X-ray nondestructive testing method for crimping quality of strain clamp of power transmission line

Abstract

The invention relates to the technical field of nondestructive testing of power equipment, and discloses an X-ray nondestructive testing method for crimping quality of a strain clamp of a power transmission line. The method comprises the following steps: fusing a topological structure, operation and maintenance records and meteorological data to generate an intelligent detection task list; the pose of the X-ray device is automatically calibrated, and multi-angle transmission imaging is carried out; performing standardization processing on the image, and inputting the image into a pre-trained multi-scale convolutional neural network to realize pixel-level defect segmentation; and calculating a crimping quality comprehensive evaluation index in combination with the three-dimensional defect depth and the service degradation factor, and generating a graded early warning signal to be pushed to an operation and maintenance platform. According to the method, through full-process closed-loop management, the detection precision, efficiency and response speed are remarkably improved, and safe and reliable operation of the power transmission line is guaranteed.

Description

Technical Field

[0001] This invention belongs to the field of non-destructive testing technology for power equipment, specifically relating to an X-ray non-destructive testing method for the crimping quality of tension clamps in transmission lines. Background Technology

[0002] With the continuous expansion of the power grid and the increasing standards for transmission line operation and maintenance, higher requirements are being placed on the crimping quality inspection of key fittings such as tension clamps. X-ray non-destructive testing technology, due to its ability to penetrate metal structures and visually reveal internal defects (such as insufficient crimping, cracks, and porosity), has become an important means of ensuring the mechanical strength and electrical performance of power lines. However, current testing practices often focus on the technical implementation of single aspects, lacking a systematic integration of the entire testing process, resulting in limitations in testing efficiency, data value extraction capabilities, and decision support levels.

[0003] X-ray non-destructive testing of tension clamps in transmission lines involves multiple stages, including test plan development, on-site operation execution, image data analysis, and maintenance decision support. Ideally, each stage should form an information loop, achieving dynamic feedback and collaborative optimization. However, in current practices, test plans are often independently prepared based on static ledgers, failing to fully integrate the three-dimensional spatial information of the line and historical defect data; on-site operations rely on manual experience to operate equipment, making it difficult to standardize shooting angles and exposure parameters, and the operation process is disconnected from the back-end system; although the acquired X-ray images can identify local defects, they are stored in isolation, failing to correlate and analyze with line topology, service environment, and previous test results; the final evaluation conclusions are mostly qualitative judgments, lacking quantitative prediction of the evolution trend of crimping quality, making it difficult to provide a reliable basis for maintenance priority ranking or life assessment.

[0004] Existing technologies exhibit significant disconnects between the aforementioned stages: inspection task planning is decoupled from the physical line status, on-site data collection lacks contextual semantic support, analysis results cannot feed back into subsequent inspection strategy optimization, and the decision-making process lacks an intelligent reasoning foundation based on multi-source data fusion. This "information silo" phenomenon not only reduces the efficiency of inspection resource utilization but also hinders the transformation from a "passive defect detection" to a "proactive risk prediction" operation and maintenance model. Especially in high-reliability scenarios such as ultra-high voltage and inter-regional interconnection, there is an urgent need for a new inspection method that can connect the entire inspection chain and achieve data-driven closed-loop optimization to improve the accuracy, foresight, and intelligence level of crimping quality assessment. Summary of the Invention

[0005] This invention provides an X-ray non-destructive testing method for the crimping quality of tension clamps in transmission lines. It aims to solve systemic technical problems in existing technologies, such as low testing efficiency, poor data consistency, insufficient defect identification accuracy, and delayed maintenance response, caused by the isolation of various stages in testing plan development, on-site operation, data analysis, and decision support. By constructing a closed-loop testing system integrating intelligent planning of testing tasks, standardized execution of on-site operations, automated image data analysis, and dynamic assessment of defect risks, it achieves full-process, high-precision, and traceable non-destructive testing of the crimping quality of tension clamps.

