A tension clamp X-ray information analysis system
Through the X-ray information analysis system, combined with composite imaging technology and artificial intelligence algorithms, the problems of low detection accuracy and inability to intuitively display the internal structure in traditional detection methods are solved, efficient detection and timely warning of tension clamps are realized, and power is guaranteed. System security and reliability.
Patent Information
- Application Number
- CN202411295708.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-18
- Publication Date
- 2024-11-19
- Estimated Expiration
- 2044-09-18
AI Technical Summary
The traditional tension-resistant wire clip detection method has subtle defects such as low detection accuracy, low efficiency, lots of manual intervention, and the inability to accurately identify tiny cracks and corrosion marks, resulting in the inability to detect and deal with potential risks in a timely manner, affecting the safety and reliability of transmission lines.
Using X-ray information analysis system, including X-ray imaging module, image analysis module, three-dimensional reconstruction module, monitoring and early warning module and augmented reality interaction module, efficient detection is achieved through composite imaging technology, artificial intelligence algorithm, three-dimensional reconstruction algorithm and reinforcement learning algorithm. , accurate analysis and timely warning.
It improves the efficiency and accuracy of tension clamp detection, can intuitively display the internal structure and defect location, reduce manual misjudgment, achieve timely early warning and fault prevention, and ensure the safe operation of the power system.
Smart Images

Figure CN118823013B_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of X-ray information analysis, in particular to an X-ray information analysis system for a tension clamp. Background Art
[0002] Strain clamps are important components in power transmission lines, mainly used to fix conductors and support towers, ensuring that the conductors can withstand a certain amount of tension and maintain the stability of the line. Traditional methods for detecting tension clamps mainly rely on visual inspection and mechanical measurement, which usually have problems such as low detection accuracy, low efficiency, and frequent manual intervention. In addition, traditional detection methods cannot accurately evaluate the defects inside the tension clamps, making it difficult to detect potential structural problems in a timely manner, resulting in threats to the safety and reliability of power transmission lines.
[0003] With the development of modern science and technology, X-ray detection technology has been widely used in the field of industrial non-destructive testing. X-rays can penetrate objects and present their internal structural information in images, making it possible to detect internal defects of tension clamps. However, the current X-ray detection of tension clamps still has the following problems: the existing technology relies on manual inspection and simple imaging technology, and has problems such as low detection efficiency, high labor costs, and high misjudgment rate. In particular, for subtle defects such as tiny cracks and corrosion marks, traditional methods are difficult to accurately identify, resulting in potential risks that cannot be discovered and handled in a timely manner; existing detection technologies usually only provide flat images, which cannot intuitively display the internal structure and defect location of the tension clamp, which brings difficulties to engineers' analysis and evaluation, and affects the accuracy and effectiveness of maintenance and repair work. Summary of the invention
[0004] To solve the above problems, the present invention provides an X-ray information analysis system for tension clamps, which solves the problems of low efficiency, high misjudgment rate and inability to intuitively display the internal structure and defect location of tension clamps in X-ray detection of tension clamps, realizes efficient detection, accurate analysis and timely warning of tension clamps, and provides comprehensive and reliable decision-making support for maintenance personnel.
[0005] To achieve the above object, the technical solution adopted by the present invention is:
[0006] A tension clamp X-ray information analysis system, comprising an X-ray imaging module, an image analysis module, a three-dimensional reconstruction module, a monitoring and early warning module and an augmented reality interaction module which are sequentially connected in communication;
[0007] The X-ray imaging module is used to obtain X-ray image data of the tension clamp by composite imaging technology, and transmit the X-ray image data to the image analysis module;
[0008] The image analysis module is used to process and analyze the received X-ray image data using an artificial intelligence algorithm, automatically identify tiny cracks, corrosion marks and other potential defects, and generate an image analysis report;
[0009] The three-dimensional reconstruction module is used to construct a three-dimensional model of the tension clamp using a three-dimensional reconstruction algorithm based on the analysis results generated by the image analysis module, and to display a three-dimensional visualization image of its internal structure;
[0010] The monitoring and early warning module is used to monitor and dynamically adjust the early warning threshold in real time through a reinforcement learning algorithm, automatically generate early warning information when an abnormality is detected, and notify relevant personnel;
[0011] The augmented reality interaction module is used to receive the warning information and three-dimensional model data of the monitoring and warning module, and provide an interactive interface through augmented reality technology.
