Power transmission line ground wire heating detection method based on infrared image space positioning

Abstract

This invention discloses a method for detecting overheating of transmission line conductors and ground wires based on infrared image spatial positioning. It comprises an infrared image acquisition module, a conductor and related component extraction module, a conductor and ground wire judgment module, and a conductor and ground wire overheating detection module. The conductor and related component extraction module includes a conductor extraction module, a U-shaped hanging ring extraction module, and a tower head extraction module. This achieves fully automated detection throughout the entire process, from automatic access to UAV infrared images, to precise segmentation of key components, to intelligent identification of conductors and ground wires based on spatial topology, and finally to comprehensive judgment of multi-component related overheating. The beneficial effects of this invention are: it can integrate infrared temperature information and image spatial structure information to accurately distinguish between conductors and ground wires and drain wires; it can perform integrated positioning and correlation analysis of abnormal overheating of conductors and ground wires with their associated tower damage fittings, achieving a more accurate and reliable diagnosis of conductor and ground wire overheating defects and their root causes.

Description

Technical Field

[0001] This invention relates to the technical field of infrared image detection of power grid transmission lines, and in particular to a method for detecting the heating of conductors and ground wires of transmission lines based on spatial positioning of infrared images. Background Technology

[0002] As the physical carrier of electrical energy transmission in a power system, the reliability and stability of transmission lines directly affect the safety of the entire power grid. Line fittings, including various metal components connecting, fixing, and protecting the ground line, are prone to abnormal heating during long-term operation due to aging, corrosion, poor contact, or overload. This not only reflects energy loss but also serves as an early sign of potential faults or even electrical fires. Therefore, regular and effective inspection of transmission line fittings is a crucial step in achieving condition monitoring and preventing accidents. Among numerous detection technologies, infrared thermal imaging technology, with its advantages of non-contact, visualization, and quantitative temperature measurement, has become the mainstream and standardized method for transmission line condition inspection, especially for detecting heating defects. Aerial photography using drones equipped with infrared thermal imagers can efficiently and safely acquire infrared image data of large areas of the lines.

[0003] Among various hardware heating defects, abnormal heating of the ground wire is relatively common. As the main current discharge path, the ground wire's heating may originate from increased contact resistance at the clamp, increased local resistance due to broken strands or damage, or corona or arcing with adjacent components. Accurately identifying and locating the heating point of the ground wire is crucial for timely troubleshooting and preventing the escalation of faults. However, current automated or semi-automated analysis based on infrared images faces several significant technical bottlenecks in detecting ground wire heating: Target confusion leads to identification difficulties: In infrared images, especially overhead or side views of complex tower structures, ground wires and drain wires (jump wires) are often highly similar in shape, direction, and temperature readings. Traditional image recognition algorithms based on shape, texture, or simple temperature thresholds struggle to reliably and accurately distinguish them. In reality, due to their structural characteristics and the kinetic relationship of the current path during normal operation, drain wires have a much lower probability of heating up than ground wires, which carry the main load current. Misjudging drain wires as heated ground wires will generate numerous false positive alarms, interfering with maintenance personnel's judgment and reducing detection efficiency; conversely, correctly identifying them may lead to missed detection of true defects.

[0004] The lack of a clear correlation between the root cause of overheating makes it difficult to accurately pinpoint the defect: Overheating of the grounding wire is often not an isolated phenomenon; its heat source is frequently and closely related to the heat generated by the hardware at the tower top grounding point or suspension point. For example, oxidation of the contact surface inside the clamp or loosening of the screw can cause a sharp increase in resistance, and the heat will first be conducted to the connected end of the grounding wire, causing an abnormal temperature rise in that section of the grounding wire. If the detection method only focuses on identifying the "grounding wire" itself as the heat source, severing its spatial and thermal conduction connection with nearby critical hardware, it is impossible to accurately trace and locate the most fundamental defect. This may lead to biased repair methods, addressing only the surface symptoms without resolving the core fault.

[0005] Existing methods do not adequately utilize spatial contextual information. Current mainstream detection processes rely on analyzing isolated two-dimensional infrared images of targets acquired by UAVs. These methods lack an understanding of the spatial structure of transmission lines, failing to construct the topological connections and precise spatial orientations between conductors, ground wires, connecting hardware, and towers. This makes the aforementioned target confusion and root cause correlation analysis difficult to achieve.

[0006] In summary, existing UAV-based infrared imagery-based power transmission line overheating detection technologies, when targeting specific objects like conductors and ground wires, suffer from limitations. These limitations stem from visual confusion between conductors / ground wires and drain wires in two-dimensional images, as well as the lack of analysis of the spatial topology and thermal logic relationships between the overheated conductor / ground wire and its associated hardware. Consequently, the accuracy of detection results, the precision of defect location, and the interpretability of alarms are all unsatisfactory. Maintenance personnel still need to invest significant effort in manual verification and review, and the level of automation needs improvement. Summary of the Invention

[0007] The present invention aims to overcome the above-mentioned deficiencies in the prior art and provides a method for detecting the heating of transmission line conductors and ground wires based on infrared image spatial positioning, which can realize the diagnosis of heating defects and their root causes.

