Method and system for detecting contact rail geometry and surface defects

Abstract

This invention provides a method and system for detecting contact rail geometric parameters and surface defects, applicable to the field of intelligent detection technology for urban rail transit operation and maintenance and rail power supply systems. The method includes controlling a laser ranging sensor and image acquisition device mounted on an inspection trolley to scan the contact rail, simultaneously acquiring geometric data and surface image data; calculating the contact rail geometric parameters based on the geometric data; identifying the types of defects on the contact rail surface and locating their positions using an improved deep learning model based on the surface image data; and outputting detection results including geometric parameters and surface defect information. In this way, automated, quantitative, and comprehensive detection of key geometric parameters and typical surface defects of the contact rail can be achieved, improving the accuracy and efficiency of contact rail operating status assessment.

Description

Technical Field

[0001] This invention relates to the field of intelligent detection technology for urban rail transit operation and maintenance and rail power supply systems, and particularly to a method and system for detecting contact rail geometric parameters and surface defects. Background Technology

[0002] With the rapid development of urban rail transit, the scale and operating mileage of subway lines are constantly increasing, leading to increasingly higher requirements for the safety and reliability of the power supply system. As a key component of the subway power supply system, the operating status of the contact rail directly affects the stability of train current collection and the safe and reliable operation of the subway system.

[0003] During long-term operation, the contact rail is prone to various deterioration problems due to factors such as train vibration, current-carrying friction, environmental erosion, and installation errors. These problems mainly fall into two categories: geometric parameter deviation and surface defects. Geometric parameter deviation is mainly manifested as exceeding the limits of key parameters such as guide height and pull-out value, which may lead to unstable current collection or even current collection interruption. Surface defects include cracks, ablation, pits, rust, and scratches. In severe cases, they can exacerbate current-carrying wear, reduce the service life of the contact rail, and even cause power supply safety hazards.

[0004] Currently, the methods for detecting the condition of contact rails are relatively limited, and engineering projects still rely mainly on manual inspections and experience-based judgments. This often results in low efficiency, high labor intensity, strong subjectivity, and difficulty in achieving continuous and quantitative assessments. Existing detection methods tend to focus on a single indicator or localized defects, making it difficult to conduct systematic and coordinated comprehensive detection of both contact rail geometric parameters and surface defects simultaneously. This fails to meet the actual needs of high-frequency, refined, and intelligent operation and maintenance management for subway lines.

[0005] Therefore, there is an urgent need to propose a technical solution that can uniformly and systematically detect the geometric parameters and surface defects of the contact rail, so as to achieve efficient acquisition and objective evaluation of the contact rail's operating status and provide reliable technical support for the safe operation and maintenance decision-making of the subway power supply system. Summary of the Invention

[0006] This invention provides a method and system for detecting contact rail geometric parameters and surface defects, solving the technical problems of existing contact rail detection mainly relying on manual inspection, having limited detection methods, low efficiency, and difficulty in systematically evaluating geometric parameters and surface defects.

[0007] According to a first aspect of the present invention, a method for detecting contact rail geometry parameters and surface defects is provided. The method includes: The laser rangefinder and image acquisition device installed on the inspection trolley are controlled to scan the contact rail and simultaneously acquire the geometric data and surface image data of the contact rail. Based on the geometric data, the geometric parameters of the contact rail are calculated; Based on the surface image data, an improved deep learning model is used to identify the types of defects on the contact rail surface and locate their positions. The output includes the detection results, which include the geometric parameters and surface defect information.

[0008] In addition to the aspects and any possible implementations described above, a further implementation is provided in which acquiring the geometric data of the contact rail includes: Before scanning, the inspection trolley is controlled to move the laser ranging sensor to a preset reference position; The laser rangefinder is controlled to perform a lateral scan in a direction perpendicular to the extension of the contact rail, and the distance data between the laser rangefinder and the contact rail surface is continuously collected. Simultaneously acquire horizontal displacement data of the laser rangefinder during the scanning process; Specifically, based on the comparison result between the distance data and the preset threshold, the edge position of the contact rail is dynamically determined, and the distance data and horizontal displacement data are recorded as effective geometric data within the effective edge interval.

