Power transmission line abrasion detection method and system

Abstract

The invention relates to the technical field of catenary state detection, in particular to a power transmission line abrasion detection method and system.The power transmission line abrasion detection system comprises a transferring mechanism used for bearing a detection mechanism to move along a catenary, and the detection angle of the detection mechanism is adjusted through a multi-degree-of-freedom mechanical arm; the detection mechanism comprises a camera module and a point cloud scanning part; the detection mechanism is in communication connection with the data processing device; the data processing device comprises an edge computing platform, a communication module and an alarm module; and a contact network detection model in the edge computing platform processes the data information and outputs a detection report of the contact network in real time. Through stable positioning, multi-degree-of-freedom adjustment, time sequence modeling of machine vision and efficient fusion of point cloud data of the transfer mechanism and the mechanical arm on the detection mechanism, high-precision and low-delay abrasion detection is automatically achieved in the whole process, the precision reaches the micron level, an accurate and real-time monitoring means is provided for railway maintenance, and the working efficiency is improved. And the safety and the reliability of a railway system can be improved.

Description

Technical Field

[0001] This invention relates to the field of overhead contact line condition detection technology, and in particular to a method and system for detecting wear on power transmission lines. Background Technology

[0002] With the continuous increase in train speed, the wear and tear on the overhead contact line is severe, affecting the reliability of railway operation. Complex measurement environments, lighting conditions, and changes in the shape of the overhead contact line can easily lead to increased errors or missing data in the acquisition of contour data, resulting in measurement results with poor robustness, low accuracy, and high failure rate.

[0003] In existing technologies, manually inspecting the wear of the contact network at close range is costly and cannot guarantee safety. On the other hand, monitoring the wear of the contact network by installing detection sensors is difficult due to the long distance, the influence of complex environments, low detection accuracy, and poor reliability of the detection results.

[0004] To address the problems in the existing technology, this invention provides a method and system for detecting wear on power transmission lines. Summary of the Invention

[0005] The purpose of this invention is to provide a method and system for detecting wear on power transmission lines, so as to solve the technical problems of low efficiency and low accuracy of manual or sensor detection in complex environments in the prior art.

[0006] The technical solution of this invention is: a power transmission line wear detection system, comprising a transfer mechanism suspended on the contact network, the transfer mechanism being used to support a detection mechanism moving along the contact network, and a multi-degree-of-freedom robotic arm mounted on the transfer mechanism adjusting the detection angle of the detection mechanism; the detection mechanism includes a camera module and a point cloud scanning component for collecting data information from the contact network; the detection mechanism is communicatively connected to a data processing device to transmit the collected data information to the data processing device, the data processing device including an edge computing platform, a communication module, and an alarm module; the edge computing platform has a built-in contact network detection model for processing the collected data information, and the alarm module issues an early warning when the obtained data is abnormal or exceeds a preset wear threshold.

[0007] Preferably, the overhead contact line detection model includes: The preprocessing module preprocesses the acquired images and point cloud data, filtering and reducing noise. The feature extraction module performs feature extraction on the preprocessed catenary image and point cloud information, including structural feature extraction and intensity feature extraction; The fusion module guides the fusion of structural and intensity features based on confidence levels, and the integrated features generate a unified representation. The temporal modeling module is used to model the dynamic evolution of local structures over time. It takes a unified representation of feature integration as input and outputs time-aware features. The change perception module maps the output of the time-series modeling module into point-level deformation responses. Combined with the original point cloud, it forms a visualized 3D cloud map through color mapping, revealing the microstructural changes on the surface of the contact wire.

[0008] Preferably, the temporal modeling module includes a multi-head attention unit, a feedback network, a residual structure, and a layer normalization unit.

[0009] Preferably, the unified representation of feature integration is as follows: three-dimensional coordinates and intensities extracted from the original input point cloud data, the three-dimensional coordinates and intensities are constructed into multimodal input pairs, and input into the temporal modeling module to ensure synchronization between geometry and sensor signals.

