Vehicle collector shoe detection method, device and equipment and storage medium

Abstract

The invention discloses a vehicle collector shoe detection method, device and equipment and a storage medium, is applied to vehicle collector shoe detection equipment comprising a collector shoe acquisition unit, a collector shoe image storage and recognition unit and data transmission equipment, and relates to the technical field of automatic detection. Comprising the following steps: generating a pulse signal by using magnetic steel in a collector shoe acquisition unit when a vehicle passes, and sending the pulse signal to an area-array camera in the collector shoe acquisition unit to acquire an image of a collector shoe to obtain a target collector shoe image; transmitting and storing the target collector shoe image to a collector shoe image storage and identification unit by using data transmission equipment; a deep neural network classifier in the collector shoe image storage and recognition unit is used for carrying out component positioning on the target collector shoe image to obtain a component positioning result, geometric dimension, inclination angle and surface defect detection is carried out on the component positioning result in sequence, and an abrasion value detection result, a crack detection result and a notch anomaly detection result are obtained; the efficiency of detecting the collector shoe is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of automatic detection, in particular to a vehicle pantograph detection method, device, equipment and storage medium. BACKGROUND

[0002] At present, in the long-term operation of the urban rail vehicle, the pantograph of the vehicle will be worn and aged, and if it is not timely maintained, it will cause damage to the pantograph and affect the operation of the vehicle. At present, the pantograph detection method is mainly manual, which has high operation intensity, high cost, low efficiency, poor real-time performance and high safety hazards.

[0003] From the above, how to improve the detection efficiency of the pantograph of the vehicle in the pantograph detection process of the vehicle is a problem to be solved at present. SUMMARY

[0004] Therefore, the purpose of the present application is to provide a vehicle pantograph detection method, device, equipment and storage medium, which can improve the detection efficiency of the pantograph of the vehicle in the pantograph detection process of the vehicle. The specific scheme is as follows:

[0005] In a first aspect, the present application provides a vehicle pantograph detection method applied to a vehicle pantograph detection device including a pantograph collection unit, a pantograph image storage and identification unit and a data transmission device, comprising:

[0006] The magnetic steel in the pantograph collection unit generates a pulse signal corresponding to the vehicle when the vehicle passes, and sends the pulse signal to the area array camera in the pantograph collection unit, so as to collect the image of the pantograph by using the area array camera and the light supplementing device in the pantograph collection unit, and obtain a target pantograph image;

[0007] The data transmission device connected between the pantograph collection unit and the pantograph image storage and identification unit transmits and stores the target pantograph image to the pantograph image storage and identification unit;

[0008] The deep neural network classifier in the pantograph image storage and identification unit is used to locate the components of the target pantograph image, and obtain a component positioning result, and an abnormal detection algorithm is used to detect the component positioning result in sequence in terms of geometric size, inclination angle and surface defect, and obtain respectively corresponding pantograph wear value detection result, crack detection result and gap abnormality detection result.

[0009] Optionally, the magnetic steel in the power shoe collection unit generates a pulse signal corresponding to the vehicle when the vehicle passes, and sends the pulse signal to the area array camera in the power shoe collection unit to collect the image of the power shoe with the light supplementing device in the power shoe collection unit, and obtain the target power shoe image, including:

[0010] The magnetic steel is arranged on both sides of the wheel track, and the magnetic steel generates a pulse signal corresponding to the vehicle when the vehicle passes the track, and then sends the pulse signal to the area array camera in the power shoe collection unit; the area array camera is a camera meeting the preset high-definition condition;

[0011] The light adjusting algorithm, lens aperture control algorithm and light supplementing device in the power shoe collection unit are used to adjust the image brightness under the preset strong backlight condition to obtain the target image brightness, then the area array camera is called and the image is captured from the top and side of the power shoe based on the target image brightness and the preset trigger and high-speed snapshot mode to obtain the to-be-processed power shoe image;

[0012] The ROI technology is used to analyze the image of the power shoe key area in the to-be-processed power shoe image to obtain the image analysis result, and the to-be-processed power shoe image is adjusted based on the image analysis result to obtain the target power shoe image; the target power shoe image includes the geometric features and surface state corresponding to the power shoe.