[0009] In addition to the aspects described above and any possible implementations, a further implementation is provided, wherein dynamically determining the edge position of the contact rail based on the comparison result of the distance data and a preset threshold includes: When the difference between the distance data and the reference distance exceeds the preset threshold for the first time, it is determined that the outer edge of the contact rail has been reached and valid data is recorded; wherein, the reference distance is the distance relative to the surface of the running rail measured by the laser rangefinder at the reference position; When the difference exceeds the preset threshold again, it is determined that the inner edge of the contact rail has been reached and recording stops.

[0010] In addition to the aspects described above and any possible implementations, a further implementation is provided in which calculating the geometric parameters of the contact rail based on the geometric data includes: Based on valid distance data and the vertical distance between the laser rangefinder sensor reference position and the running track surface, multiple intermediate values ​​of guide height measurement are calculated, and the median of the multiple intermediate values ​​of guide height measurement is taken as the final guide height parameter. Based on valid horizontal displacement data and the horizontal distance between the laser rangefinder sensor reference position and the inner edge of the running track, multiple pull-out measurement intermediate values ​​are calculated, and the median of the multiple pull-out measurement intermediate values ​​is taken as the final pull-out value parameter.

[0011] In addition to the aspects and any possible implementations described above, a further implementation is provided, wherein the method further includes: Based on the guide height parameters at multiple consecutive detection locations, the maximum deviation, average absolute deviation, and root mean square error between these parameters and the preset reference values ​​are calculated to characterize the flatness of the contact rail.

[0012] In addition to the aspects and any possible implementations described above, a further implementation is provided, wherein the improved deep learning model is an improved YOLOv11 object detection network, comprising: A high-resolution feature branch is introduced into the feature extraction part of the network to enhance shallow features, and the enhanced features are fused with deep semantic features to improve the detection capability of small-scale diseases. A direction-aware enhancement module is introduced after a certain layer of the feature pyramid network to perform feature statistics and weighted reconstruction on the fused feature map in different directions, so as to improve the continuity of identification of slender diseases. An adaptive regression mechanism is adopted to dynamically adjust the weight allocation of the bounding box regression loss according to the scale of the disease target during the training process, so as to improve the localization accuracy of small-scale diseases.

[0013] According to a second aspect of the present invention, a system for detecting contact rail geometry parameters and surface defects is provided. The system includes: The acquisition module is used to control the laser rangefinder and image acquisition device installed on the inspection trolley to scan the contact rail and simultaneously acquire the geometric data and surface image data of the contact rail. A calculation module is used to calculate the geometric parameters of the contact rail based on the geometric data; The identification module is used to identify the type of damage on the contact rail surface and locate its position based on the surface image data using an improved deep learning model. The output module is used to output the detection results, including the geometric parameters and surface defect information.

[0014] According to a third aspect of the present invention, an electronic device is provided. The electronic device includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the program to implement the method described above.

[0015] According to a fourth aspect of the invention, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the method as described in the first and / or second aspects of the invention.

[0016] This invention utilizes an inspection trolley equipped with a laser ranging sensor and an image acquisition device to simultaneously acquire contact rail geometric data and surface images. By combining laser ranging with threshold determination of edge positions, median statistics, and deviation calculation, key geometric parameters such as guide height, pull-out value, and flatness are accurately obtained. Simultaneously, based on an improved YOLOv11 network that incorporates high-resolution feature branches, a direction perception enhancement module, and an adaptive regression mechanism, various surface defects, including small-scale and slender ones, are efficiently identified. This achieves automated comprehensive detection of contact rail geometric parameters and surface defects, effectively solving the problems of low efficiency, high labor intensity, strong subjectivity, and difficulty in systematic evaluation using single detection methods in traditional manual inspections. It significantly improves detection accuracy and stability, providing comprehensive and reliable technical support for the safe operation, refined operation and maintenance, and maintenance decision-making of subway power supply systems.