[0010] Preferably, the multi-degree-of-freedom robotic arm is configured as a six-degree-of-freedom robotic arm, including multiple rotatable joints, multiple links of different lengths, and a cantilever. A piezoelectric actuator is installed at the end of the cantilever; an encoder and a corresponding servo motor are installed at the rotatable joint; and an accelerometer and a gyroscope are installed on the rotatable joint or the cantilever.

[0011] Preferably, the transfer mechanism includes a support frame, which is installed on an elevated platform, overhead bridge, or support pole along the railway line; The support frame is made of lightweight, high-strength materials, including carbon fiber or aluminum alloy.

[0012] Preferably, the camera module includes an industrial camera and an optical filter; The point cloud scanning component is configured as a laser structured light module or a lidar sensor.

[0013] A method for detecting wear on power transmission lines, applied to the aforementioned power transmission line wear detection system, comprising: The testing agency collects data from the overhead contact line, including images and point cloud data; Extract structural and intensity features from the data and perform information filtering; The edge computing platform processes the collected data, generates wear detection reports in real time, and uploads the data to the monitoring center via the communication module; When wear or abnormalities exceeding a preset threshold are detected, an early warning signal is issued, and timely measures are taken.

[0014] Preferably, the method for processing the collected data by the edge computing platform includes: Preprocess the acquired images and point cloud data, filtering and reducing noise; Feature extraction is performed on the preprocessed catenary images and point cloud information, including structural feature extraction and intensity feature extraction; Confidence scores guide the fusion of structural and strength features, resulting in a unified representation. Using a unified representation of feature integration as input, and through temporal modeling, the dynamic changes of the overhead contact line structure over time are constructed. The time-aware features output from the modeling are mapped into point-level deformation responses, and combined with the original point cloud, a visualized 3D cloud map is formed through color mapping; revealing the microstructural changes on the surface of the contact wire.

[0015] Compared with the prior art, the advantages of the present invention are: The transmission line wear detection system provided by this invention achieves real-time monitoring of railway catenary with an accuracy of 0.01mm through a transfer mechanism, a robotic arm for stable suspension and positioning of the detection mechanism, multi-degree-of-freedom adjustment, active vibration control, and efficient fusion of machine vision and point cloud data.

[0016] The catenary detection model integrates point cloud processing, temporal modeling, and multimodal fusion. By combining geometric information and sensor intensity information, it improves the expressive ability of catenary surface features. Through DeformTrack using the Transformer model, it enhances the temporal perception of minute deformations on the catenary surface. From raw data input to cloud map output, the entire process is automated without human intervention, achieving high-precision, low-latency wear detection and generating high-quality color 3D cloud maps. This provides accurate and real-time monitoring methods for railway maintenance, helping to improve the safety and reliability of the railway system. Attached Figure Description

[0017] The present invention will be further described below with reference to the accompanying drawings and embodiments: Figure 1 This is a schematic diagram illustrating the structural principle of the power transmission line wear detection system described in this invention. Figure 2 This is a schematic diagram of the overhead contact line detection model described in this invention; Figure 3 This is a flowchart of a method for wear detection using the contact wire detection model described in this invention. Figure 4 This is a schematic diagram illustrating the internal working principle of the overhead contact line detection model described in this invention. Detailed Implementation

[0018] The present invention will be further described in detail below with reference to specific embodiments: like Figure 1As shown, a power transmission line wear detection system includes a transfer mechanism suspended from the contact wire. The transfer mechanism carries the detection mechanism and moves it along the contact wire. A multi-degree-of-freedom robotic arm mounted on the transfer mechanism adjusts the detection angle of the detection mechanism. The detection mechanism is communicatively connected to a data processing device, transmitting the collected data to the data processing device. The data processing device includes an edge computing platform, a communication module, and an alarm module. The edge computing platform processes the collected data; if the obtained data is abnormal or exceeds a preset wear threshold, the alarm module issues a warning. The transfer mechanism includes a support frame, which is installed on elevated platforms, overhead bridges, or support poles along the railway line to ensure the stability of the entire system in complex environments. For example, pulleys are installed at the bottom of a liftable robotic arm, and these pulleys are secured to the contact wire without affecting its movement along the contact wire. (See Appendix) Figure 1 As shown in the diagram, the bottom transfer mechanism is configured as a support plate / frame, with rollers (or pulleys) installed at the bottom of the support frame. The robotic arm is fixed to the transfer mechanism (the robotic arm and transfer mechanism can be detachable separate structures or integrated structures). The support frame is made of lightweight, high-strength materials, such as carbon fiber or aluminum alloy, to ensure that the structure has sufficient rigidity while facilitating installation and maintenance.