[0008] To achieve the above objectives, the present invention adopts the following technical solution: The method for detecting conductor and ground wire heating in transmission lines based on infrared image spatial positioning comprises an infrared image acquisition module, a conductor and related component extraction module, a conductor and ground wire judgment module, and a conductor and ground wire heating detection module. The conductor and related component extraction module includes a conductor extraction module, a U-shaped hanging ring extraction module, and a tower head extraction module. The specific operation steps are as follows: (1) The infrared image acquisition module is responsible for acquiring raw data. This module automatically reads the infrared thermal image to be analyzed uploaded by the UAV inspection system from the designated cloud storage service. After reading, it performs basic format parsing and caching on the image to provide standardized input data for subsequent processing modules. (2) The conductor and related component extraction module, based on the conductor extraction module, U-shaped hanging ring extraction module and tower head extraction module, uses the trained model algorithm to segment and detect the conductor component set from the acquired infrared image. S_wire U-shaped hanging ring component assembly S_clamp and tower head component assembly S_tower ; (3) The conductor and ground wire judgment module integrates the spatial positions of the conductor component, U-shaped hanging ring component and tower head component in the infrared image to determine whether they belong to the conductor and ground wire, thereby obtaining the conductor and ground wire in the infrared image; (4) The ground wire heating detection module extracts the temperature of each component of the ground wire, extracts and judges the extreme temperature values, and makes a comprehensive judgment on the heating of the ground wire based on the spatial position relationship.

[0009] This invention provides an intelligent detection method that integrates infrared temperature information and image spatial structure information, aiming to accurately distinguish between ground wires and drain wires, and to perform integrated positioning and correlation analysis of abnormal heating of ground wires and their associated tower damage fittings, thereby achieving a more accurate and reliable diagnosis of ground wire heating defects and their root causes.

[0010] Preferably, in step (1), the infrared image acquisition module realizes the automated reception and identification management of infrared images transmitted back by the UAV inspection system. This module connects with the enterprise-level object storage service OBS through an interface, continuously listens for and acquires infrared image data files uploaded by the UAV inspection platform. Each infrared image is stored with its file identifier MFID as a unique file name. After receiving the infrared image file, the module directly parses the MFID from its file name. This MFID serves as the core identifier of the image in the entire system and is persisted in the metadata as a key index record.

[0011] Preferably, in step (2), the wire extraction module identifies and extracts the wire target from the complex background. This module is based on a pre-trained deep learning target detection and segmentation model, trained using an infrared image dataset labeled with the wire region. For the input infrared image, the model performs wire target detection and pixel-level segmentation, outputting a set of all wire instances in the image. The segmented set of wire instances is as follows:

[0012] in, Representing the j The bounding box coordinates of a wire instance. x_min and y_min The x and y coordinates of the top-left vertex of the bounding box are defined. x_max and y_maxThe x and y coordinates of the lower right corner vertex of the bounding box are defined; M_wj This represents its corresponding binary segmentation mask matrix. N_wire This indicates the total number of wire instances.

[0013] Preferably, in step (2), the U-shaped hanging ring extraction module identifies the key connection suspension components of the ground wire from the complex background. Based on a specially trained target detection and segmentation model, the model detects and segments all U-shaped hanging ring instances in the input infrared image, and the resulting set of segmentation results is as follows:

[0014] in, B_ck and M_ck Representing the first k Boundary box and segmentation mask of a U-shaped hanging loop. N_clamp This represents the total number of U-shaped hanging ring instances detected.

[0015] Preferably, in step (2), the tower head extraction module identifies and segments the tower head portion in the infrared image. The training set of the target detection and segmentation model includes infrared image annotations of four tower types: drum-shaped tower head, cat-head tower head, ram-horn tower head, and zig-shaped tower head. After processing the image through this model, the output set of tower head instances is as follows:

[0016] in, B_tl and M_tl Representing the first l The bounding box and segmentation mask of each tower head instance. N_tower This represents the number of tower head instances detected.

[0017] Preferably, in step (3), the conductor / ground wire determination module receives a set of conductor instances from the conductor and related component extraction module. S_wire Collection of U-shaped hanging ring examples S_clamp With the set of tower head instances S_tower Based on the unique spatial topology of conductors and ground wires in transmission lines, logical discrimination is performed on each conductor instance to distinguish between conductors / ground wires and current-carrying wires, and to identify key components and connection endpoints associated with the conductors / ground wires; specifically: (31) To achieve the complete link judgment of "conductor-U-shaped hanging ring-tower head", that is, the conductor and ground wire are suspended and electrically grounded to the tower head through the U-shaped hanging ring. This module is for the conductor instance set S_wire Each wire instance in the array corresponds to a mask and a U-shaped loop instance. S_clamp Search for the existence of a U-shaped hanging loop instance. clamp_k This makes the wire wire_i andclamp_k Adjacent or overlapping images in image space are determined using Boolean functions:

[0018] in, M_wi This is the mask for the conductor portion. M_ck This is the mask for the U-shaped hanging loop section. F_connect Determine if there is any overlap between the two; (32) If the judgment regarding the wire and the U-shaped loop is correct, then further examine the U-shaped loop. clamp_k Gather with the tower head S_tower Any tower head instance in tower_l Whether spatial relationships exist can be determined using a formula:

[0019] in, M_ck For the U-shaped hanging ring, M_tl This is the mask for the tower head; if the determination result is True, then this U-shaped hanging ring is connected to the tower head. (33) Finally, for a certain conductor wire_i If it exists clamp_k and tower_l , making =true Then determine this wire wire_i Let the conductor be a ground wire, with the head of the conductor being the endpoint region where the conductor is connected to the U-shaped loop, and the tail being the other end of the conductor; the set of associated components of the ground wire is { wire_i , clamp_k , tower_l}; (34) The drone's shooting angle may cause the U-shaped hanging ring to be obscured or invisible. For wires that do not meet the requirements for connection with the U-shaped hanging ring, the module performs a judgment on the tower head, checks whether the wire is associated with any tower head, and determines the association function. Is it True? If the output is True, then this conductor is determined to be a ground wire. Due to the lack of intermediate hardware, the end of this ground wire is denoted as the endpoint area directly connected to the tower head; its associated component set is { wire_j , tower_m},in wire_j This represents the j-th wire instance identified from the infrared image. tower_m This represents an instance of the m-th tower head identified from the same infrared image; (35) Finally, this module outputs the corresponding wires in the image, and the output wires are represented by a set of tuples:

[0020] in, Wire_i For grounding wire; Clamp_i This is the associated U-shaped loop and mask; if it does not exist, it is None. Tower_i This refers to the associated tower head instance and mask information; Head_i The coordinates of the wire end where the conductor connects to the tower. n This indicates the number of conductors identified as ground wires in the infrared image.