[0017] It should be understood that the description in the Summary of the Invention is not intended to limit the key or essential features of the embodiments of the present invention, nor is it intended to restrict the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0018] The above and other features, advantages, and aspects of the various embodiments of the present invention will become more apparent from the accompanying drawings and the following detailed description. The drawings are provided for a better understanding of the invention and are not intended to limit the scope of the invention. In the drawings, the same or similar reference numerals denote the same or similar elements, wherein: Figure 1 A flowchart of a method for detecting contact rail geometry parameters and surface defects according to an embodiment of the present invention is shown.

[0019] Figure 2 A schematic diagram illustrating the identification of surface defects on a contact rail according to an embodiment of the present invention is shown.

[0020] Figure 3 A block diagram of a contact rail geometry and surface defect detection system according to an embodiment of the present invention is shown.

[0021] Figure 4 A block diagram of an exemplary electronic device capable of implementing embodiments of the present invention is shown. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] Furthermore, the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0024] This invention utilizes an inspection trolley equipped with a laser ranging sensor and an image acquisition device to simultaneously acquire contact rail geometric data and surface images. By combining laser ranging with threshold determination of edge positions, median statistics, and deviation calculation, key geometric parameters such as guide height, pull-out value, and flatness are accurately obtained. Simultaneously, based on an improved YOLOv11 network that incorporates high-resolution feature branches, a direction perception enhancement module, and an adaptive regression mechanism, various surface defects, including small-scale and slender ones, are efficiently identified. This achieves automated comprehensive detection of contact rail geometric parameters and surface defects, effectively solving the problems of low efficiency, high labor intensity, strong subjectivity, and difficulty in systematic evaluation using single detection methods in traditional manual inspections. It significantly improves detection accuracy and stability, providing comprehensive and reliable technical support for the safe operation, refined operation and maintenance, and maintenance decision-making of subway power supply systems.

[0025] Figure 1 A flowchart of a method 100 for detecting contact rail geometry parameters and surface defects according to an embodiment of the present invention is shown. Figure 1 As shown, method 100 includes: S101 controls the laser rangefinder and image acquisition device installed on the inspection trolley to scan the contact rail and simultaneously acquire the geometric data and surface image data of the contact rail.

[0026] In some embodiments, the inspection trolley serves as the carrier and mobile platform for the entire inspection method. Its structure is adapted to the track gauge and laying pattern of the subway track, enabling stable travel along the subway line. It can perform fixed-point stopping operations or continuous travel at a preset speed, meeting the different working conditions required for detailed inspection of key sections and rapid line-level inspection. The trolley body integrates a mounting base, drive module, stabilization and shock absorption mechanism, and power supply unit. The mounting base is used to fix the laser rangefinder, slide table, image acquisition device, and cable displacement sensor, ensuring accurate and stable installation of each device. The laser rangefinder is the core sensing component for measuring the geometric parameters of the contact rail. It accurately measures the distance between the sensor and the contact rail surface by emitting a laser beam and receiving reflected signals. This sensor is mounted on the slider of the slide table and moves synchronously with the slider, performing a lateral scanning action perpendicular to the direction of the contact rail extension. The slide table is the motion drive mechanism for the laser rangefinder. It is mounted on the mounting base of the inspection trolley, fixedly connected to the slider of the laser rangefinder, and establishes a displacement transmission relationship with the cable displacement sensor through a pull wire. Under control, the slide table drives the slider to reciprocate along a direction perpendicular to the contact rail, enabling lateral scanning from the outer edge to the inner edge of the contact rail. A cable displacement sensor is fixedly mounted at one end of the slide table, its cable connected to the slider containing the laser rangefinder, forming a displacement transmission mechanism for synchronously acquiring horizontal displacement data during the lateral scanning process. The sensor's measurement accuracy matches the slide table's motion accuracy, allowing real-time capture of slider position changes and generating a displacement sequence corresponding to the laser distance data. The image acquisition device, a core component for contact rail surface defect detection, acquires high-definition images of the contact rail surface. Mounted on the inspection trolley's base, with its lens facing the contact rail surface to ensure complete coverage of the inspection area, the device can perform continuous imaging whether the inspection trolley is moving or stationary. The acquired raw images include information on surface defects such as cracks, ablation, pits, rust, and scratches, as well as background interference.

