average crossing distribution density of the urban power distribution network. Based on the second cluster data and the second reference data, a second intermediate value is calculated: Second intermediate value = Second cluster data ÷ Second reference data; this value directly reflects the deviation between the actual crossing cluster density and the standard value: if the second intermediate value > 1 (e.g., 0.00020 ÷ 0.00010 = 2.0), it indicates that the crossing point distribution is denser than the standard state; if < 1 (e.g., 0.00008 ÷ 0.00010 = 0.8), it indicates that the distribution is sparser.

[0141] The weights of the crossing loss ratio are positively adjusted based on the second median value. The specific adjustment logic needs to be combined with the base weights related to location distance.

[0142] Base weight: the initial weight of the crossing loss ratio determined based on the location distance value;

[0143] The adjusted weight is equal to the base weight multiplied by the second intermediate value. The second intermediate value shall not exceed 1.0 to avoid weight overflow.

[0144] For example: if the base weight is 0.4, the second median is 2.0, and the clustering degree is higher than the standard, then the adjusted weight = 0.4 × 2.0 = 0.8; if the second median is 0.8, and the clustering degree is lower than the standard, then the adjusted weight = 0.4 × 0.8 = 0.32.

[0145] For special scenarios, such as when there are 5 or more crossing points within an 80-meter range, an upper limit adjustment coefficient for the weight can be added, such as 1.2, to ensure the adjustment range of the weight in extremely dense areas.

[0146] After adjustment in extreme scenarios, the weight is min(base weight × second intermediate value, 1.0) × 1.2. If the result is ≤ 1.2, the sag loss weight ratio needs to be adjusted simultaneously to ensure that the total weight sum is 1.0.

[0147] The second aggregated data is defined as the "number of effective crossing points per unit area," transforming the vague description of "dense crossings" into a calculable numerical indicator and avoiding human judgment bias. By positively adjusting the weights through the second median value, the weight of the crossing loss ratio in densely crossed areas (second median value > 1) is significantly increased, strengthening its contribution to the traveling wave waveform loss data. The weights in sparse areas (second median value < 1) are reduced, weakening secondary influences and making the loss calculation more consistent with the actual characteristics of "dense crossing superimposed distortion." The second distance range can be adjusted as needed, such as expanding the range for parallel crossing sections and shrinking the range for vertical crossing sections, ensuring that key influencing areas can be accurately captured in different crossing patterns, such as areas where urban underground cables and overhead lines are laid alternately, and single crossing points in mountainous areas.

[0148] Distribution network branches (such as T-junction branches and radial branches) are significant sources of interference for traveling wave signal propagation. Branching points cause shunting and reflection of the traveling wave, and the combined effect of changes in wire sag exacerbates waveform distortion. This step identifies the distribution density of distribution network branches and dynamically adjusts the weighting of the crossing loss ratio based on the number of branches, making the loss assessment more closely reflect the combined interference scenario of "branch + crossing." The specific steps are as follows:

[0149] Distribution network branches typically exhibit a combination of abrupt changes in support pole spacing and abnormal sag. Additional support structures are required at branch points, leading to a shortening of the spacing between adjacent poles, and the sag of the branch wire differs from that of the main line. Based on support pole distance data and wire sag data, distribution network branches are identified using the following rules:

[0150] Pole spacing change judgment: If the distance between a certain support pole (e.g., 20 meters) is much smaller than the average pole spacing of adjacent sections (e.g., 500 meters), and there is a record of wire branching out at that location, it is extracted from the support wire data and judged as a potential branch point;

[0151] Sag difference verification: Extract the sag data of the main line and the branch line at this location. If the difference between the sag value of the branch line (e.g., 1.5 meters) and the sag value of the main line (e.g., 3.0 meters) exceeds 30% of the sag of the main line (i.e., |1.5-3.0|÷3.0=50%>30%), then the location is confirmed as a distribution network branch, excluding normal sag changes caused by terrain.

[0152] For example: if the distance between support poles A5 and A6 is 25 meters (the average pole spacing between adjacent sections is 480 meters), the sag of the branch line is 1.2 meters, and the sag of the main line is 3.2 meters (difference ratio = 62.5% > 30%), then it is determined that there is a distribution network branch between A5 and A6.