[0006] The X-ray non-destructive testing method for the crimping quality of tension clamps in transmission lines according to the present invention includes the following steps: acquiring the topology information, historical operation and maintenance records, and meteorological environmental data of the transmission line to be inspected; and generating a test task list based on the topology information and historical operation and maintenance records, including priority sorting, test point coordinates, equipment configuration parameters, and safe operation windows.

[0007] According to the inspection task list, the X-ray inspection device is dispatched to the designated tower location, and the relative pose of the X-ray source and detector is automatically calibrated. Within the preset safe operating window, the X-ray inspection device is controlled to perform multi-angle transmission imaging on the target tension clamp, generating a raw X-ray image sequence. The raw X-ray image sequence is subjected to radiation dose normalization, geometric distortion correction, and background noise suppression to obtain a standardized image dataset. The standardized image dataset is input into a pre-trained crimping defect recognition model, which outputs an internal structural feature map of the crimping area and a defect confidence score. Based on the internal structural feature map and defect confidence score, combined with the material specifications of the tension clamp, crimping process standards, and service life, a comprehensive crimping quality evaluation index is calculated. A graded early warning signal is generated based on the comprehensive crimping quality evaluation index and pushed to the power grid operation and maintenance decision platform to trigger corresponding maintenance strategies.

[0008] Furthermore, the acquisition of the topology information, historical operation and maintenance records, and meteorological environmental data of the transmission line to be inspected specifically includes: extracting the tower number, span length, conductor type, and tension clamp installation location of the transmission line from the power grid geographic information system; retrieving the factory batch, crimping construction records, previous test results, and fault repair logs of each tension clamp from the equipment asset management system; and obtaining wind speed, humidity, rainfall probability, and atmospheric visibility data for the testing area within the next 72 hours from the meteorological monitoring network.

[0009] Furthermore, the generation of the inspection task list, which includes priority ranking, inspection point coordinates, equipment configuration parameters, and safe operation windows, specifically includes: determining the priority of each inspection point using a weighted scoring method based on the service life of the tension clamp, the frequency of historical defects, and the risk level of lightning strikes or icing in the section; calculating the optimal deployment location and X-ray penetration path of the X-ray inspection device based on the three-dimensional coordinates of the tower and the suspension height of the conductor; setting the X-ray tube voltage, tube current, exposure time, and detector gain parameters based on the conductor diameter, clamp material, and ambient temperature and humidity; and defining the safe operation time period each day by combining the wind speed threshold and visibility limitations in the meteorological data.

[0010] Furthermore, the automatic calibration of the relative pose of the X-ray source and the detector specifically includes: measuring the distance from the focal point of the X-ray source to the center of the tension clamp using a laser ranging module, and adjusting it to a preset focal length; determining the incident direction of the X-ray beam using an tilt sensor and an electronic compass, making it perpendicular to the pressing axis; and driving the detector plane to orthogonally align with the X-ray beam using a servo gimbal, ensuring that the imaging geometry meets the parallel projection conditions.

[0011] Furthermore, the control of the X-ray detection device for multi-angle transmission imaging of the target tension clamp specifically includes: rotating the X-ray source and detector assembly at 30-degree intervals around the pressing axis, acquiring a total of 12 frames of X-ray images from different perspectives; the exposure time of each frame is not less than 0.5 seconds, and the spatial resolution is not less than 50 micrometers; all images are accompanied by timestamps, device status codes, and environmental parameter metadata.

[0012] Furthermore, the radiation dose normalization, geometric distortion correction, and background noise suppression processing of the original X-ray image sequence specifically includes: using the response function of a flat panel detector to perform dose compensation on the grayscale value of each pixel; applying a back-projection geometric model to eliminate barrel distortion based on the focal position of the X-ray source and the pixel coordinates of the detector; and using a wavelet threshold denoising algorithm to filter out high-frequency random noise while preserving the edge details of the pressing interface.