[0027] In some embodiments, acquiring the geometric data of the contact rail includes: Before scanning, the inspection trolley is controlled to move the laser rangefinder to a preset reference position; The laser rangefinder is controlled to perform a lateral scan in a direction perpendicular to the extension of the contact rail, continuously collecting distance data between itself and the surface of the contact rail. Simultaneously acquire horizontal displacement data from the laser rangefinder during the scanning process; Specifically, based on the comparison results between distance data and preset thresholds, the edge position of the contact rail is dynamically determined, and the distance data and horizontal displacement data are recorded as valid geometric data within the effective edge interval.

[0028] In some embodiments, dynamically determining the edge position of the contact rail based on a comparison between distance data and a preset threshold includes: When the difference between the distance data and the reference distance exceeds the preset threshold for the first time, it is determined that the outer edge of the contact rail has been reached and valid data is recorded. The reference distance is the distance relative to the surface of the travel rail measured by the laser rangefinder at the reference position. When the difference exceeds the preset threshold again, it is determined that the inner edge of the contact rail has been reached and recording stops.

[0029] Specifically, before performing a lateral scan of the contact rail, the initial calibration of the reference position must be completed first. The inspection trolley is then controlled to travel to the target detection area. In fixed-point detection mode, the trolley must precisely stop at the target position; in moving detection mode, the trolley maintains a preset constant speed and prepares for data acquisition according to the trigger conditions. Subsequently, relying on the mounting base on the inspection trolley, the laser rangefinder sensor is pre-fixed to the slider of the sliding table, and the cable displacement sensor is fixedly set at one end of the sliding table, establishing a rigid displacement transmission relationship with the slider through a cable to ensure that the sensor displacement and the slider movement are completely synchronized. The sliding table is controlled to drive the slider, moving the laser rangefinder sensor to a preset reference position, which is the pre-calibrated measurement starting reference point.

[0030] Furthermore, during data acquisition, the laser ranging data needs to be analyzed in real time to dynamically determine the edge position of the contact rail and achieve accurate screening of valid data. The reference distance is the distance relative to the surface of the travel rail measured by the laser ranging sensor at the reference position. During scanning, the difference between the real-time acquired distance data and the reference distance is calculated. When this difference first exceeds a preset threshold, such as 20mm, it indicates that the laser beam has been projected from the background area onto the contact rail surface. At this point, the laser measurement position is determined to have reached the outer edge of the contact rail, and the valid data recording process is initiated, incorporating subsequently acquired distance data and corresponding horizontal displacement data into the valid dataset. As the slide continues to move, the laser ranging sensor gradually scans the contact rail body area. When the difference between the distance data and the reference distance again exceeds the preset threshold, it indicates that the laser beam has detached from the contact rail surface and has reached the inner edge of the contact rail. At this point, a command is immediately issued to stop the slide and drive the slider to reset the laser ranging sensor to the initial reference position, preparing for the next scan. This edge detection logic can accurately filter out valid geometric data that only covers the contact rail body, eliminate interference from the surrounding environment, and ensure the accuracy of subsequent parameter calculations.

[0031] S102, based on geometric data, calculates the geometric parameters of the contact rail.

[0032] In some embodiments, calculating the geometric parameters of the contact rail based on geometric data includes: Based on the effective distance data and the vertical distance between the laser rangefinder sensor reference position and the running track surface, multiple intermediate values ​​of guide height measurement are calculated, and the median of the multiple intermediate values ​​of guide height measurement is taken as the final guide height parameter. Based on valid horizontal displacement data and the horizontal distance between the laser rangefinder sensor reference position and the inner edge of the running track, multiple pull-out measurement intermediate values ​​are calculated, and the median of the multiple pull-out measurement intermediate values ​​is taken as the final pull-out value parameter.

[0033] In some embodiments, calibration instruments such as rulers and laser rangefinders are used to calibrate and measure the initial geometric parameters of the sensor reference position relative to the adjacent running track surface, such as the vertical distance and the horizontal distance relative to the inner edge of the running track. The measurement is performed three times consecutively and the average value is taken as the final calibration result.