[0021] Preferably, in step (4), the conductor heating detection module receives the output result from the conductor judgment module: that is, each infrared image contains a list of tuples that are judged to be conductors and their associated components, and performs refined temperature analysis and comprehensive logical judgment on each conductor instance in the list, and finally determines whether it belongs to the abnormal heating defect of the hardware, and accurately locates the heating area. The process is as follows: (41) Input and initialization of conductor and ground wire tuple data. Each item in the conductor and ground wire element represents a conductor and ground wire to be determined. The module obtains the temperature data matrix corresponding to the current temperature image from the infrared image acquisition stage. T_original Each pixel's value represents the absolute temperature of that point; (42) For each conductor element G_i The module performs independent temperature analysis and heat point detection on each of its constituent wires, U-shaped hanging rings, and tower head components. The module traverses the temperature sets of each component, records all pixels that exceed the threshold, and forms a preliminary set of heat points. (43) Based on the comprehensive heat generation determination of spatial topology and temperature correlation, after completing the initial screening of component heat generation, the module determines the heat generation of conductor and ground wire instances based on the thermal conductivity characteristics of conductor and ground wire defects. G_i The module comprehensively analyzes the heat dissipation information and space of all related components, and then uses the ground wire example to make a judgment. G_i The included components execute corresponding judgment logic and finally output a list of overheating defects in the ground wire.

[0022] Preferably, in step (42), the specific steps are as follows: (421) First, extract the component temperature data. Based on the mask matrix of each component, extract the temperature data from the infrared image temperature matrix. T_original Extract the temperature value of the corresponding pixel to form the temperature information of the component; for wire components... wire_i Its temperature dataset T_w Similarly, the temperature set of the U-shaped hanging ring is obtained. T_c Temperature collection at the tower head T_t ; (422) Next, threshold comparison and heat point recording are performed. The module compares the temperature value of each component with its preset heat judgment threshold. The module traverses the temperature set of each component and records all pixels that exceed the threshold to form a preliminary heat point set, as follows: For the wire component, the set of its heat points is as follows: H_w = (corresponding pixel), its heat-generating marker is F_w ; For the U-shaped hanging ring component, its set of heat points is as follows: H_c = (corresponding pixel), its heat-generating marker is F_c ; For the tower head component, its set of heat-generating points is as follows: H_t = (corresponding pixel), its heat-generating marker is F_t .

[0023] Preferably, in step (43), the module determines the ground wire instance. G_i The components included execute different decision-making logic as follows: (a) If the conductor / ground wire includes a conductor, a U-shaped hanging ring, and a tower head, and all of these components generate heat, the heat source is: F_w , F_c , F_t And it meets the following conditions:

[0024] Determine the ground wire G_i It generates heat, and the heat-generating area is a combination of the heat-generating areas of the three components; (b) If the conductor and ground wire include the conductor, U-shaped ring, and tower head, and only the conductor and U-shaped ring are heated, the heating point is... F_ w , F_c And it meets the following conditions:

[0025] Determine the ground wire G_i It generates heat, and the area of ​​heat is the connection between the wire and the U-shaped hanging ring; (c) If the conductor and ground wire include the conductor, U-shaped hanging ring, and tower head, and only the conductor, tower head, and tower head are heated, the heating point is: F_ w , F_t And it meets the following conditions:

[0026] Determine the ground wire [[ID=6 The conductor is overheating, indicating that it is heating up. The area of ​​overheating is... ​ ∪ ​In this case, although the U-shaped hanging ring was not directly detected as a hot spot, the wire and the tower head were heated at the connection point at the same time and were very close, so it can be inferred that the fault exists in the connection circuit. The U-shaped hanging ring was actually heated, but it may not have reached the threshold on the image due to heat diffusion. (d) If the conductor and ground wire include the conductor and tower head, and the conductor and U-shaped hanging ring are heating up, the heating point is... ​ , ​ And it meets the following conditions:

[0027] Determine the ground wire ​ Fever, fever area is ​ ∪ ​ This corresponds to the situation where the U-shaped hanging ring is blocked, but the connection between the conductor and the tower head is hot.

[0028] The beneficial effects of this invention are: it can integrate infrared temperature information and image spatial structure information to accurately distinguish between ground wires and drain wires, and perform integrated positioning and correlation analysis of abnormal heating of ground wires and their associated tower damage fittings, thereby achieving a more accurate and reliable diagnosis of the heating defects of ground wires and their root causes. Attached Figure Description

[0029] ​ This is a flowchart of the method of the present invention; ​ This is the overall architecture diagram of the present invention. Detailed Implementation

[0030] The present invention will now be further described with reference to the accompanying drawings and specific embodiments.

[0031] like ​ , ​ In the described embodiment, the transmission line conductor and ground wire heating detection method based on infrared image spatial positioning includes an infrared image acquisition module, a conductor and related component extraction module, a conductor and ground wire judgment module, and a conductor and ground wire heating detection module. The conductor and related component extraction module includes a conductor extraction module, a U-shaped hanging ring extraction module, and a tower head extraction module. The specific operation steps are as follows: (1) The infrared image acquisition module is responsible for acquiring raw data. This module automatically reads the infrared thermal image to be analyzed uploaded by the UAV inspection system from the designated cloud storage service. After reading, it performs basic format parsing and caching on the image to provide standardized input data for subsequent processing modules.