[0013] Furthermore, the pre-trained crimping defect recognition model is a multi-scale feature fusion convolutional neural network. Its input layer receives a standardized image dataset, extracts local texture and global morphological features through four downsampling encoding stages, and then fuses low-level edge information with high-level semantic information through skip connections. Finally, a pixel-level defect segmentation mask is generated in the output layer. The types of defects include insufficient crimping, over-crimping, cracks, pores, and foreign matter inclusions. The model training adopts joint optimization of cross-entropy loss function and Dice coefficient, and the training dataset contains no less than 15,000 sets of labeled samples.

[0014] Furthermore, the calculation of the comprehensive evaluation index for crimping quality specifically includes: defining the comprehensive evaluation index for crimping quality. It is a weighted sum of the defect area ratio, defect depth ratio, and service degradation coefficient; where the defect area ratio is the ratio of the area covered by the defect segmentation mask to the theoretical pressing area; the defect depth ratio is calculated by reconstructing the three-dimensional defect volume through the parallax information of the dual-view image; the service degradation coefficient is determined based on the service life of the tension clamp and the material fatigue curve, and the degradation coefficient of the clamp that has been in service for more than 10 years is not less than 1.2.

[0015] Furthermore, the generation of graded early warning signals and their push to the power grid operation and maintenance decision-making platform specifically includes: when the comprehensive evaluation index of crimping quality... When it is less than 0.3, it is judged as a normal state, and a green warning signal is generated; when When the value is between 0.3 and 0.6, it is considered a minor defect, generating a yellow warning signal, and a re-inspection within 3 months is recommended; when When the value is greater than 0.6, it is judged as a serious defect, a red warning signal is generated, and an emergency defect elimination work order is immediately triggered; all warning signals are accompanied by defect location coordinates, three-dimensional shape map and historical comparison trend map.

[0016] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0017] By integrating the detection plan formulation, on-site operation execution, image data analysis, and operation and maintenance decision support into a unified technical process, the data barriers and operational breakpoints between the various links in the traditional model are eliminated; the generation of the detection task list integrates equipment status, environmental constraints, and safety specifications to ensure efficient allocation of operational resources; the on-site operation process realizes automatic pose calibration and multi-angle imaging, which significantly improves the consistency and integrity of image acquisition.

[0018] A pre-trained multi-scale convolutional neural network is used to segment standardized images at the pixel level, avoiding subjective bias in manual interpretation and improving the defect recognition accuracy to over 98%. The comprehensive evaluation index for crimping quality incorporates three-dimensional defect depth and service degradation factors, making risk assessment more closely aligned with engineering realities. The hierarchical early warning mechanism is seamlessly integrated with the power grid operation and maintenance platform, achieving closed-loop management from detection to handling, significantly shortening the defect response cycle, and effectively ensuring the structural safety and operational reliability of transmission lines. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the overall technical solution architecture of the present invention;

[0020] Figure 2 This is a schematic diagram of the core principle framework of the crimping defect identification model in this invention;

[0021] Figure 3 This is a logical flowchart of the intelligent planning of detection tasks and the generation of safe operation windows in this invention;

[0022] Figure 4 This is a logical flowchart of the X-ray multi-angle transmission imaging and image standardization processing in this invention;

[0023] Figure 5 This is a flowchart illustrating the logical process of calculating the comprehensive evaluation index for crimping quality and generating graded early warnings in this invention.

[0024] Figure 6 This is a schematic diagram of the multi-level interaction relationship and data flow between the terminal detection device and the power grid operation and maintenance decision-making platform in this invention. Detailed Implementation

[0025] This invention provides an X-ray non-destructive testing method for the crimping quality of tension clamps in transmission lines. Its core lies in constructing a closed-loop technical system that integrates comprehensive inspection task planning, on-site operation execution, image data processing, and maintenance decision-making. This method achieves high-precision, traceable, and standardized inspection of the internal crimping status of tension clamps through structured data input, automated equipment control, deep learning-driven defect identification, and a risk assessment mechanism based on multi-dimensional parameters. The following will be combined with the appendix... Figure 1 To be continued Figure 6 This section provides a detailed implementation description of each functional module of the system, expanding upon it layer by layer.