[0013] Optionally, the data transmission device connected between the power shoe collection unit and the power shoe image storage and identification unit is used to transmit and store the target power shoe image to the power shoe image storage and identification unit, including:

[0014] The data transmission device is constructed based on a plurality of optical fiber switching devices, and the data transmission device is connected with the power shoe collection unit and the power shoe image storage and identification unit; the data transmission device includes a gigabit switch and an optical fiber terminal device;

[0015] The target power shoe image collected by the power shoe collection unit is transmitted to the power shoe image storage and identification unit realized by an industrial computer through the data transmission device; the power shoe image storage and identification unit includes a data storage management server, which is used to run collection software, identification software and management software, and is used to perform functions of arrival and departure vehicle judgment, image collection, vehicle number correspondence, image calculation, data storage, data forwarding and alarm setting.

[0016] Optionally, the step of using a deep neural network classifier in the image storage and recognition unit of the power receiving boot to locate the target power receiving boot image, obtaining the component location result, and using an anomaly detection algorithm to sequentially detect the geometric dimensions, tilt angle, and surface defects of the component location result, obtaining the corresponding wear value detection result, crack detection result, and notch anomaly detection result of the power receiving boot, respectively, includes:

[0017] A target deep neural network classifier is obtained by adjusting the network weight parameters of the initial deep neural network classifier using a preset neural network algorithm and based on historical sample images; the historical sample images are historical sample images corresponding to the electric boot.

[0018] The target deep neural network classifier is used to locate components in the image of the target power receiving boot, obtaining component location results including category and location information corresponding to each component of the power receiving boot. Then, based on the component location results, an anomaly detection algorithm is used to sequentially detect the geometric dimensions, tilt angle, and surface defects of the power receiving boot, obtaining corresponding wear value detection results, crack detection results, and notch anomaly detection results for the power receiving boot, respectively. Among them, the geometric dimensions include the length and thickness of the power receiving boot; the tilt angle is the angle between the boundary contour line of the lower surface of the carbon slide plate and the upper edge line of the rail; the surface defects include scratches, burns, and chips on the power receiving boot.

[0019] Optionally, the step of using an anomaly detection algorithm to sequentially detect geometric dimensions, tilt angles, and surface defects in the component positioning results to obtain corresponding wear value detection results, crack detection results, and notch anomaly detection results for the power receiving shoe, including:

[0020] The target deep neural network classifier is used to locate the position of the electric boot corresponding to the image of the electric boot, and the geometric size is detected by the anomaly detection algorithm based on the position of the electric boot and the positioning result of the component to obtain the geometric size detection result;

[0021] Using edge contour extraction technology and based on the geometric size detection results, the first boundary contour feature value corresponding to the upper surface of the carbon slide plate in the power receiving shoe and the second boundary contour feature value corresponding to the lower surface of the carbon slide plate are determined. Based on the first boundary contour feature value and the second boundary contour feature value, the first boundary contour line is determined, and then the number of pixels on the first boundary contour line is determined.

[0022] Based on the number of pixels and the first boundary contour, the thickness and length of the power receiving boot corresponding to the power receiving boot are determined, and the wear value corresponding to the power receiving boot is determined based on the thickness and length of the power receiving boot. Then, it is determined whether the wear value is greater than a preset wear threshold to obtain the wear value detection result.

[0023] The edge contour feature value corresponding to the edge of the power receiving shoe on the rail is determined, and a second boundary contour line is constructed based on the second boundary contour feature value. Then, an edge contour line is constructed based on the edge contour feature value to determine the angle between the second boundary contour line and the edge contour line, and to determine whether the angle is greater than a preset angle threshold, so as to obtain the crack detection result.

[0024] The normal state interval is stripped using the dimension stripping technology to obtain the stripped result, and it is determined whether the area of ​​the unstripped interval corresponding to the stripped result is greater than a preset area threshold to obtain the gap anomaly detection result.

[0025] If the area of ​​the unpeeled region corresponding to the peeled result is greater than the preset area threshold, a defect alarm is triggered, and the defect is labeled as a scratch, burn, or chip based on its color and contour features.

[0026] Optionally, after using the anomaly detection algorithm to sequentially detect geometric dimensions, tilt angles, and surface defects in the component positioning results to obtain the corresponding wear value detection results, crack detection results, and notch anomaly detection results for the power receiving shoe, the method further includes:

[0027] If the wear value detection result indicates that the wear value is greater than the preset wear threshold, then it is determined that the power receiving boot has worn out;

[0028] If the crack detection result indicates that the included angle is greater than the preset included angle threshold, then it is determined that there is a crack in the power receiving shoe;

[0029] If the defect detection result indicates that the area of ​​the unpeeled section corresponding to the peeled result is greater than the preset area threshold, then it is determined that there is a defect in the power receiving shoe and a defect alarm is triggered. Then, based on the defect detection result, the defect color and contour features are determined to label the defect as a scratch, burn, or chip.