[0153] Using the midpoint of the line between two paired distribution network traveling wave identification devices as the center, a preset local range of the distribution network is defined; such as 1 kilometer, which can be adjusted according to the branch density of the line: 0.5 kilometers for urban distribution networks and 2 kilometers for rural distribution networks. Within this range, all identified distribution network branches are scanned, and the total number of branches is counted; the total number of branches is denoted as K, such as 3 branches identified within a 1-kilometer range.

[0154] The delineation of the local scope should cover the areas where branches are most likely to be densely distributed in the line, such as user concentration areas, to ensure that the statistical results can reflect the branch distribution characteristics of key areas.

[0155] The preset reference branch number is the "reasonable branch density threshold" determined by the line design specifications; for example, a maximum of 2 branches per kilometer, exceeding which is considered excessive branch density. Based on the comparison between the number of branches K and the reference branch number, the following operations are performed:

[0156] If the number of branches K > the reference number of branches, such as 3 > 2: it is determined that the branches in this local area are too dense, and the traveling wave signal is prone to failure of feature extraction due to multiple shunting. The distribution network branch warning prompt is immediately triggered, and the prompt content includes the branch location, number and suggested inspection direction;

[0157] If the number of branches K ≤ the reference number of branches, such as 2 ≤ 2: the branch density is within a reasonable range, no warning is needed, and the subsequent weight adjustment steps continue.

[0158] When the number of branches does not exceed the reference number of branches, the branch ratio is calculated: Branch ratio = Number of branches K ÷ Reference number of branches; this ratio reflects the ratio of actual branch density to reasonable density; for example, when K=2 and the reference number of branches=2, the branch ratio=1.0; when K=1, the branch ratio=0.5. The weight (denoted as W1) of the crossing loss ratio is positively adjusted based on the branch ratio:

[0159] Adjusted crossing loss ratio weight = Basic crossing loss weight × (1 + branch ratio);

[0160] Among them, the basic weight of the crossing loss is the initial weight without considering the branch effect, and the adjusted weight does not exceed 1.0;

[0161] For example: if the basic weight of the crossing loss is 0.4 and the branch ratio is 1.0 (the branch density reaches the reasonable upper limit), then the adjusted weight is 0.4 × (1 + 1.0) = 0.8; if the branch ratio is 0.5 (the branch density is low), then the adjusted weight is 0.4 × (1 + 0.5) = 0.6.

[0162] Adjustment logic: The denser the branches (the larger the branch ratio), the more easily the distortion of the traveling wave at the crossing node is amplified by the branch interference. Therefore, it is necessary to increase the weight of the crossing loss ratio and strengthen its influence on the overall loss.

[0163] Branch identification is based on quantitative rules of "abrupt changes in inter-pole distance + sag difference," avoiding reliance on the subjectivity of manual annotation and ensuring the accuracy of branch identification. When branches are too dense, an early warning is issued directly. When the density is reasonable, the weight is adjusted by the branch ratio, realizing a hierarchical processing of "early warning for abnormalities and optimization for normalities," adapting to different branch scenarios. By positively correlated adjustment, the branch density is linked to the crossing loss weight, so that the assessment of crossing loss can reflect the indirect impact of branches. For example, the more branches there are, the higher the crossing loss weight, ultimately improving the accuracy of traveling wave loss calculation under complex branch layouts.

[0164] In power distribution networks, the overlapping areas of branch lines and cable crossings (such as when both overhead lines and cables cross at a branch point) are high-risk areas for traveling wave signal distortion. Relying solely on existing traveling wave identification devices may result in insufficient detection accuracy due to signal distortion. This step quantifies the comprehensive impact of this area and adds identification devices in stages to improve the monitoring capabilities of complex areas. The specific steps are as follows:

[0165] Based on the "weight of the crossing loss ratio" (denoted as W1) and the "weight of the sag loss ratio" (denoted as W2) determined in the preceding steps, calculate the weight ratio between the two:

[0166] Weight ratio = Weight W1 of crossing loss ratio ÷ Weight W2 of sag loss ratio;

[0167] The weight ratio reflects the relative strength of the influence of crossing and sag on the current traveling wave of the line: if the weight ratio is >1 (e.g., W1=0.7, W2=0.3, ratio≈2.33), it indicates that the influence of crossing is dominant; if it is <1 (e.g., W1=0.3, W2=0.6, ratio=0.5), the influence of sag is more significant.