[0032] The infrared image acquisition module automates the reception and identification management of infrared images transmitted from the UAV inspection system. This module interfaces with the enterprise-level object storage service OBS, continuously monitoring and acquiring infrared image data files uploaded by the UAV inspection platform. Each infrared image is stored with its file identifier (MFID) as a unique filename. Upon receiving an infrared image file, the module directly parses the MFID from its filename. This MFID serves as the core identifier for the image throughout the system and is persisted as a key index record in the metadata.

[0033] (2) The conductor and related component extraction module, based on the conductor extraction module, U-shaped hanging ring extraction module and tower head extraction module, uses the trained model algorithm to segment and detect the conductor component set from the acquired infrared image. ​ U-shaped hanging ring component assembly ​ and tower head component assembly ​ .

[0034] The conductor and related component extraction module is the core processing unit connecting infrared image acquisition with subsequent conductor / ground wire breakage and heat analysis. This module identifies and segments key components closely related to conductor / ground wire determination from a single input infrared thermal image, providing accurate image region and location information for subsequent spatial logic judgment. The module includes a conductor extraction module, a U-shaped hanging ring extraction module, and a tower head extraction module. It receives standardized infrared images from the infrared image acquisition module and, through a computer vision model, simultaneously detects and segments three types of targets: conductors, U-shaped hanging rings, and tower heads. Its functionality is implemented by its three sub-functional modules.

[0035] The wire extraction module identifies and extracts wire targets from complex backgrounds. It is based on a pre-trained deep learning target detection and segmentation model, trained using a large dataset of labeled infrared images of wire regions. For an input infrared image, the model performs wire target detection and pixel-level segmentation, outputting a set of all wire instances in the image. The segmented set of wire instances is as follows:

[0036] in, Representing the j The bounding box coordinates of a wire instance. ​ and ​ The x and y coordinates of the top-left vertex of the bounding box are defined. ​ and ​ The x and y coordinates of the lower right corner vertex of the bounding box are defined; ​ This represents its corresponding binary segmentation mask matrix. ​ This indicates the total number of wire instances.

[0037] The U-shaped hanging ring extraction module identifies key connecting suspension components of the ground wire from complex backgrounds. It is also based on a specially trained target detection and segmentation model. The model detects and segments all U-shaped hanging ring instances in the input infrared image, and the resulting segmentation set is as follows:

[0038] in, ​ and ​ Representing the first k Boundary box and segmentation mask of a U-shaped hanging loop. ​ This represents the total number of U-shaped hanging ring instances detected.

[0039] The tower head extraction module identifies and segments the tower heads in infrared images, providing a spatial reference coordinate system for the entire invention. The training set of the target detection and segmentation model includes infrared image annotations of four tower types: drum-shaped tower heads, cat-head tower heads, ram-horn tower heads, and zig-shaped tower heads. After image processing by this model, the output set of tower head instances is as follows:

[0040] in, ​ and ​ Representing the first l The bounding box and segmentation mask of each tower head instance. ​ This represents the number of tower head instances detected.

[0041] This module ultimately outputs a set of three target instances. ​ , ​ and ​ An organic combination of bounding boxes and pixel regions.

[0042] (3) The conductor and ground wire judgment module determines whether the conductor component, U-shaped hanging ring component and tower head component belong to the conductor and ground wire in the infrared image by comprehensively considering their spatial positions in the infrared image.

[0043] The conductor / ground wire determination module receives a set of conductor instances from the conductor and related component extraction module. ​ Collection of U-shaped hanging ring examples ​ With the set of tower head instances ​ Based on the unique spatial topology of conductors and ground wires in transmission lines, logical judgment is performed on each conductor instance to accurately distinguish between conductors / ground wires and jumper wires (junction wires), and to identify key components and connection endpoints associated with the conductors / ground wires. Specifically: (31) To achieve complete link determination of "conductor-U-shaped hanging ring-tower head", that is, the conductor and ground wire are usually suspended and electrically grounded to the tower head through a U-shaped hanging ring. Therefore, this module is for the conductor instance set ​Each wire instance in the array corresponds to a mask and a U-shaped loop instance. ​ Search for the existence of a U-shaped hanging loop instance. ​ This makes the wire ​ and ​ Adjacent or overlapping in image space (i.e., their masks) ​ and ​ If there is intersection or the boundaries are in contact, a Boolean function can be used to determine this:

[0044] in, ​ This is the mask for the conductor portion. ​ This is the mask for the U-shaped hanging loop section. ​ Determine if there is any overlap between the two.

[0045] (32) If the judgment regarding the wire and the U-shaped loop is correct, then further examine the U-shaped loop. ​ Gather with the tower head ​ Any tower head instance in ​ Whether spatial relationships exist can be determined using a formula:

[0046] in, ​ For the U-shaped hanging ring, ​ This is the mask for the tower head; if the result is True, then this U-shaped hanging ring is connected to the tower head.

[0047] (33) Finally, for a certain conductor ​ If it exists ​ and ​ , making =true Then determine this wire ​ Let the conductor be the ground wire, with the wire end defined as the region connecting the conductor to the U-shaped loop, and the wire tail as the other end of the conductor. The set of associated components of this ground wire is { ​ , ​ , ​}

[0048] (34) The drone's shooting angle may cause the U-shaped hanging ring to be obscured or invisible. For wires that do not meet the requirements for connection with the U-shaped hanging ring, the module performs a judgment on the tower head, checks whether the wire is associated with any tower head, and determines the association function. Is it True? If the output is True, then this conductor is determined to be a ground wire. Due to the lack of intermediate hardware, the end of this ground wire is denoted as the endpoint area directly connected to the tower head. Its associated component set is { ​ ,​},in ​ This represents the j-th wire instance identified from the infrared image. ​ This represents an instance of the m-th tower head identified from the same infrared image.