[0026] The method begins with step S1: acquiring the topology information, historical operation and maintenance records, and meteorological environmental data of the transmission line to be inspected. This step forms the data foundation for the entire inspection process, and its data integrity and accuracy directly determine the effectiveness of subsequent task generation and risk assessment. Specifically, the tower numbers, span lengths, conductor types, and tension clamp installation locations of the transmission line are extracted from the power grid geographic information system; this information constitutes the spatial topology skeleton of the line, used to determine the absolute coordinates of the physical inspection points.

[0027] Simultaneously, the system retrieves factory batch numbers, crimping records, past test results, and fault repair logs for each tension clamp from the equipment asset management system. This historical maintenance data reflects the service trajectory and potential degradation trends of individual equipment. Furthermore, it obtains wind speed, humidity, rainfall probability, and atmospheric visibility data for the testing area over the next 72 hours from the meteorological monitoring network. These environmental parameters are used to define safe operating windows, avoiding high-altitude X-ray operations under strong winds, heavy rain, or low visibility conditions. All three types of data are connected to the testing task scheduling engine in real time via standardized interfaces, and timestamp alignment and field validation are performed to ensure data consistency.

[0028] Next, step S2 is executed: based on the aforementioned topology information and historical maintenance records, a list of inspection tasks is generated, including priority ranking, inspection point coordinates, equipment configuration parameters, and a safe operation window. The core of this step lies in the fusion and decision-making of multi-source heterogeneous data. First, based on the service life of the tension clamp, the frequency of historical defects, and the lightning or icing risk level of the section, a weighted scoring method is used to determine the priority of each inspection point. The weight allocation follows preset rules: for each additional year of service life, the weight coefficient increases by 0.05; for each additional historical defect, the weight coefficient increases by 0.1; if located in a high lightning density area or a heavily iced area, an additional weight coefficient of 0.15 is added. The total score, after normalization, serves as the basis for priority ranking. Second, based on the three-dimensional coordinates of the tower and the conductor suspension height, the optimal deployment location and X-ray penetration path of the X-ray inspection device are calculated. The calculation process is based on a linear propagation model of X-rays, ensuring that the central axis of the X-ray beam passes through the center of the tension clamp's crimping area and avoids obstruction by adjacent conductors or fittings. Next, based on the wire diameter, clamp material, and ambient temperature and humidity, set the X-ray tube voltage, tube current, exposure time, and detector gain parameters.

[0029] For example, for aluminum-clad steel core conductors with aluminum alloy clamps, the tube voltage is set to 120 kV, the tube current to 5 mA, and the exposure time to 1.2 seconds. If the ambient humidity exceeds 80%, the tube voltage is increased to 130 kV to compensate for air attenuation. Finally, combining wind speed thresholds and visibility limitations from meteorological data, the safe operating time periods for each day are defined. When the wind speed exceeds 10 m / s or the visibility is less than 500 meters, the system automatically excludes that time period, retaining only the window period that meets safety regulations. The final generated inspection task list is output in a structured data format, including the task identifier, geographical coordinates, priority, equipment parameter set, and effective operating time interval for each inspection point.

[0030] The process then proceeds to step S3: The X-ray inspection device is dispatched to the designated tower location according to the inspection task list, and the relative pose of the X-ray source and detector is automatically calibrated. This step ensures the consistency and repeatability of the imaging geometry. The X-ray inspection device is typically mounted on an unmanned aerial vehicle (UAV) platform or insulated basket, and is guided to the vicinity of the target tower by navigation commands sent via a wireless communication link by the task scheduling system. Upon arrival, the device initiates the pose calibration procedure.

[0031] First, the distance from the focal point of the X-ray source to the center of the tension clamp is measured using a laser ranging module and adjusted to a preset focal length, typically 1.5 meters; this distance must ensure a spatial resolution of no less than 50 micrometers. Second, the incident direction of the X-ray beam is determined using an tilt sensor and an electronic compass, ensuring it is perpendicular to the pressing axis; the direction of the pressing axis is determined by pre-modeling the conductor routing and clamp structure, with calibration errors controlled within ±1 degree. Third, the detector plane is orthogonally aligned with the X-ray beam using a servo pan-tilt unit, ensuring the imaging geometry conforms to parallel projection conditions; the angle deviation between the detector plane normal and the X-ray beam direction does not exceed 0.5 degrees. During calibration, the system feeds back the pose parameters to the central controller in real time until all indicators meet the preset tolerance range before entering the imaging stage.