[0168] Combining the weight ratio and the "branch ratio," denoted as R, which is the ratio of the number of branches to the number of reference branches (e.g., 0.8), the impact value of crossing branches is calculated through weighted fusion. The calculation formula is:

[0169] Impact value of crossing a branch = (weight ratio × 0.6) + (branch ratio × 0.4);

[0170] Note: Weights of 0.6 and 0.4 can be adjusted according to line characteristics. In areas with dense branches, the branch ratio weight can be increased to 0.5.

[0171] For example: if the weight ratio is 2.33 (crossing is dominant) and the branch ratio is 0.8 (the number of branches is close to the reference value), then the impact value of the crossing branch is (2.33 × 0.6) + (0.8 × 0.4) = 1.40 + 0.32 = 1.72. This value quantifies the combined interference intensity of "branch lines + cable crossing". The larger the value, the higher the complexity of the area, and the more monitoring is needed.

[0172] Two reference impact values ​​are preset and calibrated using historical fault data:

[0173] First reference impact value: such as 1.2, which corresponds to the critical value of "significant branching and crossing interference";

[0174] Second reference influence value: such as 2.0, which corresponds to the critical value of "extremely strong branching and crossing interference", and the first reference influence value < the second reference influence value.

[0175] Based on the comparison between the impact value of the traversing branch and the reference value, perform hierarchical layout optimization:

[0176] If the influence value of crossing the branch is greater than the first reference influence value and less than or equal to the second reference influence value (e.g., 1.72 > 1.2 and less than or equal to 2.0), the distribution network branch is determined to be the main interference point. A distribution network traveling wave identification device should be added at the branch node (e.g., T-junction). The new device needs to collect signals synchronously with the existing device, and the traveling wave reflection interference caused by the branch should be reduced by comparing data from multiple devices.

[0177] If the influence value of the crossing branch is greater than the second reference influence value (e.g., 2.5 > 2.0), the cable crossing node is determined to be the core interference source. The branch and crossing have a superimposed influence. A power distribution network traveling wave identification device should be installed next to the crossing node (distance ≤ 5 meters). The new device needs to focus on the traveling wave waveform changes at the crossing point and capture waveform distortion details through high-frequency sampling.

[0178] If the influence value of crossing the branch is less than or equal to the first reference influence value, such as 1.0 ≤ 1.2, the interference intensity is within an acceptable range, no additional device is needed, and the original layout can be maintained.

[0179] The weight ratio reflects the relative impact of crossing and sag, while the branch ratio reflects the branch density. The combined value of the crossing branch impact can objectively assess the complexity of the area and avoid the subjectivity of layout based on experience. Interference levels are divided by two reference values. Low-level interference only strengthens branch monitoring, while high-level interference focuses on crossing nodes, enabling on-demand addition and improving the monitoring density of key areas while controlling costs. For areas with dense branches and frequent crossings, such as industrial park sections of urban power distribution networks, the weight of the branch ratio can be increased (e.g., 0.5) to make the crossing branch impact value more likely to trigger the addition conditions, ensuring the quality of traveling wave signal acquisition in complex scenarios.

[0180] The reference distance value is the core parameter for dynamic weight adjustment, and its rationality directly affects the accuracy of traveling wave waveform loss calculation. Different distribution network lines (such as old and new lines, mountain lines and plain lines) have significant differences in conductor aging and laying environment, making it difficult to adapt a fixed reference distance value to all scenarios. This step dynamically optimizes the reference distance value by applying a simulated ground fault signal and combining it with the actual loss difference. The specific steps are as follows:

[0181] On the line between two paired distribution network traveling wave identification devices, select two locations that meet the following conditions:

[0182] First position P1: Select a straight section of the line without obvious abnormal sag, such as the midpoint between towers A3 and A4, where the conductor sag angle is ≤3°;

[0183] Second position P2: Maintain a preset distance from the first position, such as 500 meters. The distance can be adjusted according to the line length, and the maximum distance shall not exceed 1 / 3 of the monitoring range of the device. It must also meet the requirements of "no cable crossing between the first position and the second position", "no branch line" and "no other line crossing", that is, eliminate interference factors such as crossing and branching, and only retain the influence of sag on traveling waves.