[0049] (35) Finally, this module outputs the corresponding wires in the image, and the output wires are represented by a set of tuples:

[0050] in, ​ For grounding wire; ​ This is the associated U-shaped loop and mask; if it does not exist, it is None. ​ This refers to the associated tower head instance and mask information; ​ The coordinates of the wire end where the conductor connects to the tower. n This indicates the number of conductors identified as ground wires in the infrared image.

[0051] (4) The ground wire heating detection module extracts the temperature of each component of the ground wire, extracts and judges the extreme temperature values, and makes a comprehensive judgment on the heating of the ground wire based on the spatial position relationship.

[0052] The conductor / ground wire heating detection module performs heating diagnosis and location. The module receives output from the conductor / ground wire judgment module: each infrared image contains a list of tuples identified as conductors / ground wires and their associated components. It then performs refined temperature analysis and comprehensive logical judgment on each conductor / ground wire instance in the list to ultimately determine whether it constitutes an abnormal heating defect in the hardware and accurately locates the heating area. The process is as follows: (41) Input and initialization of conductor and ground wire tuple data. Each item in the conductor and ground wire element represents a conductor and ground wire to be determined. The module obtains the temperature data matrix corresponding to the current temperature image from the infrared image acquisition stage. ​ The value of each pixel represents the absolute temperature of that point.

[0053] (42) For each conductor element ​ The module performs independent temperature analysis and heat point detection on each of its constituent components (wires, U-shaped hanging rings, and tower heads); the module traverses the temperature sets of each component, records all pixels exceeding the threshold, and forms a preliminary set of heat points. Specifically: (421) First, extract the component temperature data. Based on the mask matrix of each component (a value of 1 represents a component pixel), extract the temperature data from the infrared image temperature matrix. ​ Extract the temperature value of the corresponding pixel to form the temperature information of the component. For wire components... ​ Its temperature dataset ​ Similarly, the temperature set of the U-shaped hanging ring is obtained.​ Temperature collection at the tower head ​ .

[0054] (422) Next, threshold comparison and heat point recording are performed. The module compares the temperature value of each component with its preset heat judgment threshold. Based on engineering experience, the heat judgment threshold of the ground wire and its associated components is set to 80℃. The module traverses the temperature set of each component and records all pixels that exceed the threshold, forming a preliminary heat point set. The details are as follows: For the wire component, the set of its heat points is as follows: ​ = (corresponding pixel), its heat-generating marker is ​ ; For the U-shaped hanging ring component, its set of heat points is as follows: ​ = (corresponding pixel), its heat-generating marker is ​ ; For the tower head component, its set of heat-generating points is as follows: ​ = (corresponding pixel), its heat-generating marker is ​ .

[0055] (43) Based on the comprehensive heat generation determination of spatial topology and temperature correlation, after completing the initial screening of component heat generation, the module determines the heat generation of conductor and ground wire instances based on the thermal conductivity characteristics of conductor and ground wire defects. ​ The module comprehensively analyzes the heat dissipation information and space of all related components, and then uses the ground wire example to make a judgment. ​ The included components execute corresponding judgment logic and finally output a list of overheating defects in the ground wire.

[0056] Among them, the module is based on the ground wire instance. ​ The components included execute different decision-making logic as follows: (a) If the conductor / ground wire includes a conductor, a U-shaped hanging ring, and a tower head, and all of these components generate heat, the heat source is: ​ , ​ , ​ And it meets the following conditions:

[0057] Determine the ground wire ​ It generates heat, and the heat-generating area is a combination of the heat-generating areas of the three components.

[0058] (b) If the conductor and ground wire include the conductor, U-shaped ring, and tower head, and only the conductor and U-shaped ring are heated, the heating point is... F_ w , ​ And it meets the following conditions:

[0059] Determine the ground wire ​It generates heat, and the area of ​​heat is the connection between the wire and the U-shaped hanging ring.

[0060] (c) If the conductor and ground wire include the conductor, U-shaped hanging ring, and tower head, and only the conductor, tower head, and tower head are heated, the heating point is: F_ w , ​ And it meets the following conditions:

[0061] Determine the ground wire ​ The conductor is overheating, indicating that it is heating up. The area of ​​overheating is... ​ ∪ ​ In this case, although the U-shaped hanging ring was not directly detected as a hot spot, the wire and the tower head were both heated at the connection point and were very close together. It can be inferred that the fault exists in this connection circuit. The U-shaped hanging ring was actually heated, but the heat may not have reached the threshold on the image due to heat diffusion.

[0062] (d) If the conductor and ground wire include the conductor and tower head, and the conductor and U-shaped hanging ring are heating up, the heating point is... ​ , ​ And it meets the following conditions:

[0063] Determine the ground wire ​ Fever, fever area is ​ ∪ ​ This corresponds to the situation where the U-shaped hanging ring is blocked, but the connection between the conductor and the tower head is hot.

[0064] Specific implementation examples are as follows: Step S1: Infrared image acquisition and data preprocessing.

[0065] The system interfaces with Enterprise Object Storage Service (OBS) to continuously monitor and automatically acquire infrared thermal image files uploaded by the drone inspection platform. Each image file is named with its unique File Identifier (MFID). Upon receiving a file, the acquisition module parses the filename to obtain the MFID, which is then persistently stored as the core identifier for the image throughout the system. Subsequently, the module performs basic format parsing, decoding, and standardization caching of the image to ensure a uniform input data format for all subsequent processing modules. Simultaneously, it reads and caches the absolute temperature matrix corresponding to each pixel from the image file or associated metadata. ​ .

