A method and system for locating conductor faults based on multi-source distribution network connection devices

Abstract

This invention discloses a method and system for locating conductor faults based on multi-source distribution network connection devices. The method acquires surface visual images and synchronous current values ​​of the conductor connection devices, fuses and analyzes them to filter candidate abnormal images, accurately extracts abnormal regions using a semantic segmentation model, and generates an ordered sequence of abnormal regions based on device location information. By judging the scale of consecutive abnormalities in the sequence, different location strategies are adopted: when abnormal points are concentrated, the gradient change of the abnormal region area is analyzed to accurately locate the fault source to a specific connection device or adjacent conductor segment; when abnormal points are widely distributed, it is identified as a large-scale section fault. This invention integrates visual and electrical multi-source information, achieving early detection and accurate location of conductor faults, significantly improving the reliability of distribution network fault detection and operation and maintenance efficiency.

Description

Technical Field

[0001] This invention belongs to the field of conductor fault identification technology, and in particular relates to a method and system for conductor fault location based on a multi-source distribution network connection device. Background Technology

[0002] As the final link between the power system and users, the reliability of the distribution network directly affects the quality of power supply. During long-term operation, conductors and their connecting devices (such as clamps) are prone to failure due to aging, corrosion, mechanical stress, or lightning strikes. These failures often initially manifest as localized overheating, arcing, or physical deformation. If not located and addressed promptly, they can lead to large-scale power outages.

[0003] Existing fault location methods mainly rely on electrical quantity analysis, such as measuring fault current traveling waves and impedance values ​​to determine the fault range. However, these methods have the following limitations:

[0004] Limited accuracy: It can usually only locate the fault to a large area, and it is difficult to pinpoint the specific connection device or wire point.

[0005] Not applicable to early faults: For early latent faults that have not yet caused significant changes in electrical parameters (such as slight overheating or surface cracks), the electrical quantity analysis method is not sensitive enough to effectively identify them.

[0006] Susceptible to interference: The power distribution network has a complex structure and numerous branches, making electrical signals susceptible to interference and affecting positioning accuracy.

[0007] In recent years, with the development of image recognition technology, visual inspection methods based on drones or fixed cameras have emerged. However, these methods typically only analyze images of a single conductor, lacking correlation analysis based on multi-source power distribution network connection devices. They cannot determine the propagation path and precise source of faults from a spatial sequence perspective, resulting in a high false alarm rate. Summary of the Invention

[0008] This invention provides a method and system for locating conductor faults based on multi-source distribution network connection devices, which solves the technical problem of lacking correlation analysis based on multi-source distribution network connection devices, making it impossible to determine the propagation path and precise source of faults from a spatial sequence.

[0009] In a first aspect, the present invention provides a method for locating conductor faults based on a multi-source distribution network connection device, comprising:

[0010] Acquire surface visual images of each multi-source power distribution network connection device on the conductor, and select at least one candidate surface visual image from each surface visual image according to the preset image selection rules.

[0011] Based on a preset region extraction strategy, abnormal regions are extracted from the at least one candidate surface visual image to obtain an abnormal region image corresponding to the at least one candidate surface visual image.

[0012] Based on the location information of each multi-source power distribution network connection device, the images of each abnormal area are sorted to obtain an abnormal area image sequence, and it is determined whether the number of images of the abnormal area in the abnormal area image sequence that are consecutively located is greater than a preset number threshold.

[0013] If the number is not greater than a preset threshold, at least one target abnormal region image is extracted from the abnormal region image sequence, and the abnormal region image sequence is divided according to the at least one target abnormal region image to obtain at least one abnormal region image subsequence.

[0014] Based on the at least one abnormal region image subsequence, a preset fault location strategy is used to locate the fault source on the conductor.

[0015] Secondly, the present invention provides a conductor fault location system based on a multi-source distribution network connection device, comprising:

[0016] The acquisition module is configured to acquire surface visual images of each multi-source power distribution network connection device on the conductor, and select at least one candidate surface visual image from each surface visual image according to a preset image selection rule.

[0017] The extraction module is configured to extract abnormal regions from the at least one candidate surface visual image based on a preset region extraction strategy, so as to obtain an abnormal region image corresponding to the at least one candidate surface visual image.

[0018] The judgment module is configured to sort the images of each abnormal area according to the location information of each multi-source power distribution network connection device, obtain an abnormal area image sequence, and determine whether the number of images of the abnormal area in the abnormal area image sequence that are consecutively located is greater than a preset number threshold.

[0019] The segmentation module is configured to extract at least one target abnormal region image from the abnormal region image sequence if the number is not greater than a preset threshold, and to segment the abnormal region image sequence according to the at least one target abnormal region image to obtain at least one abnormal region image subsequence.

[0020] The positioning module is configured to locate the fault source on the conductor using a preset fault positioning strategy based on the at least one abnormal region image subsequence.

[0021] Thirdly, an electronic device is provided, comprising: at least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the steps of the conductor fault location method based on a multi-source distribution network connection device according to any embodiment of the present invention.

[0022] Fourthly, the present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein when the program instructions are executed by a processor, the processor performs the steps of the conductor fault location method based on a multi-source distribution network connection device according to any embodiment of the present invention.