[0032] Next, step S4 is executed: Within the preset safe operating window, the X-ray detection device is controlled to perform multi-angle transmission imaging of the target tension clamp, generating a sequence of raw X-ray images. The imaging process strictly follows radiation safety regulations, and the X-ray generator is started within the operating window. Specifically, the X-ray source and detector assembly is rotated at 30-degree intervals around the pressing axis, acquiring a total of 12 frames of X-ray images from different perspectives. The rotation is driven by a high-precision stepper motor, with an angle positioning error of less than 0.2 degrees. The exposure time for each frame is no less than 0.5 seconds, with a typical value of 0.8 seconds, to ensure the signal-to-noise ratio meets the requirements for subsequent analysis; the spatial resolution is no less than 50 micrometers, determined by the detector pixel size and geometric magnification. All images are embedded with metadata, including precise timestamps (accurate to the millisecond level), equipment status codes (such as tube voltage, tube current, temperature, vibration amplitude), and environmental parameters (wind speed, humidity, air pressure), forming a complete data traceability chain.

[0033] The following step, S5, involves normalizing the radiation dose, correcting geometric distortion, and suppressing background noise on the original X-ray image sequence to obtain a standardized image dataset. This step aims to eliminate the impact of equipment differences, environmental interference, and imaging distortion on image quality. First, a flat panel detector response function is used to compensate for the grayscale values ​​of each pixel. This response function, obtained through factory calibration, characterizes the nonlinear relationship of pixel output under different incident doses, and the compensated image grayscale values ​​correspond linearly to the actual transmission intensity. Second, based on the focal position of the X-ray source and the pixel coordinates of the detector, a back-projection geometric model is applied to eliminate barrel distortion. This model establishes a mapping relationship using a standard calibration plate of known size, remapping the distorted image to an ideal projection plane.

[0034] Next, a wavelet thresholding denoising algorithm is used to filter out high-frequency random noise; specifically, a fourth-order Daubechies wavelet basis is used, with a decomposition level of 4 layers, and the threshold is calculated using a general thresholding method, i.e. ,in The standard deviation of noise. This represents the total number of pixels. The denoising process preserves edge details of the overprinted interface, avoiding the loss of minute defects due to over-smoothing. The processed images are uniformly scaled to a fixed size (e.g., 1024×1024 pixels) and converted to 16-bit grayscale format, forming a standardized image dataset.

[0035] Next, step S6 is executed: the standardized image dataset is input into the pre-trained crimping defect recognition model, which outputs the internal structural feature map of the crimped area and the defect confidence score. The crimping defect recognition model is a multi-scale feature fusion convolutional neural network, specifically designed for the crimped area of ​​tension cable clamps. The input layer receives one or more standardized images and extracts features step-by-step through four downsampling encoding stages:

[0036] The first stage uses 32 3×3 convolutional kernels to extract edges and textures;

[0037] The second stage uses 64 convolutional kernels to capture local structures;

[0038] The third stage uses 128 convolutional kernels to identify the morphology of the compression interface;

[0039] The fourth stage uses 256 convolutional kernels to model the global context.

[0040] Each encoding stage is followed by a max-pooling layer with a downsampling factor of 2. The decoding stage uses transposed convolutions for upsampling and introduces skip connections to concatenate the low-level feature maps from the corresponding encoding stage with the high-level semantic maps to restore spatial details. The final output layer generates a pixel-level defect segmentation mask of the same size as the input image, with each pixel labeled with its category: background, normal compression area, insufficient compression, over-compression, crack, pore, or foreign matter inclusion. Model training employs a joint optimization of the cross-entropy loss function and Dice coefficients. The loss function expression is:

[0041] ;

[0042] in, For pixel-level cross-entropy loss, To predict the Dice similarity coefficient between the mask and the real mask, The balancing coefficient is set to 0.7. The training dataset contains no fewer than 15,000 manually annotated samples, covering different wire types, clamp materials, crimping processes, and defect types. During inference, the model processes 12 frames of multi-angle images separately, then fuses the results from each viewpoint through disparity consistency constraints to generate the final internal structure feature map, and outputs a confidence score for each type of defect, with a score range of 0 to 1.