[0030] Secondly, this application provides a vehicle electric shock shoe detection device.

[0031] A vehicle electric shock boot detection device, applicable to a system including an electric shock boot acquisition unit, an electric shock boot image storage and recognition unit, and a data transmission device, includes:

[0032] The image acquisition module for the electric receiving boot is used to generate a pulse signal corresponding to the vehicle when the magnet in the electric receiving boot acquisition unit passes by, and send the pulse signal to the area array camera in the electric receiving boot acquisition unit, so as to acquire the image of the electric receiving boot by using the area array camera and the supplementary lighting device in the electric receiving boot acquisition unit to obtain the target electric receiving boot image;

[0033] The electric boot image storage module is used to transmit and store the target electric boot image to the electric boot image storage and recognition unit using the data transmission device connected between the electric boot acquisition unit and the electric boot image storage and recognition unit.

[0034] The detection result determination module is used to locate the target electric boot image using the deep neural network classifier in the electric boot image storage and recognition unit, obtain the component location result, and use the anomaly detection algorithm to detect the geometric dimensions, tilt angle, and surface defects of the component location result in turn, so as to obtain the corresponding wear value detection result, crack detection result, and notch anomaly detection result of the electric boot respectively.

[0035] Optionally, the power receiving shoe image acquisition module includes:

[0036] A pulse signal generation unit is used to configure the magnet on both sides of the wheel track, and trigger the magnet to generate a pulse signal corresponding to the vehicle when the vehicle passes over the track, and then send the pulse signal to the area array camera in the power receiving shoe acquisition unit; the area array camera is a camera that meets preset high-definition conditions;

[0037] The image capture unit is used to adjust the image brightness under preset strong backlight conditions using a dimming algorithm, a lens aperture control algorithm and a supplementary lighting device in the power receiving boot acquisition unit to obtain the target image brightness. Then, the area array camera is called and the image is captured from the top and side angles of the power receiving boot based on the target image brightness and a preset trigger and high-speed capture method to obtain the power receiving boot image to be processed.

[0038] The image analysis result determination unit is used to perform image analysis on the key areas of the power receiving boot in the image to be processed using ROI technology, and obtain image analysis results. The image analysis results are then used to adjust the image to be processed to obtain a target power receiving boot image. The target power receiving boot image includes geometric features and surface conditions corresponding to the power receiving boot.

[0039] Thirdly, this application provides an electronic device, comprising:

[0040] Memory, used to store computer programs;

[0041] A processor is used to execute the computer program to implement the aforementioned vehicle electric shock shoe detection method.

[0042] Fourthly, this application provides a computer-readable storage medium for storing a computer program, wherein the computer program, when executed by a processor, implements the aforementioned vehicle electric shock shoe detection method.

[0043] As can be seen from the above, before performing vehicle electric shock boot inspection, this application needs to use the magnet in the electric shock boot acquisition unit to generate a pulse signal corresponding to the vehicle when it passes by, and send the pulse signal to the area array camera in the electric shock boot acquisition unit to acquire the image of the electric shock boot, thereby obtaining an initial electric shock boot image. The initial electric shock boot image is then enhanced by the supplementary lighting device in the electric shock boot acquisition unit to obtain the target electric shock boot image. The target electric shock boot image is then transmitted and stored in the electric shock boot image storage and recognition unit using the data transmission device connected between the electric shock boot acquisition unit and the electric shock boot image storage and recognition unit. The deep neural network classifier in the electric shock boot image storage and recognition unit is used to locate the component in the target electric shock boot image, thereby obtaining the component location result. An anomaly detection algorithm is then used to sequentially detect the geometric dimensions, tilt angle, and surface defects of the component location result, thereby obtaining the corresponding wear value detection result, crack detection result, and notch anomaly detection result of the electric shock boot.