[0019] The inspection unit includes a camera module and a point cloud scanning component, used to collect data information from the overhead contact line.

[0020] The camera module includes an industrial camera and a matching light filter to form a high-resolution machine vision system that captures minute wear marks on the surface of the contact wire.

[0021] The point cloud scanning component is configured as a laser structured light module or a lidar sensor to generate 3D point cloud data in real time. Through multi-view acquisition and precise calibration, a spatial resolution of 0.01mm is achieved.

[0022] The multi-degree-of-freedom robotic arm is configured as a six-degree-of-freedom arm, including multiple rotatable joints, multiple links of different lengths, and a cantilever, ensuring that the inspection mechanism can be precisely suspended above or near the contact wire. The robotic arm can be finely adjusted in the horizontal, vertical, and rotational directions to ensure that the camera and sensor components are aligned with the target area on the contact wire.

[0023] Specifically, an encoder and a corresponding servo motor are installed at the rotatable joint to perform high-precision servo control, achieve micron-level positioning adjustment, and ensure that the detection sensor can accurately locate and track minute changes in the contact wire.

[0024] A piezoelectric actuator is installed at the end of the cantilever, and an accelerometer and gyroscope are installed on the rotatable joint or cantilever to monitor the vibration of the transfer mechanism in real time. Through a closed-loop control system and electronic control feedback (such as using a piezoelectric actuator), vibration interference is automatically compensated to achieve active vibration resistance and ensure the accuracy and stability of the test data.

[0025] Regarding the physical acquisition platform of the system, high-resolution image acquisition and point cloud scanning technology, combined with high-precision positioning and active vibration resistance mechanism, are used to monitor the wear of the contact wire in real time and detect minute changes.

[0026] In terms of data processing, the system integrates multi-source data through image processing, deep learning, and point cloud registration algorithms to construct a high-precision 3D model, automatically identifying and quantifying the wear and tear of the overhead contact line. This high-precision 3D model is a contact line detection model built into an edge computing platform. It is a point cloud processing framework for detecting contact line deformation. By processing images and point cloud data of the contact line, it can detect minute deformations on the contact line surface in real time and accurately, achieving automated, high-precision, and real-time contact line deformation detection.

[0027] See appendix Figure 2 As shown, the model includes: a preprocessing module, a feature extraction module, a fusion module, and a change awareness module. The specific details are as follows: The preprocessing module preprocesses the images and point cloud data acquired from the aforementioned physical acquisition structure, filtering and reducing noise. The feature extraction module extracts features from the preprocessed contact network images and point cloud information, including structural feature extraction and intensity feature extraction. The fusion module guides the fusion of structural and intensity features based on confidence levels, and the integrated features generate a unified representation.

[0028] The temporal modeling module includes a multi-head attention unit, a feedback network, a residual structure, and a layer normalization unit. It is used to model the dynamic evolution of local structures over time, taking a unified representation of feature integration as input and outputting time-aware features. The change perception module maps the output of the temporal modeling module into point-level deformation responses, and combines them with the original point cloud to form a visualized 3D cloud map through color mapping, revealing the microstructural changes on the contact wire surface.

[0029] See appendix Figure 4 As shown, the process of wear detection using the overhead contact line inspection model is as follows: The Point-Intensity Pair Loader loads 3D coordinates (XYZ) and intensity (point cloud) in parallel from the acquired point cloud data, constructing point cloud fusion pairs as the basic input, i.e., multimodal input pairs, ensuring that geometric and intensity information are input simultaneously to maintain data integrity.