[0023] This application presents a method and system for locating conductor faults in multi-source distribution network connection devices. By integrating visual enhancement with thermal materials, multi-source information fusion, topology sequence analysis, and a multi-level location strategy, it achieves a significant breakthrough in distribution network conductor fault location technology. First, a thermally sensitive material coated on the surface of the connection device transforms abstract temperature rise information into intuitive color changes. This is combined with visual anomaly features and synchronous current values ​​for fusion judgment, greatly improving the sensitivity and reliability of early fault detection. Second, by sorting anomaly images topologically and applying continuity threshold judgment, it intelligently distinguishes between wide-area faults and local faults, effectively optimizing the allocation of operation and maintenance resources. Furthermore, by identifying the "aggravation" points of the anomaly area in the current direction, it divides the fault-affected area, laying the foundation for accurate location. Finally, by employing a multi-level location strategy including intra-sequence comparison, cross-sequence verification, and gradient analysis, it can accurately determine whether the fault source is located in the connection device itself or in the upstream conductor section. This overall solution achieves intelligent operation throughout the entire process from early warning and fault type identification to precise fault source location, significantly improving fault investigation efficiency, location accuracy, and power supply reliability, while substantially reducing operation and maintenance costs. Attached Figure Description

[0024] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0025] Figure 1 A flowchart illustrating a conductor fault location method based on a multi-source distribution network connection device, provided in an embodiment of the present invention;

[0026] Figure 2 This is a structural block diagram of a conductor fault location system based on a multi-source power distribution network connection device, provided in an embodiment of the present invention.

[0027] Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0029] Please see Figure 1 The diagram shows a flowchart of a conductor fault location method based on a multi-source distribution network connection device according to this application.

[0030] like Figure 1 As shown, the conductor fault location method based on multi-source distribution network connection device specifically includes the following steps:

[0031] Step S101: Obtain surface visual images of each multi-source power distribution network connection device on the conductor, and select at least one candidate surface visual image from each surface visual image according to the preset image selection rules.

[0032] In this step, a thermosensitive material is coated on the surface of the multi-source power distribution network connection device, and a visual image of the surface of each multi-source power distribution network connection device on the conductor is obtained, that is, the visual image of the surface of the multi-source power distribution network connection device before and after the thermosensitive material is heated.

[0033] Specifically, image feature analysis is performed on each surface visual image according to a preset image recognition model to extract abnormal visual features from each surface visual image. These abnormal visual features include at least one of color anomaly regions, structural deformation regions, or surface discharge spots. Based on these abnormal visual features, an anomaly confidence score is calculated for each surface visual image. The effective current values ​​located at each multi-source power distribution network connection device and acquired at the same time as each surface visual image are obtained. Based on the anomaly confidence score and the corresponding effective current value, a preset fusion judgment rule is used to select at least one candidate surface visual image from all surface visual images. Specifically, the fusion judgment rule involves: for each surface... The anomaly confidence score and current effective value corresponding to the visual image are normalized to obtain standardized anomaly confidence scores and standardized current effective values. A comprehensive anomaly score is calculated for each surface visual image based on the standardized anomaly confidence score and standardized current effective value, using the formula: Comprehensive Anomaly Score = α × Standardized Anomaly Confidence Score + β × Standardized Current Effective Value, where α and β are preset weighting coefficients, and α + β = 1, α > 0, β > 0. The comprehensive anomaly score is compared with a preset comprehensive score threshold. If the comprehensive anomaly score of a certain surface visual image is greater than or equal to the comprehensive score threshold, then that surface visual image is selected as a candidate surface visual image.

[0034] In one specific embodiment, to significantly improve the visual detection sensitivity of early overheating faults (usually a direct manifestation of loose wire connections and increased contact resistance), we first coat the key contact surfaces of multi-source power distribution network connection devices (such as clamps, terminals, etc.) with a special thermochromic material. This thermochromic material has reversible thermochromic properties, and its color or hue changes significantly and identifiably as the temperature rises. For example, it changes from a specific color at room temperature (such as blue) to another color (such as white) after exceeding a set threshold temperature (such as 60°C).

[0035] Real-time acquired surface visual images are input into a pre-defined image recognition model (e.g., a classification or feature extraction model based on a deep convolutional neural network (CNN)). The image recognition model is trained on a large amount of image data containing normal and abnormal connections and can automatically perform image feature analysis.

[0036] Image recognition models focus on extracting abnormal visual features from images. Due to the presence of thermosensitive materials, "color abnormalities" become the most crucial detection feature, identifying areas where the thermosensitive material has changed color due to overheating. In addition, the image recognition model also simultaneously detects "structural deformation areas" (such as metal melting or bulging caused by electric arcs or prolonged overheating) and "surface discharge spots" (such as corona discharges or electric arc spots captured at night or in dim environments).

[0037] Based on the saliency, area, and number of the extracted anomalous visual features, the image recognition model outputs a quantified anomalous confidence score (e.g., a value between 0 and 1). The higher the score, the greater the likelihood that there is an anomaly in the image.

[0038] Simultaneously, the effective current value, perfectly synchronized with the image acquisition time, is read from current sensors (such as CTs) installed in each multi-source power distribution network connection device. Overcurrent is a significant cause of overheating at connection points, and this electrical quantity serves as important auxiliary evidence for diagnosing anomalies.

[0039] For each connection device, the anomaly confidence score and RMS current value are normalized to scale them to a uniform, dimensionless range (e.g., 0-1). This eliminates the differences in dimensions and orders of magnitude between the visual scores and current values, making them comparable. Normalization formulas such as Min-Max normalization can be used.