[0066] Step S2: Extraction of wires and related components This step corresponds to the "Conductor and Related Components Extraction Module," which aims to accurately segment the conductor, U-shaped hanging ring, and tower head from the infrared image.

[0067] Model Loading and Inference: The system loads three pre-trained deep learning instance segmentation models, used for detecting conductors, U-shaped hanging rings, and tower heads, respectively. The tower head detection model covers common tower types such as drum-shaped towers, cat-head towers, ram-horn towers, and zig-shaped towers. The pre-processed infrared image I is simultaneously input into these three models for parallel inference.

[0068] Results Analysis and Structured Output: The output of each model is analyzed to obtain all detected instances in the image. For each instance, its class label, bounding box B, and binary segmentation mask M are recorded. The set of pixels with a value of 1 in mask M precisely corresponds to the contour region of that part in the image. Finally, this step outputs three structured sets of instances: Collection of wire instances: ​ Collection of U-shaped hanging loop examples: ​ Tower head instance collection: ​ Step S3: Ground wire identification and associated component identification This step corresponds to the "Conductor / Ground Wire Judgment Module," which distinguishes conductors and ground wires from wires based on spatial topological relationships.

[0069] Spatial correlation determination: traversal ​ Each wire instance in ​ Define a spatial correlation function. ​ It calculates two masks ​ and M_b The distance between corresponding pixel sets (e.g., whether the minimum Euclidean distance is less than a threshold) δ Alternatively, you can determine whether two instances are connected in the image space by checking if their boundaries touch.

[0070] Main logic judgment (conductor-U-shaped hanging ring-tower head): For wire_i ,exist S_clamp Search for instances of U-shaped hanging loops. clamp_k , making F_connect(M_wi, M_ck) = True If found, then further investigation is needed. S_tower Search for the existence of tower head instances. tower_l , making F_connect(M_ck, M_tl) = True If both conditions are met, then a decision is made. wire_i For a ground wire, its associated components are as follows: {wire_i, clamp_k, tower_l} And mark the connection end between the wire and the U-shaped hanging ring as the wire end.

[0071] Alternative logic judgment (wire-tower head): For those that do not satisfy the main logic... wire_i Directly check if it exists tower_m Make F_connect(M_wi, M_tm) = True If satisfied, then determine... wire_i For a ground wire, its associated components are as follows: {wire_i, tower_m} And mark the connection end between the conductor and the tower head as the wire end.

[0072] Result Encapsulation: All determined conductors and their associated information are encapsulated into a list of tuples. For the nth conductor, its tuple is... G_n = (Wire_n, Clamp_n, Tower_n, Head_n) ,in Clamp_n The value is None when the U-shaped loop is not visible.

[0073] Step S4: Comprehensive detection and location of conductor heating.

[0074] This step corresponds to the "Conductor and Ground Wire Heating Detection Module," which performs detailed temperature analysis and defect determination for each conductor and ground wire.

[0075] Component-level temperature extraction and initial screening: for each conductor / ground element G_i : Temperature extraction: using component masks respectively M_w , M_c (if it exists) M_t From the temperature matrix T_original Extract the temperature value of the corresponding pixel to obtain the temperature set. T_w, T_c, T_t .

[0076] Heating point detection: Set a heating threshold (e.g., for wires) θ_w=80°C Identify the pixels in the temperature set of each component that exceed the threshold, forming a set of heat-generating points. H_w, H_c, H_t and obtain fever markers F_w = (|H_w|>0), F_c = (|H_c|>0), F_t = (|H_t|>0) .

[0077] Comprehensive heat generation determination based on spatial topology: based on tuples G_n The structure of the system executes different decision logic trees. A spatial proximity function is defined. Near(H_x, H_y, d) Determine whether there exists a pair of points in two hot spot clusters whose distance is less than a threshold. d .

[0078] Situation A (including U-shaped hanging ring): like F_w ∧ F_c ∧ F_t True, and H_w The point in is located Head_n The area, and meets the requirements Near(H_ w, H_c, d1) and Near(H_c, H_t, d2) If so, it is determined to be a defect. The heating area is... H_w ∪ H_c ∪ H_t .

[0079] like F_w ∧ F_c True, and H_w The point in is located Head_n The area, and meets the requirements Near(H_w, H_c, d1) If so, it is determined to be a defect. The heating area is... H_w ∪ H_c .

[0080] like F_w ∧ F_t True, and H_w The point in is located Head_n The area, and meets the requirements Near(H_w, H_t, d3) If d3 is approximately equal to the size of the U-shaped hanging ring, it is considered a defect. The heating area is... H_w ∪ H_t .

[0081] Situation B (without U-shaped hanging loop): like F_w ∧ F_t True, and H_w The point in is located Head_n The area, and meets the requirements Near(H_w, H_t, d4) If so, it is determined to be a defect. The heating area is... H_w ∪ H_t .

[0082] If a conductor does not meet any of the above conditions, it is determined that no related abnormal heating has occurred.

[0083] Result Generation: All ground wire information identified as overheating is aggregated to generate a list of overheating defects. Each record in the list includes: ground wire identifier, associated overheating component, set of pixel coordinates of the overheating area, highest temperature value, and location.

[0084] Through the above four steps, this invention achieves fully automated detection, from automatic access to UAV infrared images, to precise segmentation of key components, to intelligent identification of conductors and ground wires based on spatial topology, and finally to comprehensive judgment of multi-component related heating. This method effectively solves the problem of confusion between conductors and ground wires, and correlates conductor and ground wire heating with its root cause hardware defects, significantly improving the accuracy of detection and the targeted nature of operation and maintenance.