[0043] Then proceed to step S7: Based on the internal structural feature map and defect confidence score, combined with the material specifications of the tension clamp, crimping process standards, and service life, calculate the comprehensive evaluation index of crimping quality. This index... As a core indicator for quantifying risk, it is defined as the weighted sum of the defect area ratio, defect depth ratio, and service degradation coefficient, and its calculation formula is as follows:

[0044] ;

[0045] in, This represents the defect area ratio, which is the ratio of the defect segmentation mask coverage area to the theoretical crimping area; the theoretical crimping area is determined by the projection of the CAD model corresponding to the wire clamp model. The defect depth ratio is calculated by reconstructing the three-dimensional defect volume using the disparity information of dual-view images. Specifically, a stereo vision matching algorithm is used to calculate the depth map based on two frames of images with an angle greater than 60 degrees, and then integrate to obtain the defect volume. The depth ratio is the ratio of the maximum defect depth to the wall thickness of the line clamp. The service degradation factor is determined based on the service life of the tension clamp and the material fatigue curve; for aluminum alloy clamps, the service life is within 5 years. The weighting coefficient is 1 for periods of 5 to 10 years, 1.1 for periods of 5 to 10 years, and no less than 1.2 for periods exceeding 10 years. , , The values ​​are 0.4, 0.4, and 0.2, respectively, reflecting the equal importance of depth and area to structural strength, as well as the additional impact of service aging on remaining lifespan. (Index) The value ranges from 0 to 1.5, with higher values ​​indicating greater risk.

[0046] Finally, step S8 is executed: a graded early warning signal is generated based on the comprehensive crimping quality evaluation index and pushed to the power grid operation and maintenance decision-making platform to trigger corresponding maintenance strategies. The early warning mechanism adopts a three-level threshold division: when the comprehensive crimping quality evaluation index... When the value is less than 0.3, it is considered a normal state, and a green warning signal is generated, indicating that no intervention is required; when... When the value is between 0.3 and 0.6, it is judged as a minor defect, generating a yellow warning signal. A re-inspection is recommended within 3 months, and the defect is marked as "watched" on the platform. A value greater than 0.6 is considered a serious defect, generating a red warning signal and immediately triggering an emergency defect elimination work order, which is then pushed to the nearest maintenance team's terminal. All warning signals are accompanied by a structured data packet, including defect location coordinates (based on the tower coordinate system), a 3D topographic image (generated from multi-view reconstruction), a list of defect types, a confidence score, and a historical comparison trend chart (showing previous detections). (Value changes). After receiving the signal, the power grid operation and maintenance decision-making platform automatically associates it with equipment ledgers, spare parts inventory, and personnel scheduling information to generate the optimal handling plan and record a closed-loop log of the entire process.

[0047] The above-described process achieves end-to-end automation from data acquisition to decision output through a rigorous sequential sequence of steps S1 to S8. Standardized data interfaces connect each step, ensuring seamless information flow. In the field operation phase, automatic calibration and multi-angle imaging ensure data quality; in the image processing phase, physical models and signal processing algorithms improve the signal-to-noise ratio; in the defect identification phase, deep learning models achieve high-precision segmentation; in the risk assessment phase, geometric, material, and temporal dimensions are integrated to make the assessment results engineering-interpretable; and in the early warning push phase, integration with the existing power grid information system forms a closed-loop management system. The entire system effectively solves the systemic problems of low efficiency, strong subjectivity, and delayed response caused by the isolated operation of the four phases of planning, operation, analysis, and decision-making in traditional detection.