[0184] Using a distribution network simulation device or a portable signal generator, apply standardized simulated ground fault signals at the first position P1 and the second position P2 respectively, ensuring that the parameters of the two signals are exactly the same; the amplitude and frequency of the simulated ground fault signal are consistent with the actual ground fault traveling wave, such as an amplitude of 10kV and a main frequency of 5kHz.

[0185] After the analog signal is applied, the traveling wave propagation data is collected by the distribution network traveling wave identification device, and the corresponding traveling wave waveform loss data is calculated according to the steps described above in this method:

[0186] Based on the analog signal at the first position P1, the first traveling wave waveform loss data L1 is calculated, such as 0.35.

[0187] Based on the analog signal at the second position P2, the loss data L2 of the second traveling wave waveform is calculated, such as 0.42.

[0188] Since there are no cables crossing at the first and second locations, the difference in loss data is only caused by the inherent characteristics of the line, such as sag distribution and conductor material uniformity, and can truly reflect the sag loss variation pattern when there is no crossing interference.

[0189] Calculate the difference between the two traveling wave waveform loss data, i.e.: First simulated difference = |first traveling wave waveform loss data L1 - second traveling wave waveform loss data L2|;

[0190] For example, if L1=0.35 and L2=0.42, then the first simulation difference = |0.35-0.42|=0.07. This difference reflects the fluctuation range of loss data caused by the inherent characteristics of the line when there is no crossing interference. The larger the difference, the more uneven the sag distribution of the line itself, and the more easily the influence of sag on the traveling wave changes with location. It is necessary to increase the reference distance value to cover more potentially related sag points. The smaller the difference, the more stable the line characteristics, and the smaller the reference distance value can be to avoid interference from irrelevant sag points.

[0191] The reference distance value is positively adjusted based on the first simulated difference, and the adjustment formula is as follows:

[0192] Adjusted reference distance value = initial reference distance value × (1 + first simulation difference);

[0193] The initial reference distance is a preset baseline value (e.g., 20 meters), and the adjustment coefficient (1 + first simulation difference) ensures that the reference distance increases as the difference increases. For example:

[0194] If the first simulation difference is 0.07 and the initial reference distance is 20 meters, then the adjusted reference distance is 20 × (1 + 0.07) = 21.4 meters (rounded to one decimal place).

[0195] If the first simulation difference = 0.03 (line characteristics are stable), then the adjusted reference distance value = 20 × (1 + 0.03) = 20.6 meters;

[0196] If the first simulation difference is 0.15 (uneven distribution of line sag), then the adjusted reference distance value is 20 × (1 + 0.15) = 23.0 meters.

[0197] The adjusted reference distance value should be limited to a preset range (e.g., 10 meters to 50 meters) to avoid the reference distance value being too large or too small due to extreme differences; the minimum should not be less than 10 meters to prevent the influence of near-distance sag from being missed; the maximum should not exceed 50 meters to avoid including irrelevant sag points.

[0198] Simulated signals are applied to line segments without cable crossings or branches to ensure that the first simulated difference only reflects loss fluctuations related to sag, avoiding interference from other factors with the adjustment logic. By using the rule that a larger loss difference corresponds to a larger reference distance, the reference distance adaptively matches the line's sag distribution characteristics. For lines with uneven sag, the reference range is expanded to ensure no associated sag points are missed. For lines with stable characteristics, the range is narrowed to reduce redundant data. For older lines, sag is prone to local anomalies due to aging, and the simulated difference is usually large; the adjusted reference distance can cover more potential affected points. For newly built lines, with uniform sag, the reference distance remains at a smaller level, improving computational efficiency.

[0199] The positive correlation coefficient between the location distance value and the crossing loss ratio (hereinafter referred to as the "positive correlation coefficient") is a key parameter for adjusting the crossing loss weight. This coefficient needs to match the actual crossing loss characteristics of the line; different cable crossing scenarios (such as overhead lines crossing perpendicularly and parallel to cables) have significantly different effects on traveling waves, making it difficult to adapt a fixed coefficient. This step dynamically optimizes this coefficient by applying a simulated fault signal to the line section containing cable crossings, combined with the loss difference. The specific steps are as follows:

[0200] On the line between two paired distribution network traveling wave identification devices, select two locations that meet the following conditions:

[0201] The third location, P3, is located on one side of the cable crossing node, such as 30 meters upstream of the crossing node, and the line type is Class I wire, such as an overhead line.