[0040] A weighted summation method is used to fuse the normalized visual and electrical information. The calculated comprehensive anomaly score of each image is then compared with a preset comprehensive score threshold (e.g., 0.75). If the comprehensive anomaly score of a certain surface visual image is greater than or equal to the threshold, it is determined that the connection device is highly likely to be in an abnormal state, and this image is then selected as a candidate surface visual image for subsequent detailed analysis.

[0041] In summary, firstly, coating the surface of the connecting device with a thermosensitive material transforms the abstract "temperature" information into intuitive "color" changes, significantly enhancing the sensitivity and recognizability of visual inspection for early overheating faults. This allows both the naked eye and image recognition models to detect anomalies in their early stages. Secondly, by fusing the anomaly confidence score of the surface visual image with the effective current value at the same time, a comprehensive anomaly score is constructed, achieving cross-dimensional collaborative verification of physical appearance anomalies and electrical operating parameters. This not only significantly improves the accuracy and reliability of initial fault screening and effectively avoids visual false alarms caused by environmental factors such as changes in lighting and obstruction by foreign objects, but also overcomes the limitation that relying solely on current parameters cannot effectively identify non-overcurrent overheating faults such as increased contact resistance. Finally, this step provides a high-quality, high-reliability candidate image set for the entire fault location method, laying a solid and reliable data foundation for subsequent precise location, thereby achieving early warning and accurate diagnosis of conductor faults in the distribution network.

[0042] Step S102: Based on a preset region extraction strategy, perform abnormal region extraction on the at least one candidate surface visual image to obtain an abnormal region image corresponding to the at least one candidate surface visual image.

[0043] In this step, the candidate surface visual image is input into a pre-trained anomaly region segmentation model, which is a semantic segmentation model based on an encoder-decoder structure. The anomaly region segmentation model processes the image and outputs a pixel-level segmentation mask of the same size as the candidate surface visual image, wherein each pixel is labeled as belonging to or not belonging to an anomaly region. Based on the segmentation mask, the corresponding anomaly region is extracted from the original candidate surface visual image to generate an independent anomaly region image.

[0044] In one specific embodiment, the candidate surface visual images selected in step S101 are used as input data and fed into a pre-trained anomaly region segmentation model. Before input, the images need to undergo uniform preprocessing operations, including but not limited to: scaling the size to the fixed resolution required by the model (such as 512x512 pixels), grayscale conversion or RGB three-channel normalization, and pixel value normalization to the [0,1] interval, in order to improve the processing efficiency and generalization ability of the model.

[0045] The anomaly region segmentation model used in this implementation is a semantic segmentation model based on an encoder-decoder structure, such as U-Net or its variants. This model structure is particularly suitable for pixel-level image segmentation tasks.

[0046] The encoder (downsampling path) typically consists of a series of convolutional and pooling layers, responsible for extracting multi-level features from the input image. Its function is to progressively capture contextual information in the image, identifying high-level semantic features such as "anomalies," but this also reduces the spatial resolution of the feature maps.

[0047] Decoder (upsampling path): Typically composed of a series of transposed convolutional layers or upsampling layers, it is responsible for progressively restoring the deep, abstract feature maps extracted by the encoder to the original image size. Its function is to achieve precise localization, mapping abstract semantic information back to every pixel of the original image.

[0048] Skip connections: one of the core features of the model, it merges the high-resolution and low-resolution feature maps of the corresponding level of the encoder with the feature maps of the same level of the decoder. This effectively compensates for the spatial details lost during downsampling, enabling the model to simultaneously consider both global context and precise local localization.

[0049] After forward propagation processing by the model, a segmentation mask with the exact same size as the input candidate image is output. This mask is a two-dimensional matrix, where each pixel is labeled with a category: for example, "1" represents that the pixel belongs to an anomalous region (such as the color-changing area of ​​a thermosensitive material or a discharge spot), and "0" represents that it does not belong to an anomalous region. This allows the boundaries of anomalous regions to be accurately delineated.

[0050] After obtaining the segmentation mask, it is used as a "guide" to the original candidate surface visual image. Specifically, the pixel regions in the original candidate image corresponding to all pixels marked "1" (i.e., abnormal regions) in the segmentation mask are completely cropped or copied.

[0051] Finally, for each candidate surface visual image, one or more independent images containing only the anomalous region are generated, i.e., anomalous region images. These images remove most of the normal background, allowing the focus to be entirely on the fault features themselves.

[0052] Step S103: Sort the images of each abnormal area according to the location information of each multi-source power distribution network connection device to obtain an abnormal area image sequence, and determine whether the number of images of the abnormal area in the abnormal area image sequence that are consecutively located is greater than a preset number threshold.

[0053] In this step, the node numbers of each multi-source distribution network connection device are obtained from the preset distribution network topology database; based on the node numbers and the node order from the power source side to the load side, the abnormal area images corresponding to each multi-source distribution network connection device are sorted to obtain the abnormal area image sequence.

[0054] In one specific embodiment, after determining whether the number of consecutive abnormal region images in the abnormal region image sequence is greater than a preset number threshold, if it is greater than the preset number threshold, then it is determined that the entire conductor is faulty, and a conductor fault abnormal signal is generated.