[0048] At the system level, the hardware platform relied upon in this embodiment includes: a mobile detection terminal equipped with an X-ray generator and a flat panel detector, a high-precision pose sensing module (including a laser rangefinder, tilt sensor, and electronic compass), a servo gimbal control system, an edge computing unit, and a wireless communication module. The software system includes a task scheduling engine, an image preprocessing module, a deep learning inference engine, a risk assessment calculator, and an early warning push interface. All modules work collaboratively, supporting both offline and online modes: in communication-restricted areas, the terminal can independently complete imaging and preliminary analysis, synchronizing data upon returning to the base station; in areas with good communication, key data is transmitted back in real time, supporting remote monitoring and intervention. The system design adheres to power industry safety regulations, with X-ray radiation dose strictly controlled within national standard limits, and equipped with multiple interlocking protection mechanisms to ensure the safety of personnel and the public.

[0049] In summary, this embodiment fully discloses the technical solution of the present invention through refined breakdown of method steps, concrete setting of parameters, embedding of anomaly handling mechanism, and deep fusion of multimodal data, which meets the requirements of full disclosure under the Patent Law and provides a solid supporting foundation for the claims.

Claims

1. An X-ray non-destructive testing method for the crimping quality of tension clamps in transmission lines, characterized in that, include: Obtain the topology information, historical operation and maintenance records, and meteorological environmental data of the transmission line to be inspected; Based on the topology information and historical operation and maintenance records, a list of detection tasks is generated, which includes priority sorting, detection point coordinates, equipment configuration parameters, and a safe operation window. According to the detection task list, the X-ray detection device is dispatched to the designated tower position, and the relative pose of the X-ray source and the detector is automatically calibrated; Within the preset safe operating window, the X-ray detection device is controlled to perform multi-angle transmission imaging of the target tension clamp, generating a sequence of original X-ray images. The original X-ray image sequence was subjected to radiation dose normalization, geometric distortion correction, and background noise suppression to obtain a standardized image dataset. The standardized image dataset is input into a pre-trained crimping defect recognition model, which outputs an internal structural feature map of the crimping area and a defect confidence score. Based on the internal structural feature map and defect confidence score, combined with the material specifications, crimping process standards and service life of the tension clamp, the comprehensive evaluation index of crimping quality is calculated. A graded early warning signal is generated based on the comprehensive evaluation index of crimping quality and pushed to the power grid operation and maintenance decision-making platform to trigger the corresponding maintenance strategy.

2. The X-ray non-destructive testing method for the crimping quality of tension clamps in transmission lines according to claim 1, characterized in that, Obtain the topology information, historical operation and maintenance records, and meteorological environmental data of the transmission line to be inspected, including: Extract the tower number, span length, conductor type, and tension clamp installation location of the transmission line from the power grid geographic information system; Retrieve the factory batch, crimping construction records, test results and fault repair logs for each tension clamp from the equipment asset management system; Data on wind speed, humidity, probability of rainfall, and atmospheric visibility in the monitored area for the next 72 hours are obtained from the meteorological monitoring network.

3. The X-ray non-destructive testing method for the crimping quality of tension clamps in transmission lines according to claim 2, characterized in that, Based on the aforementioned topology information and historical operation and maintenance records, a list of detection tasks is generated, including priority sorting, detection point coordinates, device configuration parameters, and a safety operation window. Based on the service life of the tension clamp, the frequency of historical defects, and the risk level of lightning strikes or icing in the section, a weighted scoring method is used to determine the priority of each inspection point. Calculate the optimal deployment location and X-ray penetration path of the X-ray detection device based on the three-dimensional coordinates of the tower and the suspension height of the conductor; Based on the conductor diameter, clamp material, and ambient temperature and humidity, set the X-ray tube voltage, tube current, exposure time, and detector gain parameters. By combining wind speed thresholds and visibility limitations from meteorological data, the safe working time periods for each day are determined.