Claims

1. A method for detecting heating of a ground wire of a power transmission line based on spatial positioning of infrared images, characterized in that, It consists of an infrared image acquisition module, a conductor and related component extraction module, a conductor and ground wire identification module, and a conductor and ground wire heating detection module. The conductor and related component extraction module includes a conductor extraction module, a U-shaped hanging ring extraction module, and a tower head extraction module. The specific operation steps are as follows: (1) The infrared image acquisition module is responsible for acquiring raw data. This module automatically reads the infrared thermal image to be analyzed uploaded by the UAV inspection system from the designated cloud storage service. After reading, it performs basic format parsing and caching on the image to provide standardized input data for subsequent processing modules. (2) The wire and its related components extraction module extracts the wire component set from the acquired infrared image based on the wire extraction module, the U-shaped hanger ring extraction module, and the tower head extraction module according to the trained model algorithm S_wire , the U-shaped hanger ring component set S_clamp , and the tower head component set S_tower ; (3) The conductor and ground wire judgment module integrates the spatial positions of the conductor component, U-shaped hanging ring component and tower head component in the infrared image to determine whether they belong to the conductor and ground wire, thereby obtaining the conductor and ground wire in the infrared image; (4) The ground wire heating detection module extracts the temperature of each component of the ground wire, extracts and judges the extreme temperature values, and makes a comprehensive judgment on the heating of the ground wire based on the spatial position relationship.

2. The method of claim 1, wherein the step of detecting the heat of the ground wire of the power transmission line based on the infrared image space positioning is characterized by In step (1), the infrared image acquisition module realizes the automated reception and identification management of infrared images transmitted back by the UAV inspection system. This module connects with the enterprise-level object storage service OBS through the interface to continuously monitor and acquire infrared image data files uploaded by the UAV inspection platform. Each infrared image is stored with its file identifier MFID as a unique filename. After receiving the infrared image file, the module directly parses the MFID from its filename. This MFID serves as the core identifier of the image in the entire system and is persisted to the metadata as a key index record.

3. The method for detecting heating of transmission line conductors and ground wires based on infrared image spatial positioning according to claim 1, characterized in that, in In step (2), the wire extraction module identifies and extracts wire targets from a complex background. It is based on a pre-trained deep learning target detection and segmentation model, trained using an infrared image dataset labeled with wire regions. For the input infrared image, the model performs wire target detection and pixel-level segmentation, outputting a set of all wire instances in the image. The segmented set of wire instances is as follows: in, Representing the j The bounding box coordinates of a wire instance. x_min and y_min The x and y coordinates of the top-left vertex of the bounding box are defined. x_max and y_max The x and y coordinates of the lower right corner vertex of the bounding box are defined; M_wj This represents its corresponding binary segmentation mask matrix. N_wire This indicates the total number of wire instances.

4. The method for detecting heating of transmission line conductors and ground wires based on infrared image spatial positioning according to claim 1, characterized in that, in In step (2), the U-shaped hanging ring extraction module identifies the key connection and suspension components of the ground wire from the complex background. Based on a specially trained target detection and segmentation model, the model detects and segments all U-shaped hanging ring instances in the input infrared image, and the resulting set of segmentation results is as follows: in, B_ck and M_ck Representing the first k Boundary box and segmentation mask of a U-shaped hanging loop. N_clamp This represents the total number of U-shaped hanging ring instances detected.

5. The method for detecting heating of transmission line conductors and ground wires based on infrared image spatial positioning according to claim 1, characterized in that, In step (2), the tower head extraction module identifies and segments the tower head portion in the infrared image. The training set of the target detection and segmentation model includes infrared image annotations of four tower types: drum-shaped tower head, cat-head tower head, ram-horn tower head, and zig-shaped tower head. After processing the image by this model, the output set of tower head instances is as follows: in, B_tl and M_tl Representing the first l The bounding box and segmentation mask of each tower head instance. N_tower This represents the number of tower head instances detected.

6. The method for detecting heating of transmission line conductors and ground wires based on infrared image spatial positioning according to claim 1, characterized in that, in In step (3), the conductor / ground wire determination module receives a set of conductor instances from the conductor and related component extraction module. S_wire Collection of U-shaped hanging ring examples S_clamp With the set of tower head instances S_tower Based on the unique spatial topology of conductors and ground wires in transmission lines, logical discrimination is performed on each conductor instance to distinguish between conductors / ground wires and current-carrying wires, and to identify key components and connection endpoints associated with the conductors / ground wires; specifically: (31) To achieve the complete link judgment of "conductor-U-shaped hanging ring-tower head", that is, the conductor and ground wire are suspended and electrically grounded to the tower head through the U-shaped hanging ring. This module is for the conductor instance set S_wire Each wire instance in the array corresponds to a mask and a U-shaped loop instance. S_clamp Search for the existence of a U-shaped hanging loop instance. clamp_k This makes the wire wire_i and clamp_k Adjacent or overlapping images in image space are determined using Boolean functions: in, M_wi This is the mask for the conductor portion. M_ck This is the mask for the U-shaped hanging loop section. F_connect Determine if there is any overlap between the two; (32) If the judgment regarding the wire and the U-shaped loop is correct, then further examine the U-shaped loop. clamp_k Gather with the tower head S_ tower Any tower head instance in tower_l Whether spatial relationships exist can be determined using a formula: in, M_ck For the U-shaped hanging ring, M_tl This is the mask for the tower head; if the determination result is True, then this U-shaped hanging ring is connected to the tower head. (33) Finally, for a certain conductor wire_i If it exists clamp_k and tower_l , making Then determine this wire wire_i Let the conductor be a ground wire, with the head of the conductor being the endpoint region where the conductor is connected to the U-shaped loop, and the tail being the other end of the conductor; the set of associated components of the ground wire is { wire_i , clamp_k , tower_l }; (34) The drone's shooting angle may cause the U-shaped hanging ring to be obscured or invisible. For wires that do not meet the requirements for connection with the U-shaped hanging ring, the module performs a judgment on the tower head, checks whether the wire is associated with any tower head, and determines the association function. Is it True? If the output is True, then this conductor is determined to be a ground wire. Due to the lack of intermediate hardware, the end of this ground wire is denoted as the endpoint area directly connected to the tower head; its associated component set is { wire_j , tower_m },in wire_j This represents the j-th wire instance identified from the infrared image. tower_m This represents an instance of the m-th tower head identified from the same infrared image; (35) Finally, this module outputs the corresponding wires in the image, and the output wires are represented by a set of tuples: in, Wire_i For grounding wire; Clamp_i This is the associated U-shaped loop and mask; if it does not exist, it is None. Tower_i This refers to the associated tower head instance and mask information; Head_i The coordinates of the wire end where the conductor connects to the tower. n This indicates the number of conductors identified as ground wires in the infrared image.