[0055] For example, the system automatically retrieves the node numbers corresponding to all multi-source distribution network connection devices for all generated anomaly area images from a pre-defined distribution network topology database. This database is the core asset management database of the power system, clearly defining the unique numbers of all equipment in the power grid (including switches, towers, connection devices, etc.) and their connection relationships, forming a complete power grid topology.

[0056] Based on the topology, the power supply side (e.g., substation outlet) and load side (e.g., user access point) of the entire line are identified, and the dominant direction of the current is determined. All abnormal region images are then sorted according to the node sequence from the power supply side to the load side. For example, if a line has connection devices with node numbers A01 (power supply side), A02, A03, and A04 (load side), and abnormal region images are generated for A01, A03, and A04, then the sorted abnormal region image sequence is obtained.

[0057] Analyze the image sequence of the abnormal region and determine whether the number of abnormal region images that are consecutive in position (i.e., adjacent nodes in the topology) in the sequence is greater than a preset number threshold.

[0058] The preset number threshold is usually set based on the total length of the line and the node density. Its value may be an absolute number (such as "5 consecutive nodes") or a proportion relative to the total number of nodes in the line (such as "more than 70% of the nodes").

[0059] If the number of consecutive abnormal images in the sequence exceeds a preset threshold—for example, on a line with 10 nodes, if 7 consecutive nodes from A02 to A08 show abnormalities—this strongly indicates that the fault is not localized but affects the entire line. In this case, the system will determine it as a global, wide-area fault (such as overall line overload, common environmental erosion, etc.) and directly generate a "conductor fault anomaly signal," prompting maintenance personnel to investigate and replace the entire line without needing to perform subsequent fine-tuning steps.

[0060] Conversely, if the number of consecutive abnormal images is not greater than (i.e. less than or equal to) the threshold, it indicates that the abnormality is localized, and the system will proceed to step S104 for more in-depth fault source localization analysis.

[0061] In summary, by determining whether the number of consecutive abnormal images exceeds a preset threshold, this method achieves rapid and effective initial screening of fault types, accurately distinguishing between wide-area line faults and local faults. This judgment is crucial, as it immediately terminates unnecessary fine-grained localization processes for global faults, directly leading to efficient operation and maintenance decisions (such as full-line overhaul), while ensuring that computational resources are concentrated on in-depth analysis of local faults. This greatly improves the decision-making intelligence and operation and maintenance efficiency of the entire fault localization method, avoiding misjudgments and resource waste.

[0062] Step S104: If the number is not greater than a preset threshold, at least one target abnormal region image is extracted from the abnormal region image sequence, and the abnormal region image sequence is divided according to the at least one target abnormal region image to obtain at least one abnormal region image subsequence.

[0063] In this step, taking the current transmission direction in the conductor as the positive direction, the first abnormal region image in the abnormal region image sequence is defined as the starting abnormal region image, and the last abnormal region image in the abnormal region image sequence is defined as the ending abnormal region image. It is determined whether the area of ​​the first abnormal region in the first abnormal region image in the abnormal region image sequence is greater than the area of ​​the starting abnormal region in the starting abnormal region image, wherein the first abnormal region image is the abnormal region image adjacent to the starting abnormal region image. If it is greater than the area of ​​the starting abnormal region in the starting abnormal region image, then the first abnormal region image is defined as the target abnormal region image; otherwise, it is not positioned as a target abnormal region image. The abnormal region image sequence is divided using each target abnormal region image as a dividing point to obtain at least one abnormal region image subsequence.

[0064] In one specific embodiment, the direction of current transmission is defined as the positive direction (e.g., from the substation to the user end). Along this positive direction, the first image in the sorted sequence of abnormal region images (i.e., the abnormal image closest to the power source side) is defined as the starting abnormal region image, and the last image in the sequence (i.e., the abnormal image closest to the load side) is defined as the ending abnormal region image.

[0065] Starting with the initial abnormal region image, it is compared with the next adjacent abnormal region image (i.e., the first abnormal region image) along the positive direction of the current. The core comparison parameter is the area of ​​the abnormal region (this area data has been calculated by pixels when the abnormal region image is generated in step S102).

[0066] If the area of ​​the anomalous region in the first anomalous region image is larger than the area of ​​the anomalous region in the initial anomalous region image, it indicates that the degree of anomalousness has intensified during the propagation process from upstream to downstream. This point of intensification is often a significant marker of the propagation of fault effects; therefore, this first anomalous region image is defined as a target anomalous region image.

[0067] If the area does not increase, it indicates that the degree of anomaly has stabilized or weakened, and the node is not marked as a target image. Following this logic, each pair of adjacent images in the sequence (i.e., the i-th image and the (i+1)-th image) will be compared sequentially to extract all target anomaly region images in the sequence that meet the "anomaly intensification" condition.

[0068] Using all identified target anomaly region images as key segmentation points, the original anomaly region image sequence is segmented into several anomaly region image subsequences. The segmentation rule is that each subsequence must begin with a target anomaly region image, and the subsequence contains this target image and all subsequent anomaly images up to the next target image. This rule ensures that each subsequence contains only one starting point of "anomaly aggravation," thus representing a relatively independent fault-affected region.

[0069] Suppose an abnormal sequence ordered by current direction is [ImgA, ImgB, ImgC, ImgD, ImgE]. Comparison reveals that the area of ​​ImgB is greater than the area of ​​ImgA, and the area of ​​ImgD is greater than the area of ​​ImgC. Therefore, ImgB and ImgD are identified as the target images. Using them as segmentation points, the original sequence is divided into: [ImgA], [ImgB, ImgC], and [ImgD, ImgE].