[0012] Further, the composite imaging technology includes multi-angle imaging, digital radiography and computed tomography imaging;
[0013] The multi-angle imaging is used to obtain images of the tension clamp from multiple angles;
[0014] The digital radiography is used to generate high-resolution digital images;
[0015] The computer tomography imaging is used to obtain a tomographic image of the tension clamp, providing detailed information and in-depth analysis of its internal structure.
[0016] Furthermore, the operation process of the image analysis module includes the following steps:
[0017] Acquire X-ray image data and perform preprocessing operations, using adaptive histogram equalization image enhancement technology to improve image contrast and details;
[0018] Apply a deep learning algorithm based on a multi-scale convolutional neural network to extract multi-scale features from the pre-processed images, including features of tiny cracks, corrosion marks, and other potential defects;
[0019] Use the temporal convolutional network to analyze the defect changes in the image sequence, generate the defect development curve, and evaluate the severity of the defect and its evolution trend;
[0020] Based on the analysis results of the defect development curve, an image analysis report is generated, which includes the defect location, type and severity of the X-ray image.
[0021] Furthermore, the formula of the defect development curve is as follows:
[0022]
[0023] in, Indicates the degree of defect development at time t. The larger the value, the more serious the defect. Indicates the initial defect level; represents the exponential growth coefficient of crack development; represents the corrosion influence coefficient; Indicates time The development function of corrosion traces at each moment; t represents the time variable; Represents the small time increment used in the integral calculation to cumulatively calculate the effect of corrosion traces on the overall defect over time.
[0024] Furthermore, the formula of the multi-scale convolutional neural network is as follows:
[0025]
[0026] in, represents the defect feature map after processing by the multi-scale convolutional neural network, including the location, shape and severity of tiny cracks, corrosion marks and other potential defects; x represents the input X-ray image data; n represents the number of convolutional layers; Represents the weight parameter of the i-th convolution layer, which is used to adjust the response strength of the convolution kernel to the input image; Represents the feature extraction function of the i-th convolution layer on the input x.
[0027] Furthermore, the operation process of the 3D reconstruction module includes the following steps:
[0028] receiving and preprocessing X-ray image data from an image analysis module;
[0029] Using the point cloud generation algorithm, the pre-processed X-ray image data is converted into three-dimensional point cloud data;
[0030] Adaptive Delaunay triangulation algorithm is used to automatically adjust the mesh density according to the density of point cloud data to generate a three-dimensional mesh model to display the complex geometric structure of the tension clamp;
[0031] Apply intelligent marking algorithm on 3D mesh model to automatically mark the location and type of defects identified by image analysis module, use color coding to distinguish different types of defects, and use finite element analysis method to dynamically optimize the model;
[0032] The final generated 3D model and its defect information are displayed in multiple levels through an augmented reality interface, providing 3D visualization images and detailed defect analysis reports.
[0033] Furthermore, the formula of the Delaunay triangulation algorithm is as follows:
[0034]
[0035] in, Delaunay triangulation results represent triangulated mesh models generated by triangulating point cloud data, showing the complex geometric structure of the tension clamp; Represents three points in the point cloud data , and P represents the three-dimensional point cloud data, which is converted from the pre-processed X-ray image data and represents the spatial distribution points of the tension clamp; , and Represents any three points in the three-dimensional point cloud data, which are used to construct Delaunay triangles; m represents the total number of triangles in the point set P.
[0036] Furthermore, the reinforcement learning algorithm includes a deep Q network algorithm or a policy gradient algorithm.