[0034] In some embodiments, the first [parameter] on the cross-section of the contact rail Vertical distance from the point to the adjacent running track surface The calculation formula is:

[0035] in, Let be the distance from the laser rangefinder to the contact rail surface measured when scanning point i. This represents the initial vertical distance of the laser rangefinder sensor relative to the adjacent travel rail surface at the reference position. High Conductivity H The calculation formula is:

[0036] Furthermore, the first cross-section on the contact rail The horizontal distance from the point to the inner edge of the adjacent running track The calculation formula is:

[0037] in, This refers to the initial horizontal distance parameter of the laser rangefinder sensor at its reference position relative to the inner edge of the adjacent travel track. For the laser rangefinder to scan to the first At the specified point, horizontal displacement data is synchronously collected by the rope displacement sensor; Pull out value L The calculation formula is:

[0038] Multiple samples collected from the same location and By performing statistical analysis and taking the median, the guide height parameter at the detection location is finally obtained. and pull-out value parameters .

[0039] In some embodiments, based on the guide height parameters at multiple consecutive detection locations, the maximum deviation, average absolute deviation, and root mean square error between these parameters and a preset reference value are calculated to characterize the flatness of the contact rail.

[0040] Specifically, the maximum deviation value Mean absolute deviation and root mean square error This can be expressed by the following formula:

[0041]

[0042]

[0043] in, This is the design reference value for the contact rail guide height, which is generally 200mm, and N is the number of measurement points.

[0044] S103, based on surface image data, uses an improved deep learning model to identify the types of defects on the contact rail surface and locate their positions.

[0045] In some embodiments, the improved deep learning model is an improved YOLOv11 object detection network, comprising: A high-resolution feature branch is introduced into the feature extraction part of the network to enhance shallow features, and the enhanced features are fused with deep semantic features to improve the detection capability of small-scale diseases. A direction-aware enhancement module is introduced after a certain layer of the feature pyramid network to perform feature statistics and weighted reconstruction on the fused feature map in different directions, so as to improve the continuity of identification of slender diseases. An adaptive regression mechanism is adopted to dynamically adjust the weight allocation of the bounding box regression loss according to the scale of the disease target during the training process, so as to improve the localization accuracy of small-scale diseases.

[0046] In some embodiments, an image acquisition device is used to acquire images of the contact rail surface, and preprocessing operations such as cropping, scale normalization, and brightness adjustment are performed on the images to enhance the features of the defects and reduce background interference. The preprocessed contact rail surface images are then input into an improved YOLOv11 target detection network, and multi-scale feature information is extracted through the network's feature extraction structure to characterize the spatial morphological features of the contact rail surface defects.

[0047] In some embodiments, a high-resolution feature branch is derived from the shallow feature output of the backbone network. While maintaining high spatial resolution, this high-resolution feature branch enhances shallow texture and edge information through at least one convolutional layer and a feature aggregation module to obtain small-target enhanced features. These small-target enhanced features are then used as high-resolution input in subsequent multi-scale feature fusion to reduce the loss of small-scale disease features caused by downsampling. The output of the high-resolution feature branch is concatenated or fused with the upsampled mid-to-high-level semantic features at the feature fusion node. The fused features are used to generate detection feature maps for small-scale disease targets, thereby improving the recall rate and detection stability of small-target diseases such as cracks, scratches, and minor ablation.

[0048] In some embodiments, the direction-aware feature enhancement module performs feature statistics and modeling on the input feature map in different directions, including both the track-direction and the vertical direction. Based on the direction statistics results, it generates direction weight coefficients and performs direction-related weighted reconstruction on the input feature map to enhance the response intensity of slender defects in their main extension direction. The direction-aware feature enhancement module is executed on the high-resolution detection feature map and / or medium-resolution detection feature map after multi-scale feature fusion, to suppress interference from metal textures, reflections, and complex backgrounds on defect identification on the contact rail surface while preserving details of small targets, thereby improving the detection consistency and boundary continuity of directional defects such as cracks and scratches.

[0049] In some embodiments, an adaptive weighting mechanism related to the target scale is introduced during bounding box regression. This mechanism dynamically adjusts the weight allocation of regression error terms based on the scale relationship between the predicted and labeled bounding boxes. This increases the weight of the center position error term and / or the width-to-height consistency error term when the target scale is small, thereby reducing bounding box jitter for small targets and improving localization accuracy under high overlap thresholds. The adaptive localization regression mechanism is used to update network parameters during the training phase and achieves accurate localization of small-scale disease targets during the inference phase by obtaining more stable bounding box outputs.