[0202] The fourth position, P4, is located on the other side of the cable crossing node, such as 30 meters downstream of the crossing node, and the line type is Class II wire, such as cable.

[0203] There is only one cable crossing node between the third and fourth positions, with no other crossings or branches, and both positions are equidistant from the crossing node, ensuring that the crossing node is the only major source of interference in the analog signal propagation path.

[0204] Using the same analog signal generating device as in the above sub-steps, apply analog ground fault signals with the same parameters (e.g., amplitude 10kV, main frequency 5kHz) to the third position P3 and the fourth position P4 respectively, ensuring that the initial characteristics of the two signals are consistent.

[0205] After the analog signal is applied, the propagation data of the traveling wave after passing through the cable crossing node is collected by the distribution network traveling wave identification device, and the corresponding traveling wave waveform loss data is calculated according to the steps described above in this method:

[0206] Based on the analog signal at the third position P3, the traveling wave propagates from the overhead line to the cable and passes through the crossing node, and the third traveling wave waveform loss data L3 is calculated, such as 0.58.

[0207] Based on the analog signal at the fourth position P4, the traveling wave propagates from the cable to the overhead line and passes through the crossing node, and the fourth traveling wave waveform loss data L4 is calculated, such as 0.65.

[0208] Among them, since there is only one cable crossing node between the third and fourth positions, the difference in the two loss data is mainly caused by the reflection and refraction effects of the crossing node, which can truly reflect the "actual impact of cable crossing on traveling wave loss".

[0209] Calculate the difference between the two traveling wave waveform loss data, i.e.: Second simulated difference = |Third traveling wave waveform loss data L3 - Fourth traveling wave waveform loss data L4|;

[0210] For example, if L3 = 0.58 and L4 = 0.65, then the second simulated difference = |0.58 - 0.65| = 0.07. This difference directly reflects the actual impact of cable crossing on traveling wave loss. The larger the difference, the more significant the traveling wave distortion caused by the crossing node, requiring an increase in the positive correlation coefficient to enhance the sensitivity of adjusting the weight of crossing loss based on location distance. Conversely, the smaller the difference, the weaker the impact of crossing, allowing for a reduction in the coefficient to avoid excessive weight amplification.

[0211] The positive correlation coefficient is adjusted based on the second simulated difference, using the following formula:

[0212] Adjusted positive correlation coefficient = initial positive correlation coefficient × (1 + second simulation difference);

[0213] The initial positive correlation coefficient is a preset baseline value, such as 0.05 / meter. This means that for every 1 meter increase in location distance, the weight of the crossing loss ratio increases by 0.05. The adjustment coefficient (1 + second simulation difference) ensures that the coefficient increases with the magnitude of the crossing impact. For example:

[0214] If the second simulation difference is 0.07 and the initial positive correlation coefficient is 0.05 / meter, then the adjusted positive correlation coefficient is 0.05 × (1 + 0.07) = 0.0535 / meter, rounded to four decimal places.

[0215] If the second simulation difference is 0.12, the impact of crossing is significant. For example, if the parallel crossing is 10 meters, the adjusted positive correlation coefficient is 0.05 × (1 + 0.12) = 0.056 / meter. At this time, the position distance value is more sensitive to the adjustment of the crossing weight. For example, the weight corresponding to an 8-meter distance is 8 × 0.056 = 0.448, which is 12% higher than the initial coefficient of 0.4.

[0216] If the second simulation difference is 0.03, the impact of crossing is relatively weak. If it is a vertical crossing, the adjusted positive correlation coefficient is 0.05 × (1 + 0.03) = 0.0515 / meter, to avoid excessive weighting.

[0217] The adjusted positive correlation coefficient needs to be limited to a preset range, such as 0.02 / meter to 0.1 / meter, to prevent extreme values ​​from causing abnormal weight adjustment: the lower limit of 0.02 / meter ensures that effective weight can still be generated for close-distance crossings, and the upper limit of 0.1 / meter avoids excessive weight for long-distance crossings.