[0070] In summary, by capturing the spatial inflection points of anomalous area area, this method can intelligently identify abrupt changes that may occur as fault effects propagate along distribution network lines. These locations are often key points where fault energy is concentratedly released or the nature of the fault changes. This sequence partitioning not only significantly reduces the computational complexity of subsequent fault source localization algorithms, but more importantly, it creates the necessary conditions for accurate fault source localization. This is because it effectively distinguishes different fault influence domains, allowing the localization analysis to focus on the local interval most likely containing the true fault source. This avoids mutual interference between multiple fault influence intervals, greatly improving the accuracy and reliability of the final fault localization results.

[0071] Step S105: Based on the at least one abnormal region image subsequence, a preset fault location strategy is used to locate the fault source on the conductor.

[0072] In this step, the area of ​​the abnormal region corresponding to each abnormal region image in a certain abnormal region image subsequence is obtained;

[0073] Identify the source anomalous region image with the largest anomalous region area in a certain anomalous region image subsequence, and preliminarily locate the multi-source distribution network connection device corresponding to the source anomalous region image as a potential fault source device.

[0074] Calculate the average area of ​​a certain abnormal region in a subsequence of images of an abnormal region.

[0075] Calculate a certain area ratio between the area of ​​an abnormal region in a certain source abnormal region image and the area of ​​a certain average abnormal region, and determine whether the area ratio is less than a preset ratio threshold.

[0076] If a certain area ratio is not less than a preset ratio threshold, then the location of a potential fault source device is determined to be a fault source location.

[0077] If a certain area ratio is less than a preset ratio threshold, then calculate the other average abnormal area of ​​all abnormal area images in another abnormal area image subsequence, and determine whether the other area ratio of the abnormal area of ​​a certain source abnormal area image to the other average abnormal area is less than a preset ratio threshold.

[0078] If another area ratio is not less than a preset ratio threshold, then the location of a potential fault source device is determined to be a fault source location.

[0079] If the other area ratio is less than the preset ratio threshold, the fault source is determined to be located on the conductor section between the upstream adjacent multi-source distribution network connection devices of a certain potential fault source device based on the gradient change of the abnormal area area along the current transmission direction in a certain abnormal area image subsequence.

[0080] In one specific embodiment, for a certain abnormal region image subsequence (denoted as subsequence X) that needs to be analyzed, the abnormal region area corresponding to each abnormal region image in the subsequence is first obtained (this data has been obtained by pixel statistics in step S102).

[0081] Within this subsequence X, the image with the largest abnormal area is identified and denoted as a source abnormal area image. The multi-source distribution network connection device corresponding to this image is initially located as a potential fault source device (e.g., device X).

[0082] Calculate the average area of ​​a certain abnormal region in all abnormal region images within subsequence X.

[0083] Calculate the area ratio R1 = (area of ​​the source abnormal region image) / (average area of ​​the abnormal region in subsequence X).

[0084] If R1 ≥ the threshold, it indicates that the anomaly of the potential fault source device is significantly higher than that of other devices in its subsequence, and it is highly likely to be the fault source itself. Therefore, the location of device X can be directly determined as a fault source location.

[0085] If R1 < threshold, then another abnormal region image subsequence (denoted as subsequence Y) is selected. Subsequence Y is usually the upstream neighboring subsequence of the current subsequence X, because the effects of the fault usually propagate along the current direction.

[0086] Calculate the average area of ​​another abnormal region in the images of all abnormal regions within subsequence Y.

[0087] The area ratio R2 is calculated as follows: (Area of ​​the source abnormal region image) / (Average area of ​​the abnormal region in subsequence Y).

[0088] If R² ≥ the threshold, it indicates that the anomaly of the potential fault source device is significant, even if it is not prominent within its own subsequence, compared to the upstream normal or slightly abnormal segments. Therefore, device X is still determined to be a fault source location.

[0089] If R² < threshold, gradient analysis and conductor segment location are performed: if both ratio tests fail, it indicates that the "source" of the fault may not be device X itself, but rather the conductor upstream of it. The heat or electromagnetic effects generated by the faulty conductor affect the nearest downstream device, causing it to exhibit the largest abnormal area.

[0090] At this point, the system analyzes the gradient change of the area of ​​the abnormal region in subsequence X along the direction of current transmission. Specifically, it checks whether the area of ​​the abnormal region shows a stable increasing trend from the starting point of subsequence X (i.e., a target abnormal image) to X.

[0091] If a significant increasing gradient is confirmed, the fault source is ultimately determined to be located on the conductor section between the potential fault source device and its upstream adjacent multi-source distribution network connection device. For example, the fault location result can be precisely stated as "the fault point is located on the conductor between node A03 and node A04".

[0092] By employing a two-tiered internal and external ratio verification method, the reliability of the location results is significantly improved. When both levels of verification fail, analyzing the gradient changes in the abnormal area further pinpoints the fault source from the device to a specific conductor segment, achieving a leapfrog improvement in location accuracy. This rigorous, multi-layered judgment mechanism ensures that, regardless of whether the fault is a simple single-point fault or a complex multi-point impact scenario, this method can provide highly reliable location conclusions. This provides maintenance personnel with extremely accurate repair targets, significantly shortening fault diagnosis and repair time and improving power supply reliability.