[0037] Further, the augmented reality interaction module includes an augmented reality display unit and an interaction control unit;
[0038] The augmented reality display unit is used to display the three-dimensional model and warning information on the mobile device;
[0039] The interactive control unit is used to interact with the system through gestures or voice commands.
[0040] The beneficial effects of the present invention are:
[0041] The present invention obtains high-quality X-ray image data through composite imaging technology, and the application of composite imaging technology makes the image acquisition clearer. Through the combination of X-ray imaging module and image analysis module, the system can efficiently obtain and process X-ray image data of tension clamp. Combined with the artificial intelligence algorithm in the image analysis module, it can quickly and accurately identify tiny cracks, corrosion marks and other potential defects in the tension clamp, greatly improve the detection efficiency and accuracy, and reduce the possibility of manual misjudgment. Using the three-dimensional reconstruction algorithm, the image analysis results are constructed into a three-dimensional model of the tension clamp, providing a three-dimensional visualization image of the internal structure, helping relevant personnel to more intuitively understand and evaluate the state of the tension clamp, so as to make more accurate maintenance or replacement decisions. The monitoring and early warning module monitors the state of the tension clamp in real time through the reinforcement learning algorithm, and dynamically adjusts the early warning threshold. When an abnormal situation is detected, it can automatically generate early warning information and notify relevant personnel. In this way, timely response can be achieved, the occurrence of faults can be effectively prevented, and the safe operation of the power system can be guaranteed. The augmented reality interaction module combines the early warning information with the three-dimensional model data and provides an interactive interface through augmented reality technology. This not only improves the user's operating experience, but also enables engineers to view and operate 3D models in a real environment, allowing for more intuitive fault diagnosis and maintenance planning. BRIEF DESCRIPTION OF THE DRAWINGS
[0042] Figure 1 It is a module schematic diagram of a tension clamp X-ray information analysis system of the present invention.
[0043] Figure 2 It is a flowchart of the operation process of the image analysis module in the present invention.
[0044] Figure 3 It is a flow chart of the operation process of the three-dimensional reconstruction module in the present invention. DETAILED DESCRIPTION
[0045] See also Figure 1-Figure 3 As shown, the present invention relates to an X-ray information analysis system for a tension clamp.
[0046] Example
[0047] A tension clamp X-ray information analysis system, comprising an X-ray imaging module, an image analysis module, a three-dimensional reconstruction module, a monitoring and early warning module and an augmented reality interaction module which are sequentially connected in communication;
[0048] The X-ray imaging module is used to obtain X-ray image data of the tension clamp by composite imaging technology, and transmit the X-ray image data to the image analysis module;
[0049] Wherein, the composite imaging technology includes multi-angle imaging, digital radiography and computed tomography imaging;
[0050] The multi-angle imaging is used to obtain images of the tension clamp from multiple angles;
[0051] The digital radiography is used to generate high-resolution digital images;
[0052] The computer tomography imaging is used to obtain a tomographic image of the tension clamp, providing detailed information and in-depth analysis of its internal structure.
[0053] Specifically, a multi-angle imaging device is used to perform a 360-degree full-scale scan of the tension clamp. The imaging device consists of multiple X-ray sources and detectors, and can capture images at different angles simultaneously. The tension clamp is fixed in the center position by a rotating platform, which can be rotated at a preset angle during the imaging process to obtain images at different angles. The multi-angle image capture process uses high frame rate continuous shooting technology to ensure that a large amount of image data is obtained in a short time, and each frame of the image has a high resolution.
[0054] Digital radiography (DR) technology is used to capture the attenuation image of X-rays passing through the tension clamp using a high-sensitivity digital detector. The digital detector has high resolution and high dynamic range, which can clearly display the tiny structure and details of the tension clamp. The image processing software pre-processes the collected digital images, including denoising, contrast enhancement and artifact correction, to ensure the clarity and accuracy of the image.