[0050] S104 outputs detection results including geometric parameters and surface defect information.

[0051] Specifically, it outputs the detection and measurement results of surface defects of the contact rail, and can be integrated with the detection results of geometric parameters for use in contact rail operation status assessment and maintenance decisions.

[0052] The following detailed description, with reference to specific embodiments, illustrates a method 100 for detecting contact rail geometric parameters and surface defects provided by this invention: Contact rail geometric parameter inspection mainly includes measuring guide height, pull-out value, and flatness parameters. Surface defects mainly include cracks and ablation. The inspection equipment is installed on an inspection trolley that can travel along the subway track. It uses a laser rangefinder, image acquisition device, displacement sensor, and sliding table mechanism to collaboratively complete geometric parameter and image acquisition. Depending on the inspection conditions, geometric parameter inspection can be carried out using either a fixed-point inspection method or a moving inspection method.

[0053] Example 1: Fixed-point detection method.

[0054] In the fixed-point inspection method, the inspection trolley stops at the target inspection position and completes the precise measurement of the contact rail geometric parameters at that position, which is suitable for accurate inspection of key sections or suspected abnormal sections.

[0055] After the inspection trolley stops at the position to be measured, the laser rangefinder sensor is installed on the slider and connected to the slide table. The pull-string displacement sensor is fixed to one end of the slide table and a displacement correlation is established between the pull-string and the slider. The laser rangefinder sensor is controlled to move to the preset reference position to obtain its initial geometric relationship parameters relative to the adjacent travel rail surface and the inner edge of the travel rail, and the measurement threshold for determining the contact rail edge position is set to 20mm.

[0056] The control slide moves perpendicular to the extension direction of the contact rail, causing the laser rangefinder to scan from the outer edge to the inner edge of the contact rail. During the scanning process, laser rangefinder data is continuously acquired. When the difference between the measured value and the reference value first exceeds a threshold of 20mm, it is determined that the laser measurement position has reached the outer edge of the contact rail, and valid data is recorded. When the difference again exceeds the threshold of 20mm, it is determined that the laser measurement position has reached the inner edge of the contact rail, and the control slide stops moving.

[0057] Based on the recorded laser ranging data, the vertical distance between the contact rail and the running rail surface at the detection point is calculated by combining the geometric relationship parameters of the reference position. The results of multiple measurements are statistically processed, and the median is taken as the guide height value at the detection point.

[0058] While the guide height measurement is being completed, the horizontal displacement data of the slider is simultaneously collected by the pull-wire displacement sensor during the movement of the slide table. Based on the measurement results of the displacement sensor and the horizontal geometric relationship between the reference position of the laser rangefinder sensor and the inner edge of the travel rail, the horizontal offset distance of the contact rail at the detection point is calculated, and the median of multiple measurement results is taken to obtain the pull-out value at the detection point.

[0059] Statistical analysis was performed on the guide height values ​​obtained at multiple fixed detection locations. The flatness of the contact rail along the track direction was characterized by calculating the maximum deviation, average absolute deviation, and root mean square error.

[0060] Example 2: Motion detection method.

[0061] In the mobile inspection method, the inspection trolley travels continuously along the subway track at a preset speed, and collects geometric parameters simultaneously during the journey, which is suitable for rapid line-level inspection.

[0062] During the inspection vehicle's movement, the laser ranging sensor remains in the working position, and the slide performs a lateral scanning action according to a preset cycle or trigger conditions to continuously acquire laser ranging data and displacement sensor data at different mileage locations.

[0063] The continuously collected data is segmented according to mileage or time sequence. Within each sampling interval, the same edge determination logic and parameter calculation method as the fixed-point detection method are used to obtain the guide height value and pull-out value of the corresponding segment.

[0064] The guide height values ​​obtained continuously along the line direction are statistically segmented, and the flatness of the contact rail is comprehensively evaluated by indicators such as maximum deviation, average absolute deviation and root mean square error, thereby realizing the geometric state analysis at the line level.