[0044] Therefore, this application first utilizes the magnet in the electric shock boot acquisition unit to generate a pulse signal corresponding to the vehicle as it passes by. This pulse signal is then sent to the area array camera in the electric shock boot acquisition unit to acquire an initial image of the electric shock boot. The initial image is then enhanced using a supplementary lighting device in the electric shock boot acquisition unit to obtain the target image of the electric shock boot. Secondly, the target image of the electric shock boot is transmitted and stored in the electric shock boot image storage and recognition unit using a data transmission device connected between the electric shock boot acquisition unit and the electric shock boot image storage and recognition unit. Finally, a deep neural network classifier in the electric shock boot image storage and recognition unit performs component localization on the target image of the electric shock boot, obtaining the component localization result. An anomaly detection algorithm is then used to sequentially detect the geometric dimensions, tilt angle, and surface defects of the component localization result, obtaining the corresponding wear value detection result, crack detection result, and notch anomaly detection result of the electric shock boot, respectively. This improves the efficiency of electric shock boot detection in vehicle inspection, thereby enhancing the user experience. Attached Figure Description

[0045] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0046] Figure 1 This is a flowchart of a vehicle electric sensor shoe testing method disclosed in this application;

[0047] Figure 2This application discloses a flowchart of a specific vehicle electric sensor shoe testing method.

[0048] Figure 3 This is a schematic diagram illustrating a specific process for image acquisition and recognition disclosed in this application;

[0049] Figure 4 This is a flowchart of a method disclosed in this application for recognizing an image to be recognized acquired by an image acquisition unit of an electric receiving boot using an image storage and recognition unit;

[0050] Figure 5 A schematic diagram of the captured image of the power receiving boot;

[0051] Figure 6 This is a schematic diagram of the structure of a vehicle electric shock boot detection device disclosed in this application;

[0052] Figure 7 This is a structural diagram of an electronic device disclosed in this application. Detailed Implementation

[0053] 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.

[0054] Currently, during long-term operation, the power receiving shoes of urban rail vehicles wear out and age. If not inspected and repaired in a timely manner, this can damage the power receiving shoes and affect vehicle operation. Currently, power receiving shoe inspection is mainly done manually, which is labor-intensive, costly, inefficient, lacks real-time performance, and poses significant safety hazards. Therefore, this application provides a method for inspecting vehicle power receiving shoes, which improves the efficiency of power receiving shoe inspection.

[0055] See Figure 1 As shown, this invention discloses a method for detecting vehicle electric shock boots, applied to a vehicle electric shock boot detection device including an electric shock boot acquisition unit, an electric shock boot image storage and recognition unit, and a data transmission device, comprising:

[0056] Step S11: When a vehicle passes by, the magnet in the electric boot acquisition unit generates a pulse signal corresponding to the vehicle, and sends the pulse signal to the area array camera in the electric boot acquisition unit. The area array camera and the supplementary lighting device in the electric boot acquisition unit are used to acquire the image of the electric boot to obtain the target electric boot image.

[0057] In this embodiment, the flowchart for testing the vehicle's power receiving shoe is shown below. Figure 2 As shown: The vehicle electric shock boot detection equipment is used to detect the condition of the electric shock boots during vehicle entry / exit or operation. The equipment can automatically calculate the wear value, cracks, gaps, and other defects of the electric shock boots. Specifically, the equipment includes an electric shock boot acquisition unit, an electric shock boot image storage and recognition unit, and a data transmission device. The electric shock boot acquisition unit includes a magnet, an area array camera, and a supplementary lighting device.

[0058] It is worth mentioning that magnets are installed on both sides of the track. When a wheel passes over the track, the magnets are triggered to emit pulse signals to the area scan camera. Using this high-precision triggering device, high-definition cameras can be used to acquire high-definition images, reducing the time overhead of the image triggering algorithm, improving the real-time response of the system, and significantly reducing the amount of image data acquired on-site. The area scan camera, located on both sides of the track, begins image acquisition of the power receiving shoe upon receiving the pulse signal emitted by the magnets.

[0059] Furthermore, the area scan camera in this embodiment can be a high-definition area scan camera because it has the following advantages: it employs superior camera dimming and lens aperture control algorithms. When strong backlighting occurs, the camera automatically adjusts the image brightness. It uses ROI technology to perform focused image analysis on key areas. The image brightness in these key areas is automatically adjusted using aperture and dimming algorithms, while strong light in other locations does not affect the imaging. Therefore, the power receiving shoe acquisition unit can use precise triggering combined with high-speed snapshot to acquire the original high-definition image of the power receiving shoe.