[0218] Choosing the path from the third location to the crossing node to the fourth location ensures that the second simulation difference only reflects the loss impact of cable crossing, excluding interference from other factors such as branches and dense sag, making the adjustment logic more accurate. By using the rule that the greater the crossing impact, the higher the coefficient, the positive correlation coefficient adaptively matches the crossing type. Parallel crossings (strong impact) correspond to high coefficients, and vertical crossings (weak impact) correspond to low coefficients, ensuring that the adjustment range of the crossing weight based on location distance closely matches reality. For the "multiple cable parallel crossings" scenario commonly seen in urban power distribution networks, the simulation difference is usually large, and the adjusted coefficient can enhance the weight sensitivity and avoid underestimating the crossing loss. For the "single vertical crossing" scenario less common in rural power distribution networks, the coefficient is kept at a low level to prevent weight redundancy.

[0219] Among these issues, the sag of the line increases the positioning error, with an amplitude exceeding 5%. In distribution network traveling wave detection, cable crossing refers to the phenomenon where a fault traveling wave signal propagates in a distribution network containing both cables and overhead lines. Due to the significant difference in wave impedance between the two mediums (cable and overhead line), the traveling wave undergoes reflection, refraction, and waveform distortion at the junction (crossing point). Specifically, the wave impedance of cables is typically much smaller than that of overhead lines (e.g., cable wave impedance is approximately 10-50Ω, while overhead line wave impedance is approximately 200-400Ω). When a traveling wave enters from one medium to the other, a significant change in wave impedance occurs at the crossing point. This change causes some of the traveling wave energy to be reflected back to the original line, while some energy is refracted into the other line, resulting in waveform distortion (e.g., a slower wavefront, attenuated amplitude, and increased oscillation), and altering the propagation speed and direction of the traveling wave. This phenomenon interferes with the accurate extraction of traveling wave characteristics (e.g., wavefront arrival time, polarity, and amplitude), increasing the difficulty of fault location, and is one of the key problems that needs to be addressed in distribution network traveling wave detection. There is a cable crossing on the line, and the signal is attenuated by 67% after passing through the cable once.

[0220] This application also discloses a distribution network fault early warning device based on traveling wave waveform recognition, including a processor, wherein the processor executes the steps of the distribution network fault early warning method based on traveling wave waveform recognition as described in any of the above embodiments.

[0221] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. A method for early warning of distribution network faults based on traveling wave waveform recognition, characterized in that, The application between two paired distribution network traveling wave identification devices includes the following steps: Based on the distribution network fault warning prompt issued by the distribution network traveling wave identification device, the distribution network line data of the power line between two paired distribution network traveling wave identification devices is obtained, wherein the distribution network fault warning prompt is triggered according to the traveling wave waveform identification. Extract the associated support location data and support cable data from the power distribution network line data; The distance data between multiple support rods is calculated based on the support position data; Based on the support wire data, wire sag data is matched from a preset wire database; The sag loss data is calculated based on the distance data between the support rods and the wire sag data. The first type of wire and the second type of wire are identified from the support wire data; Based on the association, a first type of support and a second type of support are extracted from the support location data. The first type of support is associated with the first type of cable, and the second type of support is associated with the second type of cable. The wire crossing nodes are matched based on the spatial orientation of the first type of support and the second type of support; The crossing loss data is calculated based on the wire crossing nodes; Calculate the travel wave waveform loss data based on the sag loss data and the crossing loss data; If the traveling wave waveform loss data is greater than the preset waveform loss reference data, a waveform identification warning will be issued.

2. The distribution network fault early warning method based on traveling wave waveform recognition according to claim 1, characterized in that, The step of calculating the sag loss data based on the distance data between the support rods and the wire sag data further includes the following sub-steps: The distance data between each of the support rods is grouped in a one-to-one correspondence with the sag data of the wire. Calculate the inter-rod loss ratio between the support rod distance data and the preset reference rod distance data; Calculate the sag loss ratio between the wire sag data and the preset reference sag data; Temporary sag loss data is calculated by weighted average of the inter-bar loss ratio and the sag loss ratio; By iterating through the distance data between each set of support rods and the sag data of the wire, multiple temporary sag loss data are obtained, and the sag loss data is calculated based on the multiple temporary sag loss data.