[0030] The GeoAlign & Denoiser module (GeoAlign module, normalization, denoising and ROI extraction module) performs coordinate normalization and denoising on point cloud data, and extracts regions of interest (ROI), achieving geometric standardization and ROI filtering, completing the unification of the coordinate system and the removal of redundant points, and improving the quality of downstream features.

[0031] PatchDecompNet divides the preprocessed point cloud into multiple local patches, enhancing the model's ability to perceive changes in local structure, allowing each local region to be processed independently, and reducing computational complexity.

[0032] In time series analysis, a patch refers to dividing continuous time series data into several local segments, each called a patch. These segments typically have a fixed length and optional degree of overlap. By analyzing the dynamic behavior of these local segments, we can better capture local patterns and global dependencies in the time series.

[0033] Each segment is fed into two sub-modules of the SparseFusion Encoder: the 3D convolutional module (SparseConv3D) and the multilayer sensing module (IntensityMLP). Sparse geometric features (i.e., structural features) are extracted in the SparseConv3D module, and intensity sensing features are extracted in the multilayer sensing module.

[0034] The features extracted from both methods are then fused using a confidence-guided fusion module to achieve multimodal fusion, capturing more feature information and improving the model's expressive power. Dynamically weighted merging of modal information enables feature integration and generates a unified representation at the fragment level.

[0035] The fused patch representation is fed into the DeformTrack module for sequence modeling. The DeformTrack module is a temporal modeling module based on a customized Transformer network (a Transformer-based temporal modeler) used to model the dynamic evolution of local structures over time, i.e., to capture the dynamic features of the contact wire surface changing over time, thereby improving the temporal accuracy of contact wire wear detection. The modeling process focuses on the cross-frame dynamic behavior of the patch, and its structure includes multi-head attention units, a feedforward network, a residual structure, and layer normalization units (LayerNorm).

[0036] The time-aware features output from the modeling are directly mapped into point-level deformation responses. These responses, combined with the original point cloud, are used to generate a visualized color 3D cloud map in the HeatCloud Renderer through color mapping. This reveals the microstructural changes on the contact wire surface and enables spatial visualization of the contact wire surface deformation.

[0037] The overhead contact line detection model has point-to-frame level structural perception capabilities, integrating sparse geometry, sensor strength, and temporal dynamic information, enabling high-precision detection and visualization of minute deformations in the overhead contact line structure; it features a compact structure, decoupled modules, and strong adaptability.

[0038] The built-in edge computing platform processes the collected data quickly, generates wear detection reports in real time, and uploads the data to the monitoring center via wireless network or dedicated communication module. The integrated intelligent judgment algorithm immediately issues an early warning signal when wear or abnormalities exceeding the preset threshold are detected, assisting railway maintenance departments in taking timely measures.

[0039] This detection system achieves real-time monitoring of railway catenary with an accuracy of 0.01mm through stable suspension positioning, multi-degree-of-freedom adjustment, active vibration control, and efficient fusion of machine vision and point cloud data.

[0040] The catenary detection model integrates point cloud processing, temporal modeling, and multimodal fusion. By combining geometric information and sensor intensity information, it improves the expressive ability of catenary surface wear characteristics. Through DeformTrack using the Transformer model, it enhances the temporal perception of minute deformations on the catenary surface. From raw data input to cloud map output, the entire process is automated without manual intervention, achieving high-precision, low-latency wear detection and generating high-quality color 3D cloud maps. This provides accurate and real-time monitoring methods for railway maintenance, helping to improve the safety and reliability of the railway system.

[0041] The above embodiments are merely illustrative of the technical concept and features of the present invention, intended to enable those skilled in the art to understand the content of the present invention and implement it accordingly, and should not be construed as limiting the scope of protection of the present invention. It will be apparent to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects. The scope of the present invention is defined by the appended claims rather than the foregoing description, and thus all changes falling within the meaning and scope of the equivalents of the claims are intended to be included within the present invention.