[0060] In one specific implementation, the magnet, area scan camera, and supplementary lighting equipment can be installed on columns on both sides of the vehicle track. For example, each column can be equipped with three area scan cameras, one magnet, and one supplementary light. When the electric receiving shoe enters the field of view of the area scan camera, the corresponding wheel will trigger the magnet, which will then send a pulse signal to the three area scan cameras on that side. The area scan cameras can then capture high-definition images of the electric receiving shoe from the top and two sides. It is worth noting that after the vehicle arrives, the wheel triggers the magnet once, the magnet sends a pulse signal once, and the area scan camera captures an image once. Therefore, for a single vehicle, the wheel will trigger the magnet multiple times, and the area scan camera will receive multiple pulse signals, resulting in multiple image capture operations.

[0061] Specifically, the magnets in the electric shock boot acquisition unit generate pulse signals corresponding to the vehicle when it passes by, and send these pulse signals to the area scan camera in the electric shock boot acquisition unit. The area scan camera and the supplementary lighting equipment in the electric shock boot acquisition unit then acquire images of the electric shock boot to obtain the target electric shock boot image. This can include: placing magnets on both sides of the wheel track, triggering the magnets to generate pulse signals corresponding to the vehicle when it passes by, and then sending the pulse signals to the area scan camera in the electric shock boot acquisition unit; the area scan camera is a camera that meets preset high-definition conditions; and utilizing dimming algorithms and lens aperture control algorithms... The supplementary lighting device in the power receiving boot acquisition unit adjusts the image brightness under preset strong backlight conditions to obtain the target image brightness. Then, an area array camera is called and images are captured from the top and side angles of the power receiving boot based on the target image brightness and preset triggering and high-speed capture methods to obtain the power receiving boot image to be processed. The ROI technology is used to perform image analysis on the key areas of the power receiving boot in the image to be processed to obtain the image analysis results. The image analysis results are used to adjust the image to be processed to obtain the target power receiving boot image. The target power receiving boot image includes the geometric features and surface state corresponding to the power receiving boot.

[0062] Step S12: Use the data transmission device connected between the electric boot acquisition unit and the electric boot image storage and recognition unit to transmit and store the target electric boot image to the electric boot image storage and recognition unit.

[0063] In this embodiment, the electric boot image storage and recognition unit is connected to the data transmission device to receive the electric boot image to be identified captured by the area scan camera and to identify the abnormal state of the electric boot. That is, the electric boot image storage and recognition unit can be implemented using an industrial control computer, which can also serve as a data storage management server, running acquisition software, recognition software, and management software. The computer automatically performs functions such as vehicle arrival / departure judgment, image acquisition, vehicle number matching, automatic image calculation, data storage, forwarding, and alarm settings. The data transmission device consists of fiber optic switching equipment, such as a gigabit switch, which uses fiber optic terminal equipment to achieve network transmission, transmitting the electric boot image captured by the area scan camera to the electric boot storage and recognition unit.

[0064] Specifically, the transmission of the target electric boot image to the electric boot image storage and recognition unit using a data transmission device connected between the electric boot acquisition unit and the electric boot image storage and recognition unit can include: constructing a data transmission device based on several fiber optic switching devices, and connecting the data transmission device to the electric boot acquisition unit and the electric boot image storage and recognition unit; the data transmission device includes gigabit switches and fiber optic terminal equipment; the data transmission device is used to transmit the target electric boot image acquired by the electric boot acquisition unit to the electric boot image storage and recognition unit implemented using an industrial control computer; the electric boot image storage and recognition unit includes a data storage management server, used to run acquisition software, recognition software and management software, and used to perform functions such as vehicle arrival and departure judgment, image acquisition, vehicle number matching, image calculation, data storage, data forwarding and alarm setting.

[0065] Furthermore, the electric shock shoe detection device in this embodiment may also include a system startup unit for detecting an approaching vehicle signal and starting the system. That is, after receiving the approaching vehicle signal, the system software starts, waiting for the area scan camera to transmit image data. Once the electric shock shoe enters the camera's field of view, the corresponding wheel will trigger the magnet, and the area scan camera will immediately begin acquiring images. Figure 3 As shown, the specific process of image acquisition and recognition in this embodiment is as follows: Before the train reaches the power receiving shoe acquisition unit, the wheels pass the triggering device, triggering the magnet to send a pulse signal to the area array camera. The camera captures an image of the power receiving shoe and uploads it to the power receiving shoe storage and recognition unit for recognition and comparison via a gigabit switch. If an abnormal state of the power receiving shoe is found after comparison, the power receiving shoe image storage and recognition unit will issue an abnormality alarm. After the train passes, the image acquisition operation ends, and the power receiving shoe acquisition unit will automatically shut down.