[0065] Example 3: Detection of surface defects on contact rails.

[0066] During the inspection trolley's movement or stationary state, the image acquisition device continuously images the contact rail surface to obtain raw image data. Preprocessing operations such as cropping, scale normalization, brightness adjustment, and noise suppression are then performed on the images to enhance defect characteristics and reduce background interference.

[0067] The preprocessed image is input into the backbone structure of the target detection network for feature extraction, and a high-resolution feature branch is derived from the shallow feature output to preserve the texture and edge information of small-scale defects on the contact rail surface. A direction-aware feature enhancement module is introduced into the fused detection feature map. By statistically modeling and weighted reconstruction of features in different directions, the response of slender defects such as cracks and scratches in their main extension direction is enhanced, while suppressing the interference of metallic reflection and complex background on the detection results.

[0068] A small-target adaptive positioning regression strategy is employed in the detection head. The regression constraint weights are dynamically adjusted based on the scale of the defect target, improving the boundary positioning accuracy of small-scale defect targets. The system outputs the defect category, spatial location, and corresponding confidence score, enabling automatic identification of surface defects on the contact rail. Figure 2 As shown.

[0069] According to embodiments of the present invention, automated integrated detection of contact rail geometric parameters and surface defects is realized, effectively solving the problems of low efficiency, strong subjectivity, and single detection in traditional manual inspection, significantly improving detection accuracy and efficiency, and providing reliable technical support for the safe operation and refined maintenance of subway power supply systems.

[0070] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that the present invention is not limited to the described order of actions, because according to the present invention, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily essential to the present invention.

[0071] The above is an introduction to the method embodiments. The following describes the present invention further through device embodiments.

[0072] Figure 3 A block diagram of a contact rail geometry and surface defect detection system 300 according to an embodiment of the present invention is shown. Figure 3 As shown, system 300 includes: The acquisition module 301 is used to control the laser ranging sensor and image acquisition device installed on the inspection trolley to scan the contact rail and simultaneously acquire the geometric data and surface image data of the contact rail. The calculation module 302 is used to calculate the geometric parameters of the contact rail based on the geometric data; The identification module 303 is used to identify the type of defects on the contact rail surface and locate their positions based on surface image data using an improved deep learning model. Output module 304 is used to output the detection results, including geometric parameters and surface defect information.

[0073] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the described module can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0074] According to embodiments of the present invention, the present invention also provides an electronic device, a readable storage medium, and a computer program product.

[0075] Figure 4 A schematic block diagram of an electronic device 400 that can be used to implement embodiments of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0076] Electronic device 400 includes a computing unit 401, which can perform various appropriate actions and processes according to a computer program stored in ROM 402 or a computer program loaded into RAM 403 from storage unit 408. RAM 403 can also store various programs and data required for the operation of electronic device 400. The computing unit 401, ROM 402, and RAM 403 are interconnected via bus 404. I / O interface 405 is also connected to bus 404.

[0077] Multiple components in electronic device 400 are connected to I / O interface 405, including: input unit 406, such as keyboard, mouse, etc.; output unit 407, such as various types of displays, speakers, etc.; storage unit 408, such as disk, optical disk, etc.; and communication unit 409, such as network card, modem, wireless transceiver, etc. Communication unit 409 allows electronic device 400 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0078] The computing unit 401 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 401 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 401 performs the various methods and processes described above, such as method 100. For example, in some embodiments, method 100 may be implemented as a computer software program tangibly contained in a machine-readable medium, such as storage unit 408. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 400 via ROM 402 and / or communication unit 409. When the computer program is loaded into RAM 403 and executed by the computing unit 401, one or more steps of method 100 described above may be performed. Alternatively, in other embodiments, the computing unit 401 may be configured to perform method 100 by any other suitable means (e.g., by means of firmware).

[0079] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0080] The program code used to implement the methods of the present invention can be written in any combination of one or more programming languages. This program code can be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing device, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code can be executed entirely on the machine, partially on the machine, as a standalone software package partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0081] In the context of this invention, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. Machine-readable media can include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0082] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including voice input, speech input, or tactile input).