[0055] The tension clamp is scanned layer by layer using computed tomography (CT) technology. The CT scan rotates around the tension clamp through the X-ray source, and the detector on the opposite side receives the transmitted X-ray signal. The system reconstructs the scan data of each layer to generate a tomographic image of the tension clamp, providing detailed information on its internal structure. Advanced image reconstruction algorithms, such as iterative reconstruction algorithms or filtered back-projection algorithms, are used to convert the original scan data into high-quality tomographic images.
[0056] The X-ray imaging module is equipped with a high-speed data transmission interface to transmit the acquired image data to the image analysis module in real time. The data transmission adopts an encryption protocol to ensure the security and integrity of the data. At the same time, the system has a built-in large-capacity storage device for storing raw image data and processed image data, which is convenient for subsequent analysis and backtracking. To ensure imaging accuracy, the X-ray imaging module performs system calibration regularly. The calibration process includes the output power calibration of the X-ray source, the sensitivity calibration of the detector, and the calibration of the imaging geometric parameters. The module is equipped with an automatic calibration function, which can automatically adjust the system parameters according to the preset calibration standards to ensure the stability and consistency of the imaging quality.
[0057] The image analysis module is used to process and analyze the received X-ray image data using an artificial intelligence algorithm, automatically identify tiny cracks, corrosion marks and other potential defects, and generate an image analysis report;
[0058] The operation process of the image analysis module includes the following steps:
[0059] Acquire X-ray image data and perform preprocessing operations, using adaptive histogram equalization image enhancement technology to improve image contrast and details;
[0060] It should be noted that the median filter method is used to process the image. The median filter replaces each pixel value with the median of the neighboring pixel values to remove random noise and retain the edge details of the image. The Gaussian filter is also a commonly used denoising method. It reduces the impact of noise by performing Gaussian smoothing on the image.
[0061] Image enhancement: Adopt the adaptive histogram equalization (CLAHE) technology to improve the contrast of the image. CLAHE enhances the local details of the image through local histogram equalization, making tiny cracks and corrosion marks more obvious. The specific implementation steps of CLAHE include: dividing the image into multiple small blocks, performing histogram equalization on each small block independently, and then recombining these small blocks into an enhanced image.
[0062] Apply a deep learning algorithm based on a multi-scale convolutional neural network to extract multi-scale features from the pre-processed images, including features of tiny cracks, corrosion marks, and other potential defects;
[0063] Specifically, a multi-scale convolutional neural network (CNN) is used to extract features from the preprocessed images. The multi-scale CNN model contains multiple convolutional layers and pooling layers to extract features at different scales. The initial convolutional layer extracts low-level features such as edges and textures; the intermediate layer extracts intermediate features such as shapes and patterns; and the high-level convolutional layer extracts high-level features such as the specific form of the defect. The CNN model is trained with a large amount of labeled training data. The training data includes various types of defect images and their annotation information, such as crack locations, corrosion areas, etc. During the training process, the cross entropy loss function is used to measure the difference between the model prediction results and the actual annotations, and the network weights are adjusted through the back propagation algorithm to gradually improve the model recognition accuracy. After the multi-scale feature extraction is completed, the CNN model generates feature maps, which contain detailed information on subtle defects such as tiny cracks and corrosion marks for subsequent analysis.
[0064] Use the temporal convolutional network to analyze the defect changes in the image sequence, generate the defect development curve, and evaluate the severity of the defect and its evolution trend;
[0065] Specifically, a temporal convolutional network (TCN) is used to analyze a sequence of continuously shot X-ray images. The TCN model processes and analyzes defect changes in an image sequence through multiple temporal convolutional layers. The TCN model extracts defect features from each frame of the image, generates time series data, and records the state changes of the defect at different time points. Based on the time series data, the TCN model generates a defect development curve that displays information such as the crack propagation rate and the corrosion area growth. According to the defect development curve, the severity of the defect is evaluated. Quantitative indicators such as the rate of change of the crack length and the expansion rate of the corrosion area are used to measure the severity of the defect and its future evolution trend.