[0066] Step S13: Use the deep neural network classifier in the image storage and recognition unit of the power receiving boot to locate the target power receiving boot image, obtain the component location result, and use the anomaly detection algorithm to detect the geometric size, tilt angle and surface defects of the component location result in sequence, to obtain the corresponding wear value detection result, crack detection result and notch anomaly detection result of the power receiving boot respectively.

[0067] In this embodiment, the flowchart of recognizing the image to be recognized acquired by the electric shock boot acquisition unit using the electric shock boot image storage and recognition unit is as follows: Figure 4As shown: First, a deep neural network classifier is obtained by training with sample images; wherein, in this embodiment, the network weight parameters can be obtained by training with a neural network algorithm. Then, the neural network classifier is used to locate components in the acquired images of the power receiving boot. Next, high-definition images acquired by a high-definition area scan camera are input into the trained neural network classifier, thereby outputting the category and location information of each target component. Finally, an anomaly detection algorithm is used to determine anomalies in the located components.

[0068] Specifically, the deep neural network classifier in the image storage and recognition unit of the power receiving boot is used to locate the components in the target power receiving boot image, obtaining the component location results. Then, an anomaly detection algorithm is used to sequentially detect the geometric dimensions, tilt angle, and surface defects of the component location results, obtaining the corresponding wear value detection results, crack detection results, and notch anomaly detection results for the power receiving boot, respectively. This can include: using a preset neural network algorithm and adjusting the network weight parameters of the initial deep neural network classifier based on historical sample images to obtain the target deep neural network classifier; the historical sample images are historical sample images corresponding to the power receiving boot. The target deep neural network classifier is used to locate the components of the target electric receiving boot image, obtaining component location results including the category and location information of each component of the electric receiving boot. Then, based on the component location results, anomaly detection algorithms are used to detect the geometric dimensions, tilt angle, and surface defects of the electric receiving boot in sequence, obtaining the corresponding wear value detection results, crack detection results, and notch anomaly detection results of the electric receiving boot, respectively. Among them, the geometric dimensions include the length and thickness of the electric receiving boot; the tilt angle is the angle between the boundary contour line of the lower surface of the carbon slide plate and the upper edge line of the rail; surface defects include scratches, burns, and chips on the electric receiving boot.

[0069] It is worth mentioning that the detection range corresponding to the embodiments of this application includes the geometric dimensions, tilt angle, surface scratches, burns, and chipping of the electric receiving shoe. Among them, the geometric dimensions of the electric receiving shoe refer to its length and thickness, and detecting these geometric dimensions can determine whether the wear of the electric receiving shoe exceeds the limit.

[0070] The process for detecting the geometric dimensions of the electric receiving boot includes the following steps: First, in this embodiment, a neural network classifier is used to locate the image position of the electric receiving boot in the image, and the contour information of the electric receiving boot is extracted based on the edge contour. Specifically, since the material of the electric receiving boot is a carbon skateboard, the boundary contour feature values ​​of the upper surface and the lower surface of the carbon skateboard are picked up, and the boundary contour lines of the upper and lower surfaces of the carbon skateboard are obtained by calculation. Then, the number of pixels between the upper and lower surfaces is calculated, and the thickness H of the carbon skateboard is calculated based on the number of pixels. The wear value of the electric receiving boot can be determined by the thickness H. Next, the left and right boundary contour feature values ​​are picked up, the left and right boundary contour lines are obtained by calculation, and the number of pixels between the left and right boundaries is calculated. The length L of the electric receiving boot is calculated based on the number of pixels.

[0071] Furthermore, the calculation process for detecting the tilt angle of the electric receiving shoe includes: picking up the boundary contour feature value of the lower surface of the carbon slide plate and the upper edge contour feature value of the rail, and obtaining the boundary contour line of the lower surface of the carbon slide plate and the upper edge line of the rail by calculation. The tilt angle of the electric receiving shoe can be determined by calculating the angle between the two lines. When the electric receiving shoe passes through the detection point, its tilt angle in the image should be within a certain range. If the tilt angle exceeds a certain range, the posture of the electric receiving shoe is abnormal.