7. The method for detecting heating of transmission line conductors and ground wires based on infrared image spatial positioning according to claim 1, characterized in that, in In step (4), the ground wire heating detection module receives the output result from the ground wire judgment module: that is, each infrared image contains a list of tuples that are judged to be ground wires and their associated components, and performs fine temperature analysis and comprehensive logical judgment on each ground wire instance in the list, and finally determines whether it belongs to the abnormal heating defect of the hardware, and accurately locates the heating area. The process is as follows: (41) Input and initialization of conductor and ground wire tuple data. Each item in the conductor and ground wire element represents a conductor and ground wire to be determined. The module obtains the temperature data matrix corresponding to the current temperature image from the infrared image acquisition stage. T_original Each pixel's value represents the absolute temperature of that point; (42) For each conductor element G_i The module performs independent temperature analysis and heat point detection on each of its constituent wires, U-shaped hanging rings, and tower head components. The module traverses the temperature sets of each component, records all pixels that exceed the threshold, and forms a preliminary set of heat points. (43) Based on the comprehensive heat generation determination of spatial topology and temperature correlation, after completing the initial screening of component heat generation, the module determines the heat generation of conductor and ground wire instances based on the thermal conductivity characteristics of conductor and ground wire defects. G_i The module comprehensively analyzes the heat dissipation information and space of all related components, and then uses the ground wire example to make a judgment. G_i The included components execute corresponding judgment logic and finally output a list of overheating defects in the ground wire.

8. The method for detecting heating of transmission line conductors and ground wires based on infrared image spatial positioning according to claim 7, characterized in that, In step (42), specifically: (421) First, extract the component temperature data. Based on the mask matrix of each component, extract the temperature data from the infrared image temperature matrix. T_original Extract the temperature value of the corresponding pixel to form the temperature information of the component; for wire components... wire_i Its temperature dataset T_w Similarly, the temperature set of the U-shaped hanging ring is obtained. T_c Temperature collection at the tower head T_t ; (422) Next, threshold comparison and heat point recording are performed. The module compares the temperature value of each component with its preset heat judgment threshold. The module traverses the temperature set of each component and records all pixels that exceed the threshold to form a preliminary heat point set, as follows: For the wire component, the set of its heat points is as follows: H_w = (corresponding pixel), its heat-generating marker is F_w ; For the U-shaped hanging ring component, its set of heat points is as follows: H_c = (corresponding pixel), its heat-generating marker is F_c ; For the tower head component, its set of heat-generating points is as follows: H_t = (corresponding pixel), its heat-generating marker is F_t .

9. The method for detecting heating of transmission line conductors and ground wires based on infrared image spatial positioning according to claim 8, characterized in that, in In step (43), the module determines the ground wire instance. G_i The components included execute different decision-making logic as follows: (a) If the conductor / ground wire includes a conductor, a U-shaped hanging ring, and a tower head, and all of these components generate heat, the heat source is: F_w , F_c , F_t And it meets the following conditions: Determine the ground wire G_i It generates heat, and the heat-generating area is a combination of the heat-generating areas of the three components; (b) If the conductor and ground wire include the conductor, U-shaped ring, and tower head, and only the conductor and U-shaped ring are heated, the heating point is... F_w , F_c And it meets the following conditions: Determine the ground wire G_i It generates heat, and the area of ​​heat is the connection between the wire and the U-shaped hanging ring; (c) If the conductor and ground wire include the conductor, U-shaped hanging ring, and tower head, and only the conductor, tower head, and tower head are heated, the heating point is: F_w , F_ t And it meets the following conditions: Determine the ground wire G_i The conductor is overheating, indicating that it is heating up. The area of ​​overheating is... H_w ∪ H_t In this case, although the U-shaped hanging ring was not directly detected as a hot spot, the wire and the tower head were heated at the connection point at the same time and were very close, so it can be inferred that the fault exists in the connection circuit. The U-shaped hanging ring was actually heated, but it may not have reached the threshold on the image due to heat diffusion. (d) If the conductor and ground wire include the conductor and tower head, and the conductor and U-shaped hanging ring are heating up, the heating point is... F_w , F_t And it meets the following conditions: Determine the ground wire G_i Fever, fever area is H_w ∪ H_t This corresponds to the situation where the U-shaped hanging ring is blocked, but the connection between the conductor and the tower head is hot.

Images