3. The distribution network fault early warning method based on traveling wave waveform recognition according to claim 2, characterized in that, The step of calculating the travel wave waveform loss data based on the sag loss data and the crossing loss data further includes the following sub-steps: Based on the support position data, the sag position corresponding to the sag loss data and the crossing position corresponding to the crossing loss data are identified; the sag position is the position of the sag point closest to the crossing position and whose sag angle is greater than the preset sag reference angle; Calculate the positional distance between the sag position and the crossing position; Calculate the crossing loss ratio between the crossing loss data and the preset crossing reference data; The travel wave waveform loss data is calculated by weighting the sag loss ratio and the crossing loss ratio. If the location distance value is greater than the preset reference distance value, then the weight of the sag loss ratio is 0, and the weight of the crossing loss ratio is 1. Otherwise, the weight of the crossing loss ratio is adjusted according to the positive correlation between the location distance value and the crossing loss ratio.

4. The distribution network fault early warning method based on traveling wave waveform recognition according to claim 3, characterized in that, The method also includes the following steps: Within a predetermined first distance range around the sag point, multiple sag points are identified, and first clustered data of the sag distribution is calculated based on the sag positions within the first distance range. A first intermediate value is calculated based on the first aggregated data and a preset first reference data, and the weight of the sag loss ratio is adjusted according to the positive correlation of the first intermediate value.

5. The distribution network fault early warning method based on traveling wave waveform recognition according to claim 3, characterized in that, The method also includes the following steps: Within a predetermined second distance range around the wire crossing node, multiple wire crossing nodes are identified, and second cluster data of the crossing distribution is calculated based on the crossing positions within the second distance range; A second intermediate value is calculated based on the second aggregated data and a preset second reference data, and the weight of the crossing loss ratio is adjusted according to the positive correlation of the second intermediate value.

6. The distribution network fault early warning method based on traveling wave waveform recognition according to claim 3, characterized in that, The method also includes the following steps: The distribution network branch is identified based on the distance data between the support rods and the wire sag data; Within a preset local area of ​​the distribution network, calculate the number of branches of the identified distribution network branches; If the number of branches exceeds the preset reference number of branches, a distribution network branch early warning will be issued. Otherwise, calculate the branch ratio of the number of branches to the reference number of branches, and adjust the weight of the crossing loss ratio according to the positive correlation of the branch ratio.

7. The distribution network fault early warning method based on traveling wave waveform recognition according to claim 6, characterized in that, The method also includes the following steps: Calculate the weight ratio of the crossing loss ratio to the weight of the sag loss ratio; The impact value of crossing branches is calculated based on the weight ratio and the branch ratio; If the influence value of the crossing branch is greater than the preset first reference influence value, then the distribution network traveling wave identification device is added at the distribution network branch; If the influence value of the crossing branch is greater than the preset second reference influence value, then the distribution network traveling wave identification device is added next to the wire crossing node; The first reference influence value is less than the second reference influence value.

8. The distribution network fault early warning method based on traveling wave waveform recognition according to claim 3, characterized in that, The method also includes the following steps: A simulated ground fault signal is applied at the first location on the distribution network line, and the corresponding first traveling wave waveform loss data is calculated. A simulated ground fault signal is applied at the second location on the distribution network line, and the corresponding second traveling wave waveform loss data is calculated. There is no cable crossing between the first position and the second position. Calculate the first simulated difference between the first traveling wave waveform loss data and the second traveling wave waveform loss data. The reference distance value is adjusted based on the positive correlation of the first simulated difference.

9. The distribution network fault early warning method based on traveling wave waveform recognition according to claim 3, characterized in that, The method also includes the following steps: The method also includes the following steps: A simulated ground fault signal is applied at the third location on the distribution network line, and the corresponding third traveling wave waveform loss data is calculated. A simulated ground fault signal is applied at the fourth location on the distribution network line, and the corresponding fourth traveling wave waveform loss data is calculated. A cable crosses between the third position and the fourth position, and the second simulated difference between the third and fourth traveling wave waveform loss data is calculated. The positive correlation coefficient between the location distance value and the crossing loss ratio is adjusted based on the second simulated difference positive correlation.

10. A distribution network fault early warning device based on traveling wave waveform recognition, characterized in that, The system includes a processor that performs the steps of the distribution network fault early warning method based on traveling wave waveform recognition as described in any one of claims 1-9.

Images