[0066] Based on the analysis results of the defect development curve, an image analysis report is generated, wherein the image analysis report includes the defect location, type and severity of the X-ray image;
[0067] Furthermore, the formula of the multi-scale convolutional neural network is as follows:
[0068]
[0069] in, represents the defect feature map after processing by the multi-scale convolutional neural network, including the location, shape and severity of tiny cracks, corrosion marks and other potential defects; x represents the input X-ray image data; n represents the number of convolutional layers; Represents the weight parameter of the i-th convolution layer, which is used to adjust the response strength of the convolution kernel to the input image; Represents the feature extraction function of the i-th convolution layer on the input x.
[0070] The formula of the defect development curve is as follows:
[0071]
[0072] in, Indicates the degree of defect development at time t. The larger the value, the more serious the defect. Indicates the initial defect level; represents the exponential growth coefficient of crack development; represents the corrosion influence coefficient; Indicates time The development function of corrosion traces at each moment; t represents the time variable; Represents the small time increment used in the integral calculation to cumulatively calculate the effect of corrosion traces on the overall defect over time.
[0073] The three-dimensional reconstruction module is used to construct a three-dimensional model of the tension clamp using a three-dimensional reconstruction algorithm based on the analysis results generated by the image analysis module, and to display a three-dimensional visualization image of its internal structure;
[0074] The operation process of the 3D reconstruction module includes the following steps:
[0075] Receive and pre-process X-ray image data from the image analysis module; these data include image spatial information and defect annotation information.
[0076] Using the point cloud generation algorithm, the pre-processed X-ray image data is converted into three-dimensional point cloud data;
[0077] Specifically, volume rendering technology is used to convert two-dimensional X-ray image data into three-dimensional point cloud data. Volume rendering technology generates dense point cloud data by calculating the position of each pixel in three-dimensional space. The point cloud generation process includes spatial coordinate conversion and density calculation to ensure that the spatial position of each point is accurate.
[0078] Adaptive Delaunay triangulation algorithm is used to automatically adjust the mesh density according to the density of point cloud data to generate a three-dimensional mesh model to display the complex geometric structure of the tension clamp;
[0079] Specifically, the point cloud data is triangulated using an adaptive Delaunay triangulation algorithm, which automatically adjusts the density of the triangular mesh to adapt to the complex geometric structure of the tension clamp by analyzing the density of the point cloud data.
[0080] The specific steps of the adaptive Delaunay triangulation algorithm include: sampling of point cloud data, generation of initial triangular mesh, adjustment of mesh density according to point cloud density, and optimization of mesh topology.
[0081] The generated 3D mesh model can accurately display the geometric details of the tension clamp, including its external shape and internal structure. The mesh model forms a complete 3D shape by connecting the points in the point cloud to form polygonal patches.
[0082] Apply intelligent marking algorithm on 3D mesh model to automatically mark the location and type of defects identified by image analysis module, use color coding to distinguish different types of defects, and use finite element analysis method to dynamically optimize the model;
[0083] Specifically, the intelligent labeling algorithm is applied to the 3D mesh model to automatically annotate the defect location and type identified by the image analysis module. The intelligent labeling algorithm combines the defect location coordinates and the 3D mesh data to accurately locate the defect. Color coding is used to distinguish different types of defects, such as cracks, corrosion, etc. Each defect type corresponds to a different color, which is convenient for users to intuitively identify. The 3D model is dynamically optimized using the finite element analysis method. Finite element analysis evaluates the impact of defects on the structural performance of the tension clamp by performing mechanical analysis on the mesh model. The dynamic optimization process includes mesh refinement, stress distribution calculation and structural stability analysis to ensure that the 3D model can truly reflect the mechanical properties of the clamp while displaying defects.
[0084] The final generated 3D model and its defect information are displayed in multiple levels through an augmented reality interface, providing 3D visualization images and detailed defect analysis reports.