Claims

1. A power transmission line wear detection system, characterized in that, It includes a transfer mechanism, which is suspended and connected to the contact wire. The transfer mechanism is used to support the detection mechanism to move along the contact wire. The detection angle of the detection mechanism is adjusted by a multi-degree-of-freedom robotic arm set on the transfer mechanism. The testing equipment includes a camera module and a point cloud scanning component, used to collect data information from the overhead contact line; The detection mechanism is connected to the data processing device and transmits the collected data information to the data processing device. The data processing device includes an edge computing platform, a communication module, and an alarm module. The edge computing platform has a built-in contact network detection model for processing the collected data. When the obtained data is abnormal or exceeds the preset wear threshold, the alarm module will issue a warning.

2. The power transmission line wear detection system according to claim 1, characterized in that, The overhead contact line detection model includes: The preprocessing module preprocesses the acquired images and point cloud data, filtering and reducing noise. The feature extraction module performs feature extraction on the preprocessed catenary image and point cloud information, including structural feature extraction and intensity feature extraction; The fusion module guides the fusion of structural and intensity features based on confidence levels, and the integrated features generate a unified representation. The temporal modeling module is used to model the dynamic evolution of local structures over time. It takes a unified representation of feature integration as input and outputs time-aware features. The change perception module maps the output of the time-series modeling module into point-level deformation responses. Combined with the original point cloud, it forms a visualized 3D cloud map through color mapping, revealing the microstructural changes on the surface of the contact wire.

3. The power transmission line wear detection system according to claim 2, characterized in that, The temporal modeling module includes a multi-head attention unit, a feedback network, a residual structure, and a layer normalization unit.

4. The power transmission line wear detection system according to claim 2, characterized in that, The unified representation of feature integration is as follows: three-dimensional coordinates and intensities extracted from the original input point cloud data, the three-dimensional coordinates and intensities are constructed into multimodal input pairs, and input into the temporal modeling module to ensure synchronization between geometry and sensor signals.

5. The power transmission line wear detection system according to claim 1, characterized in that, The multi-degree-of-freedom robotic arm is configured as a six-degree-of-freedom robotic arm, including multiple rotatable joints, multiple links of different lengths, and a cantilever. A piezoelectric actuator is installed at the end of the cantilever; an encoder and a corresponding servo motor are installed at the rotatable joint; and an accelerometer and a gyroscope are installed on the rotatable joint or the cantilever.

6. The power transmission line wear detection system according to claim 1, characterized in that, The transfer mechanism includes a support frame, which is installed on an elevated platform, overhead bridge, or support pole along the railway line. The support frame is made of lightweight, high-strength materials, including carbon fiber or aluminum alloy.

7. The power transmission line wear detection system according to claim 1, characterized in that, The camera module includes an industrial camera and an optical filter; The point cloud scanning component is configured as a laser structured light module or a lidar sensor.

8. A method for detecting wear on power transmission lines, applied to the power transmission line wear detection system as described in any one of claims 1-7, characterized in that, include: The testing agency collects data from the overhead contact line, including images and point cloud data; Extract structural and intensity features from the data and perform information filtering; The edge computing platform processes the collected data, generates wear detection reports in real time, and uploads the data to the monitoring center via the communication module; When wear or abnormalities exceeding a preset threshold are detected, an early warning signal is issued, and timely measures are taken.

9. The method for detecting wear on power transmission lines according to claim 8, characterized in that, The methods used by edge computing platforms to process collected data include: Preprocess the acquired images and point cloud data, filtering and reducing noise; Feature extraction is performed on the preprocessed catenary images and point cloud information, including structural feature extraction and intensity feature extraction; Confidence scores guide the fusion of structural and strength features, resulting in a unified representation. Using a unified representation of feature integration as input, and through temporal modeling, the dynamic changes of the overhead contact line structure over time are constructed. The time-aware features output from the modeling are mapped into point-level deformation responses, and combined with the original point cloud, a visualized 3D cloud map is formed through color mapping; revealing the microstructural changes on the surface of the contact wire.

Images