[0093] In summary, the method presented in this application achieves a significant breakthrough in power distribution network conductor fault location technology by integrating visual enhancement with thermal materials, multi-source information fusion, topology sequence analysis, and a multi-level positioning strategy. First, by utilizing thermal materials coated on the surface of the connecting device, abstract temperature rise information is transformed into intuitive color changes. This is combined with visual anomaly features and synchronous current values ​​for fusion judgment, greatly improving the sensitivity and reliability of early fault detection. Second, by sorting anomaly images topologically and applying continuity threshold judgment, wide-area faults and local faults are intelligently distinguished, effectively optimizing the allocation of maintenance resources. Furthermore, by identifying the "aggravation" points of the anomaly area in the current direction, the fault impact range is divided, laying the foundation for accurate location. Finally, by employing a multi-level positioning strategy including intra-sequence comparison, cross-sequence verification, and gradient analysis, the fault source can be accurately determined whether it is located in the connecting device itself or in the upstream conductor section. This overall solution achieves intelligent operation throughout the entire process from early warning and fault type identification to precise fault source location, significantly improving fault investigation efficiency, location accuracy, and power supply reliability, while substantially reducing maintenance costs.

[0094] Please see Figure 2 The diagram shows a structural block diagram of a conductor fault location system based on a multi-source distribution network connection device according to this application.

[0095] like Figure 2 As shown, the conductor fault location system 200 includes an acquisition module 210, an extraction module 220, a judgment module 230, a division module 240, and a location module 250.

[0096] The acquisition module 210 is configured to acquire surface visual images of each multi-source distribution network connection device on the conductor, and select at least one candidate surface visual image from each surface visual image according to a preset image selection rule; the extraction module 220 is configured to extract abnormal regions from the at least one candidate surface visual image based on a preset region extraction strategy to obtain an abnormal region image corresponding to the at least one candidate surface visual image; the judgment module 230 is configured to sort each abnormal region image according to the location information of each multi-source distribution network connection device to obtain an abnormal region image sequence, and determine whether the number of consecutive abnormal region images in the abnormal region image sequence is greater than a preset number threshold; the division module 240 is configured to extract at least one target abnormal region image from the abnormal region image sequence if it is not greater than the preset number threshold, and divide the abnormal region image sequence according to the at least one target abnormal region image to obtain at least one abnormal region image subsequence; the positioning module 250 is configured to locate the fault source on the conductor according to the at least one abnormal region image subsequence using a preset fault positioning strategy.

[0097] It should be understood that Figure 2 The modules and references described in the document Figure 1 The steps described in the text correspond to those in the method described above. Therefore, the operations, features, and corresponding technical effects described above also apply to the method described in the text. Figure 2 The various modules in the document will not be described in detail here.

[0098] In other embodiments, the present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein when the program instructions are executed by a processor, the processor performs the conductor fault location method based on a multi-source distribution network connection device in any of the above method embodiments.

[0099] In one embodiment, the computer-readable storage medium of the present invention stores computer-executable instructions, which are configured as follows:

[0100] Acquire surface visual images of each multi-source power distribution network connection device on the conductor, and select at least one candidate surface visual image from each surface visual image according to the preset image selection rules.

[0101] Based on a preset region extraction strategy, abnormal regions are extracted from the at least one candidate surface visual image to obtain an abnormal region image corresponding to the at least one candidate surface visual image.

[0102] Based on the location information of each multi-source power distribution network connection device, the images of each abnormal area are sorted to obtain an abnormal area image sequence, and it is determined whether the number of images of the abnormal area in the abnormal area image sequence that are consecutively located is greater than a preset number threshold.

[0103] If the number is not greater than a preset threshold, at least one target abnormal region image is extracted from the abnormal region image sequence, and the abnormal region image sequence is divided according to the at least one target abnormal region image to obtain at least one abnormal region image subsequence.

[0104] Based on the at least one abnormal region image subsequence, a preset fault location strategy is used to locate the fault source on the conductor.

[0105] Computer-readable storage media may include a stored program area and a stored data area, wherein the stored program area may store an operating system and an application program required for at least one function; the stored data area may store data created based on the use of the conductor fault location system based on the multi-source distribution network connection device. Furthermore, the computer-readable storage medium may include high-speed random access memory, and may also include memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state storage device. In some embodiments, the computer-readable storage medium may optionally include memory remotely disposed relative to a processor, which can be connected to the conductor fault location system based on the multi-source distribution network connection device via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0106] Figure 3 This is a schematic diagram of the structure of the electronic device provided in the embodiment of the present invention, such as... Figure 3 As shown, the device includes a processor 310 and a memory 320. The electronic device may also include an input device 330 and an output device 340. The processor 310, memory 320, input device 330, and output device 340 can be connected via a bus or other means. Figure 3 Taking a bus connection as an example, the memory 320 is the computer-readable storage medium described above. The processor 310 executes various server functions and data processing by running non-volatile software programs, instructions, and modules stored in the memory 320, thereby implementing the conductor fault location method based on a multi-source distribution network connection device as described in the above embodiment. The input device 330 can receive input digital or character information and generate key signal inputs related to user settings and function control of the conductor fault location system based on a multi-source distribution network connection device. The output device 340 may include a display screen or other display device.