[0085] Furthermore, the formula of the Delaunay triangulation algorithm is as follows:
[0086]
[0087] in, Delaunay triangulation results represent triangulated mesh models generated by triangulating point cloud data, showing the complex geometric structure of the tension clamp; Represents three points in the point cloud data , and P represents the three-dimensional point cloud data, which is converted from the pre-processed X-ray image data and represents the spatial distribution points of the tension clamp; , and Represents any three points in the three-dimensional point cloud data, which are used to construct Delaunay triangles; m represents the total number of triangles in the point set P.
[0088] Specific calculation steps:
[0089] 1. The X-ray image data received and preprocessed from the image analysis module is converted into three-dimensional point cloud data.
[0090] 2. Traverse all points in the point set P, and for every three points , and Constructing triangles .
[0091] 3. Ensure that any triangle The circumcircle of does not contain other points. If the circumcircle contains other points, adjust the point selection and rebuild the triangle.
[0092] 4. Repeat steps 2 and 3 until all points are traversed and the final Delaunay triangulation D(P) is generated to form a triangular mesh model.
[0093] The monitoring and early warning module is used to monitor in real time and dynamically adjust the early warning threshold through a reinforcement learning algorithm, automatically generate early warning information when an abnormality is detected, and notify relevant personnel; the reinforcement learning algorithm includes a deep Q network algorithm or a policy gradient algorithm.
[0094] Specifically, real-time image data and analysis reports are obtained from the X-ray imaging module and the image analysis module, including various status information of the tension clamp. The data is preprocessed, including normalization, noise reduction and feature extraction, to ensure the stability and consistency of data input. The preprocessed data includes detailed information such as the location of micro cracks, corrosion traces, and fault images. The reinforcement learning model is designed using the deep Q network (DQN) algorithm or the policy gradient algorithm. The input of the model is the preprocessed status data, and the output is the warning decision. The reinforcement learning model is trained offline with a large amount of historical data. During the training process, the model is continuously iterated to optimize the warning strategy and improve the accuracy of the warning. The reinforcement learning model is deployed in the monitoring and warning module to receive and process the status data from the image analysis module in real time. The module continuously analyzes the real-time data, evaluates the current status of the tension clamp, and compares it with the preset threshold. According to the output of the reinforcement learning model, the warning threshold is dynamically adjusted to adapt to different working environments and usage conditions. When an abnormal state (such as crack extension, corrosion intensification, etc.) is detected, the module immediately generates a warning message. The warning information includes the abnormal type, location, severity and recommended treatment measures. Warning information is delivered to relevant personnel through the system's built-in notification mechanism. Notification methods include SMS, email, and system notifications to ensure that warning information can be delivered in a timely manner. In addition, the system will provide images and 3D models to assist engineers in further evaluation and decision-making. The monitoring and warning module is equipped with an adaptive optimization mechanism to adjust the warning threshold through continuous learning and feedback. The system records the actual effect of each warning and evaluates the accuracy and timeliness of the warning. Based on the feedback results, the reinforcement learning model automatically adjusts parameters and strategies to optimize the performance of the warning system to ensure that efficient warning capabilities can be maintained in different working environments.
[0095] In one embodiment, a monitoring and early warning module is constructed using a deep Q-network algorithm to make decisions based on the current state (such as image data, historical data) and actions (such as adjusting the early warning threshold). The Q-network is composed of a multi-layer convolutional neural network (CNN), with the input being the current state and the output being the Q value of each possible action. The training process is stabilized through experience replay and target network technology.
[0096] In one embodiment, a policy gradient algorithm is used to directly optimize the policy network to maximize the probability of selecting the best action under a given state. The policy network consists of a multi-layer feedforward neural network (FFNN), with the current state as input and the probability distribution of the action as output. Advantage function and signal normalization technology are used to improve the algorithm stability and convergence speed.
[0097] The augmented reality interaction module is used to receive the warning information and three-dimensional model data of the monitoring and warning module, and provide an interactive interface through augmented reality technology;
[0098] Wherein, the augmented reality interaction module includes an augmented reality display unit and an interaction control unit;
[0099] The augmented reality display unit is used to display the three-dimensional model and warning information on the mobile device;
[0100] The interactive control unit is used to interact with the system through gestures or voice commands.