[0083] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0084] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact via communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other. Servers can be cloud servers, servers in distributed systems, or servers incorporating blockchain technology.

[0085] It should be understood that the various forms of processes shown above can be used to reorder, add, or delete steps. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this invention can be achieved, and this is not limited herein.

[0086] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for detecting contact rail geometric parameters and surface defects, characterized in that, include: The laser rangefinder and image acquisition device installed on the inspection trolley are controlled to scan the contact rail and simultaneously acquire the geometric data and surface image data of the contact rail. Based on the geometric data, the geometric parameters of the contact rail are calculated; Based on the surface image data, an improved deep learning model is used to identify the types of defects on the contact rail surface and locate their positions. The output includes the detection results, which include the geometric parameters and surface defect information.

2. The method according to claim 1, characterized in that, The acquisition of the contact rail's geometric data includes: Before scanning, the inspection trolley is controlled to move the laser ranging sensor to a preset reference position; The laser rangefinder is controlled to perform a lateral scan in a direction perpendicular to the extension of the contact rail, and the distance data between the laser rangefinder and the contact rail surface is continuously collected. Simultaneously acquire horizontal displacement data of the laser rangefinder during the scanning process; Specifically, based on the comparison result between the distance data and the preset threshold, the edge position of the contact rail is dynamically determined, and the distance data and horizontal displacement data are recorded as effective geometric data within the effective edge interval.

3. The method according to claim 2, characterized in that, The dynamic determination of the contact rail edge position based on the comparison result of the distance data and the preset threshold includes: When the difference between the distance data and the reference distance exceeds the preset threshold for the first time, it is determined that the outer edge of the contact rail has been reached and valid data is recorded; wherein, the reference distance is the distance relative to the surface of the running rail measured by the laser rangefinder at the reference position; When the difference exceeds the preset threshold again, it is determined that the inner edge of the contact rail has been reached and recording stops.

4. The method according to claim 3, characterized in that, The calculation of the geometric parameters of the contact rail based on the geometric data includes: Based on valid distance data and the vertical distance between the laser rangefinder sensor reference position and the running track surface, multiple intermediate values ​​of guide height measurement are calculated, and the median of the multiple intermediate values ​​of guide height measurement is taken as the final guide height parameter. Based on valid horizontal displacement data and the horizontal distance between the laser rangefinder sensor reference position and the inner edge of the running track, multiple pull-out measurement intermediate values ​​are calculated, and the median of the multiple pull-out measurement intermediate values ​​is taken as the final pull-out value parameter.

5. The method according to claim 4, characterized in that, The method further includes: Based on the guide height parameters at multiple consecutive detection locations, the maximum deviation, average absolute deviation, and root mean square error between these parameters and the preset reference values ​​are calculated to characterize the flatness of the contact rail.

6. The method according to claim 1, characterized in that, The improved deep learning model is an improved YOLOv11 object detection network, including: A high-resolution feature branch is introduced into the feature extraction part of the network to enhance shallow features, and the enhanced features are fused with deep semantic features to improve the detection capability of small-scale diseases. A direction-aware enhancement module is introduced after a certain layer of the feature pyramid network to perform feature statistics and weighted reconstruction on the fused feature map in different directions, so as to improve the continuity of identification of slender diseases. An adaptive regression mechanism is adopted to dynamically adjust the weight allocation of the bounding box regression loss according to the scale of the disease target during the training process, so as to improve the localization accuracy of small-scale diseases.

7. A system for detecting contact rail geometric parameters and surface defects, characterized in that, include: The acquisition module is used to control the laser rangefinder and image acquisition device installed on the inspection trolley to scan the contact rail and simultaneously acquire the geometric data and surface image data of the contact rail. A calculation module is used to calculate the geometric parameters of the contact rail based on the geometric data; The identification module is used to identify the type of damage on the contact rail surface and locate its position based on the surface image data using an improved deep learning model. The output module is used to output the detection results, including the geometric parameters and surface defect information.

8. An electronic device, characterized in that, include: At least one processor; as well as A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method described in any one of claims 1-6.

9. A non-transitory computer-readable storage medium storing computer instructions, characterized in that, The computer instructions are used to cause the computer to perform the method described in any one of claims 1-6.

Images