[0107] The aforementioned electronic device can execute the method provided in the embodiments of the present invention, and has the corresponding functional modules and beneficial effects for executing the method. Technical details not described in detail in this embodiment can be found in the method provided in the embodiments of the present invention.

[0108] In one implementation, the above-described electronic device is applied to a conductor fault location system based on a multi-source distribution network connection device, for a client, and includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to:

[0109] Acquire surface visual images of each multi-source power distribution network connection device on the conductor, and select at least one candidate surface visual image from each surface visual image according to the preset image selection rules.

[0110] Based on a preset region extraction strategy, abnormal regions are extracted from the at least one candidate surface visual image to obtain an abnormal region image corresponding to the at least one candidate surface visual image.

[0111] Based on the location information of each multi-source power distribution network connection device, the images of each abnormal area are sorted to obtain an abnormal area image sequence, and it is determined whether the number of images of the abnormal area in the abnormal area image sequence that are consecutively located is greater than a preset number threshold.

[0112] If the number is not greater than a preset threshold, at least one target abnormal region image is extracted from the abnormal region image sequence, and the abnormal region image sequence is divided according to the at least one target abnormal region image to obtain at least one abnormal region image subsequence.

[0113] Based on the at least one abnormal region image subsequence, a preset fault location strategy is used to locate the fault source on the conductor.

[0114] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of various embodiments or some parts of embodiments.

[0115] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for locating conductor faults based on a multi-source distribution network connection device, characterized in that, include: Acquire surface visual images of each multi-source distribution network connection device on the conductor, and select at least one candidate surface visual image from each surface visual image according to a preset image selection rule, including: Based on a preset image recognition model, image feature analysis is performed on each surface visual image to extract abnormal visual features from each surface visual image. The abnormal visual features include at least one of color abnormality region, structural deformation region or surface discharge spot. Based on the aforementioned abnormal visual features, an anomaly confidence score is calculated for each surface visual image; Obtain the effective current value located at each multi-source power distribution network connection device at the same acquisition time as each surface visual image; Based on the anomaly confidence score and the corresponding effective current value, at least one candidate surface visual image is selected from all surface visual images using a preset fusion judgment rule. Based on a preset region extraction strategy, abnormal regions are extracted from the at least one candidate surface visual image to obtain an abnormal region image corresponding to the at least one candidate surface visual image. Based on the location information of each multi-source power distribution network connection device, the images of each abnormal area are sorted to obtain an abnormal area image sequence, and it is determined whether the number of images of the abnormal area in the abnormal area image sequence that are consecutively located is greater than a preset number threshold. If the number is not greater than a preset threshold, at least one target abnormal region image is extracted from the abnormal region image sequence, and the abnormal region image sequence is divided according to the at least one target abnormal region image to obtain at least one abnormal region image subsequence. Based on the at least one abnormal region image subsequence, a preset fault location strategy is used to locate the fault source on the conductor, including: Obtain the area of ​​the abnormal region corresponding to each abnormal region image in a certain abnormal region image subsequence; Identify the source abnormal region image with the largest abnormal region area in the image subsequence of a certain abnormal region, and preliminarily locate the multi-source power distribution network connection device corresponding to the source abnormal region image as a potential fault source device. Calculate the average area of ​​a certain abnormal region in all abnormal region images in a certain abnormal region image subsequence; Calculate a certain area ratio between the area of ​​the abnormal region in a certain source abnormal region image and the area of ​​a certain average abnormal region, and determine whether the certain area ratio is less than a preset ratio threshold. If a certain area ratio is not less than a preset ratio threshold, then the location of a certain potential fault source device is determined to be a fault source location. If the area ratio is less than a preset ratio threshold, then calculate the other average area of ​​all abnormal region images in the other abnormal region image subsequence, and determine whether the other area ratio of the abnormal region area of ​​the source abnormal region image to the other average area of ​​the abnormal region is less than the preset ratio threshold. If the other area ratio is not less than a preset ratio threshold, then the location of a potential fault source device is determined to be a fault source location. If the other area ratio is less than a preset ratio threshold, then based on the gradient change of the abnormal area area along the current transmission direction in the image subsequence of a certain abnormal area, it is determined that the fault source is located on the conductor section between the upstream adjacent multi-source distribution network connection devices of a certain potential fault source device.

2. The method for locating conductor faults based on a multi-source distribution network connection device according to claim 1, characterized in that, in, The specific fusion determination rule is as follows: The anomaly confidence score and current RMS value corresponding to each surface visual image are normalized to obtain standardized anomaly confidence scores and standardized current RMS values. The comprehensive anomaly score for each surface visual image is calculated based on the standardized anomaly confidence score and the standardized current effective value. The formula is: Comprehensive anomaly score = α × standardized anomaly confidence score + β × standardized current effective value, where α and β are preset weighting coefficients, and α + β = 1, α > 0, β > 0. The comprehensive anomaly score is compared with a preset comprehensive score threshold; If the overall anomaly score of a certain surface visual image is greater than or equal to the overall score threshold, then the certain surface visual image is selected as a candidate surface visual image.