[0101] It should be noted that the augmented reality display unit includes mobile devices (such as smartphones and tablets). These devices have high-resolution displays and powerful processing capabilities, and can run AR applications smoothly. The three-dimensional model can display the appearance and internal structure of the wire clamp, and the user can zoom in, zoom out and rotate the model through touch or gesture to obtain detailed information from different perspectives. When the monitoring and early warning module generates an early warning message, the augmented reality display unit will pop up a relevant notification on the device screen. The notification includes text descriptions, image annotations and three-dimensional model indications to help users quickly understand the problem. The user can click or gesture to select the early warning notification to view detailed information such as crack location, type, severity and recommended treatment measures.
[0102] The interactive control unit interacts with the system through natural gestures. Common gestures include click, slide, rotate, and zoom. The interactive control unit has a built-in voice recognition system that supports voice commands in multiple languages. Users can control the system through voice commands, which enhances the ease of use of the system, especially when the user's hands are occupied. The system provides instant operation feedback, including visual feedback and sound prompts. After each operation of the user, the interface will respond in time and confirm the success of the operation through animation effects and sound effects, enhancing the user experience.
[0103] In summary, the present invention uses the composite imaging technology in the X-ray imaging module, and the system can fully capture the internal structure and details of the tension clamp in a multi-angle, high-resolution and tomographic imaging manner. This multi-dimensional imaging method ensures the integrity and accuracy of the data, providing a solid foundation for subsequent analysis. The image analysis module uses artificial intelligence algorithms to perform in-depth processing and analysis of X-ray image data, and can automatically identify potential defects such as tiny cracks and corrosion marks. This module ensures the high accuracy and comprehensiveness of defect identification through multi-scale convolutional neural networks and temporal convolutional networks. The generated defect development curve can intuitively display the evolution trend of defects and provide a scientific basis for maintenance decisions. The three-dimensional reconstruction module converts two-dimensional image data into three-dimensional point cloud data, and generates an accurate three-dimensional mesh model through an adaptive Delaunay triangulation algorithm to display the complex geometric structure and internal details of the tension clamp. The module also intuitively annotates defect information through color coding and intelligent marking algorithms to assist engineers in detailed structural analysis and defect assessment.
[0104] The monitoring and early warning module in the present invention uses a reinforcement learning algorithm (such as a deep Q network algorithm or a policy gradient algorithm) to monitor the status of the tension clamp in real time, dynamically adjust the early warning threshold, and ensure the timely discovery and handling of abnormal conditions. The early warning information generated by the system contains detailed information about the defect type, location, and severity, and is quickly communicated to relevant personnel through a variety of notification methods to ensure the safe operation of the clamp. The augmented reality interaction module uses AR technology to intuitively display the three-dimensional model and early warning information on the mobile device. Users can interact with the system through gestures or voice commands, which enhances the user experience and the operability of the system. This module not only provides real-time visual feedback, but also supports multi-language voice control, which is convenient for on-site operators.
[0105] The above implementation modes are merely descriptions of the preferred implementation modes of the present invention, and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary engineering and technical personnel in the field shall fall within the protection scope determined by the claims of the present invention.