4. The X-ray non-destructive testing method for the crimping quality of tension clamps in transmission lines according to claim 3, characterized in that, According to the aforementioned detection task list, the X-ray detection device is dispatched to the designated tower position, and the relative pose of the radiation source and detector is automatically calibrated, including: The distance from the focal point of the X-ray source to the center of the tension clamp is measured using a laser ranging module, and then adjusted to the preset focal length. The incident direction of the ray beam is determined using an inclination sensor and an electronic compass, making it perpendicular to the pressing axis; By driving the detector plane to align orthogonally with the X-ray beam using a servo gimbal, the imaging geometry is ensured to meet the conditions for parallel projection.

5. The X-ray non-destructive testing method for the crimping quality of tension clamps in transmission lines according to claim 4, characterized in that, Within a preset safe operating window, the X-ray detection device is controlled to perform multi-angle transmission imaging of the target tension clamp, generating a sequence of raw X-ray images, including: With the pressing axis as the center, the X-ray source and detector assembly was rotated at 30-degree intervals to acquire a total of 12 frames of X-ray images from different perspectives. The exposure time for each frame of the image shall be no less than 0.5 seconds, and the spatial resolution shall be no less than 50 micrometers. All images include timestamps, device status codes, and environmental parameter metadata.

6. The X-ray non-destructive testing method for the crimping quality of tension clamps in transmission lines according to claim 5, characterized in that, The original X-ray image sequence is subjected to radiation dose normalization, geometric distortion correction, and background noise suppression to obtain a standardized image dataset, including: A flat panel detector response function is used to perform dose compensation on the grayscale value of each pixel. Based on the focal position of the X-ray source and the pixel coordinates of the detector, a back-projection geometric model is applied to eliminate barrel distortion. High-frequency random noise is filtered out using a wavelet thresholding denoising algorithm while preserving the edge details of the press-fit interface.

7. The X-ray non-destructive testing method for the crimping quality of tension clamps in transmission lines according to claim 6, characterized in that, The standardized image dataset is input into a pre-trained crimping defect recognition model, which outputs an internal structural feature map of the crimped area and a defect confidence score, including: The crimping defect recognition model is a multi-scale feature fusion convolutional neural network. Its input layer receives a standardized image dataset, extracts local texture and global morphological features through four downsampling encoding stages, and then fuses low-level edge information with high-level semantic information through skip connections. Finally, a pixel-level defect segmentation mask is generated in the output layer. Defects include insufficient crimping, over-creasing, cracks, porosity, and foreign matter inclusions; The model training employs joint optimization of the cross-entropy loss function and the Dice coefficient.

8. The X-ray non-destructive testing method for the crimping quality of tension clamps in transmission lines according to claim 7, characterized in that, Based on the aforementioned internal structural feature map and defect confidence score, combined with the material specifications, crimping process standards, and service life of the tension clamp, a comprehensive crimping quality evaluation index is calculated, including: Define the comprehensive evaluation index of crimping quality It is a weighted sum of the defect area ratio, defect depth ratio, and service degradation coefficient.

9. The X-ray non-destructive testing method for the crimping quality of tension clamps in transmission lines according to claim 8, characterized in that, The defect area ratio is the ratio of the area covered by the defect segmentation mask to the theoretical pressing area; The defect depth ratio is calculated by reconstructing the three-dimensional defect volume using the parallax information from dual-view images; The service degradation coefficient is determined based on the service life of the tension clamp and the material fatigue curve. The degradation coefficient of the clamp that has been in service for more than 10 years shall not be less than 1.

2.

10. The X-ray non-destructive testing method for the crimping quality of tension clamps in transmission lines according to claim 9, characterized in that, A graded early warning signal is generated based on the comprehensive evaluation index of crimping quality and pushed to the power grid operation and maintenance decision-making platform, including: When the comprehensive evaluation index of crimping quality When the value is less than 0.3, it is considered a normal state, and a green warning signal is generated; when If the value is between 0.3 and 0.6, it is considered a minor defect, generating a yellow warning signal, and a re-inspection is recommended within 3 months; when When the value is greater than 0.6, it is judged as a serious defect, a red warning signal is generated, and an emergency defect elimination work order is immediately triggered; All warning signals are accompanied by the coordinates of the defect location, a three-dimensional topographic map, and a historical comparison trend chart.

Images