3. The method for locating conductor faults based on a multi-source distribution network connection device according to claim 1, characterized in that, The step of extracting abnormal regions from the at least one candidate surface visual image based on a preset region extraction strategy to obtain an abnormal region image corresponding to the at least one candidate surface visual image includes: The candidate surface visual image is input into a pre-trained anomaly region segmentation model, which is a semantic segmentation model based on an encoder-decoder structure. The abnormal region segmentation model is used to process the data and output a pixel-level segmentation mask of the same size as the candidate surface visual image. Each pixel is labeled as belonging to or not belonging to the abnormal region. Based on the segmentation mask, the corresponding abnormal regions are extracted from the original candidate surface visual image, and independent abnormal region images are generated.

4. The conductor fault location method based on a multi-source distribution network connection device according to claim 1, characterized in that, The step of sorting the images of each abnormal area based on the location information of each multi-source distribution network connection device to obtain the abnormal area image sequence includes: Obtain the node number of each multi-source distribution network connection device from the preset distribution network topology database; Based on the node numbering and the node order from the power supply side to the load side, the abnormal area images corresponding to each multi-source distribution network connection device are sorted to obtain the abnormal area image sequence.

5. The method for locating conductor faults based on a multi-source distribution network connection device according to claim 1, characterized in that, After determining whether the number of consecutively located abnormal region images in the abnormal region image sequence is greater than a preset threshold, the method further includes: If the number exceeds a preset threshold, the entire conductor is determined to be faulty, and a conductor fault abnormality signal is generated.

6. The method for locating conductor faults based on a multi-source distribution network connection device according to claim 1, characterized in that, The step of extracting at least one target abnormal region image from the abnormal region image sequence and dividing the abnormal region image sequence according to the at least one target abnormal region image to obtain at least one abnormal region image subsequence includes: Taking the current transmission direction in the conductor as the positive direction, the first abnormal region image in the abnormal region image sequence is defined as the starting abnormal region image along the positive direction, and the last abnormal region image in the abnormal region image sequence is defined as the ending abnormal region image. Determine whether the area of ​​the first abnormal region in the first abnormal region image in the abnormal region image sequence is greater than the area of ​​the starting abnormal region image, wherein the first abnormal region image is: an abnormal region image adjacent to the starting abnormal region image; If the area of ​​the first abnormal region is greater than the area of ​​the initial abnormal region image, then the first abnormal region image is defined as the target abnormal region image; otherwise, it is not positioned as the target abnormal region image. Using each target abnormal region image as a segmentation point, the abnormal region image sequence is divided to obtain at least one abnormal region image subsequence.

7. A conductor fault location system based on a multi-source distribution network connection device, characterized in that, include: The acquisition module is configured to acquire surface visual images of various multi-source distribution network connection devices on the conductor, and select at least one candidate surface visual image from the various surface visual images according to preset image selection rules, including: Based on a preset image recognition model, image feature analysis is performed on each surface visual image to extract abnormal visual features from each surface visual image. The abnormal visual features include at least one of color abnormality region, structural deformation region or surface discharge spot. Based on the aforementioned abnormal visual features, an anomaly confidence score is calculated for each surface visual image; Obtain the effective current value located at each multi-source power distribution network connection device at the same acquisition time as each surface visual image; Based on the anomaly confidence score and the corresponding effective current value, at least one candidate surface visual image is selected from all surface visual images using a preset fusion judgment rule. The extraction module is configured to extract abnormal regions from the at least one candidate surface visual image based on a preset region extraction strategy, so as to obtain an abnormal region image corresponding to the at least one candidate surface visual image. The judgment module is configured to sort the images of each abnormal area according to the location information of each multi-source power distribution network connection device, obtain an abnormal area image sequence, and determine whether the number of images of the abnormal area in the abnormal area image sequence that are consecutively located is greater than a preset number threshold. The segmentation module is configured to extract at least one target abnormal region image from the abnormal region image sequence if the number is not greater than a preset threshold, and to segment the abnormal region image sequence according to the at least one target abnormal region image to obtain at least one abnormal region image subsequence. The positioning module is configured to locate the fault source on the conductor based on the at least one abnormal region image subsequence and using a preset fault positioning strategy, including: obtaining the abnormal region area corresponding to each abnormal region image in a certain abnormal region image subsequence; Identify the source abnormal region image with the largest abnormal region area in the image subsequence of a certain abnormal region, and preliminarily locate the multi-source power distribution network connection device corresponding to the source abnormal region image as a potential fault source device. Calculate the average area of ​​a certain abnormal region in all abnormal region images in a certain abnormal region image subsequence; Calculate a certain area ratio between the area of ​​the abnormal region in a certain source abnormal region image and the area of ​​a certain average abnormal region, and determine whether the certain area ratio is less than a preset ratio threshold. If a certain area ratio is not less than a preset ratio threshold, then the location of a certain potential fault source device is determined to be a fault source location. If the area ratio is less than a preset ratio threshold, then calculate the other average area of ​​all abnormal region images in the other abnormal region image subsequence, and determine whether the other area ratio of the abnormal region area of ​​the source abnormal region image to the other average area of ​​the abnormal region is less than the preset ratio threshold. If the other area ratio is not less than a preset ratio threshold, then the location of a potential fault source device is determined to be a fault source location. If the other area ratio is less than a preset ratio threshold, then based on the gradient change of the abnormal area area along the current transmission direction in the image subsequence of a certain abnormal area, it is determined that the fault source is located on the conductor section between the upstream adjacent multi-source distribution network connection devices of a certain potential fault source device.

8. An electronic device, characterized in that, include: At least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by a processor, it implements the method described in any one of claims 1 to 6.

Images