Claims
1. A tension clamp X-ray information analysis system, It is characterized in that It includes an X-ray imaging module, an image analysis module, a three-dimensional reconstruction module, a monitoring and early warning module, and an augmented reality interaction module which are sequentially connected in communication; The X-ray imaging module is used to obtain X-ray image data of the tension clamp by composite imaging technology, and transmit the X-ray image data to the image analysis module; The image analysis module is used to process and analyze the received X-ray image data using an artificial intelligence algorithm, automatically identify tiny cracks, corrosion marks and other potential defects, and generate an image analysis report; The three-dimensional reconstruction module is used to construct a three-dimensional model of the tension clamp using a three-dimensional reconstruction algorithm based on the analysis results generated by the image analysis module, and to display a three-dimensional visualization image of its internal structure; The monitoring and early warning module is used to monitor and dynamically adjust the early warning threshold in real time through a reinforcement learning algorithm, automatically generate early warning information when an abnormality is detected, and notify relevant personnel; The augmented reality interaction module is used to receive the warning information and three-dimensional model data of the monitoring and warning module, and provide an interactive interface through augmented reality technology; The operation process of the image analysis module includes the following steps: Acquire X-ray image data and perform preprocessing operations, using adaptive histogram equalization image enhancement technology to improve image contrast and details; Apply a deep learning algorithm based on a multi-scale convolutional neural network to extract multi-scale features from the pre-processed images, including features of tiny cracks, corrosion marks, and other potential defects; Use the temporal convolutional network to analyze the defect changes in the image sequence, generate the defect development curve, and evaluate the severity of the defect and its evolution trend; Based on the analysis results of the defect development curve, an image analysis report is generated, wherein the image analysis report includes the defect location, type and severity of the X-ray image; The formula of the defect development curve is as follows: ; in, Indicates the degree of defect development at time t. The larger the value, the more serious the defect. Indicates the initial defect level; represents the exponential growth coefficient of crack development; represents the corrosion influence coefficient; Indicates time The development function of corrosion traces at each moment; t represents the time variable; Represents the small time increment used in the integral calculation, which is used to cumulatively calculate the impact of the development of corrosion traces over time on the overall defect; The operation process of the 3D reconstruction module includes the following steps: receiving and preprocessing X-ray image data from an image analysis module; Using the point cloud generation algorithm, the pre-processed X-ray image data is converted into three-dimensional point cloud data; Adaptive Delaunay triangulation algorithm is used to automatically adjust the mesh density according to the density of point cloud data to generate a three-dimensional mesh model to display the complex geometric structure of the tension clamp; Apply intelligent marking algorithm on 3D mesh model to automatically mark the location and type of defects identified by image analysis module, use color coding to distinguish different types of defects, and use finite element analysis method to dynamically optimize the model; The final generated 3D model and its defect information are displayed in multiple levels through an augmented reality interface, providing 3D visualization images and detailed defect analysis reports; The formula for the Delaunay triangulation algorithm is as follows: ; in, Delaunay triangulation results represent triangulated mesh models generated by triangulating point cloud data, showing the complex geometric structure of the tension clamp; Represents three points in the point cloud data , and P represents the three-dimensional point cloud data, which is converted from the pre-processed X-ray image data and represents the spatial distribution points of the tension clamp; , and Represents any three points in the three-dimensional point cloud data, which are used to construct Delaunay triangles; m represents the total number of triangles in the point set P.
2. A tension clamp X-ray information analysis system according to claim 1, It is characterized in that The composite imaging technology includes multi-angle imaging, digital radiography and computed tomography imaging; The multi-angle imaging is used to obtain images of the tension clamp from multiple angles; The digital radiography is used to generate high-resolution digital images; The computer tomography imaging is used to obtain a tomographic image of the tension clamp, providing detailed information and in-depth analysis of its internal structure.
3. The X-ray information analysis system for tension clamps according to claim 1, It is characterized in that The formula of the multi-scale convolutional neural network is as follows: ; in, represents the defect feature map after processing by the multi-scale convolutional neural network, including the location, shape and severity of tiny cracks, corrosion marks and other potential defects; x represents the input X-ray image data; n represents the number of convolutional layers; Indicates The weight parameter of the layer convolution is used to adjust the response strength of the convolution kernel to the input image; Indicates Layer convolution on the input The feature extraction function.
4. The X-ray information analysis system for tension clamps according to claim 1, It is characterized in that The reinforcement learning algorithm includes a deep Q network algorithm or a policy gradient algorithm.
5. The X-ray information analysis system for tension clamps according to claim 1, It is characterized in that The augmented reality interaction module includes an augmented reality display unit and an interaction control unit; The augmented reality display unit is used to display the three-dimensional model and warning information on the mobile device; The interactive control unit is used to interact with the system through gestures or voice commands.