[0072] It is worth mentioning that, for detecting surface scratches, burns, and chipping on the electric receiving boot, since the electric receiving boot is made of carbon sliding plate material, its outline and the imaging of each surface should be relatively uniform under normal conditions. Therefore, dimensional peeling technology is used to gradually peel off the normal state area. When the area of ​​the unpeeled area reaches the set threshold, a defect alarm can be triggered. Then, based on the color and outline characteristics of the defect, it is labeled as a scratch, burn, or chipping abnormality.

[0073] Specifically, anomaly detection algorithms are used to sequentially detect geometric dimensions, tilt angles, and surface defects in the component positioning results, obtaining corresponding wear value detection results, crack detection results, and notch anomaly detection results for the power receiving boot. This can include: using a target deep neural network classifier to locate the power receiving boot position corresponding to the power receiving boot image; using anomaly detection algorithms and based on the power receiving boot position and component positioning results to perform geometric dimension detection, obtaining geometric dimension detection results; using edge contour extraction technology and based on the geometric dimension detection results to determine the first boundary contour feature value corresponding to the upper surface of the carbon slide plate and the second boundary contour feature value corresponding to the lower surface of the carbon slide plate in the power receiving boot; using the first boundary contour feature value and the second boundary contour feature value to determine the first boundary contour line; then determining the number of pixels on the first boundary contour line; and based on the number of pixels and the first boundary contour line, determining the thickness of the power receiving boot and the wear value of the power receiving boot. The system calculates the length of the electric shoe and determines the wear value corresponding to it based on the shoe's thickness and length. It then checks if the wear value exceeds a preset wear threshold to obtain the wear value detection result. Next, it determines the edge contour feature value corresponding to the edge of the electric shoe on the rail and constructs a second boundary contour line based on this feature value. Then, it constructs an edge contour line based on the edge contour feature value to determine the angle between the second boundary contour line and the edge contour line. It checks if the angle exceeds a preset angle threshold to obtain the crack detection result. Finally, it uses dimensional peeling technology to peel the normal state interval of the component positioning result, obtaining the peeled result. It then checks if the area of ​​the unpeeled interval corresponding to the peeled result exceeds a preset area threshold to obtain the defect anomaly detection result. If the area of ​​the unpeeled interval corresponding to the peeled result exceeds the preset area threshold, a defect alarm is triggered, and the defect is categorized as a scratch, burn, or chip based on its color and contour features.

[0074] In one specific embodiment, the vehicle electric shock boot detection device in this application can utilize deep neural network and machine vision technology to fuse electric shock boot images. Within the train's operating speed range of 1-35 km / h, as long as each detection item of the electric shock boot reaches a certain range, the device can detect the corresponding detection results. For example: wear detection accuracy ≥ 0.5 mm, tilt angle detection accuracy ≥ 5°, crack detection accuracy: length greater than 20 mm, width greater than 2 mm, chip detection accuracy ≥ 10 mm * 10 mm, burn detection accuracy ≥ 20 mm * 20 mm, foreign object detection accuracy: objects with a color different from the surface of the electric shock boot, and an object area ≥ 40 mm * 40 mm.

[0075] In another specific embodiment, the electric boot acquisition unit can acquire color images. The acquired images are high-definition and true-color, and are then transmitted to the electric boot storage and identification unit via a data transmission device. The electric boot storage and identification unit can locate the position of the electric boot based on deep neural networks and machine vision technology, and detect the wear degree (thickness H) and length (L) of the electric boot. A schematic diagram of the acquired electric boot image is shown below. Figure 5 As shown, 1 is the wear boundary of the power receiving shoe, and 2 is the wear surface of the power receiving shoe.

[0076] Specifically, after using an anomaly detection algorithm to sequentially detect geometric dimensions, tilt angles, and surface defects in the component positioning results to obtain the corresponding wear value detection results, crack detection results, and notch anomaly detection results for the power receiving shoe, the process can further include: if the wear value detection result indicates that the wear value is greater than a preset wear threshold, then the power receiving shoe is determined to be worn; if the crack detection result indicates that the included angle is greater than a preset included angle threshold, then the power receiving shoe is determined to have a crack; if the notch anomaly detection result indicates that the area of ​​the unpeeled section corresponding to the peeled result is greater than a preset area threshold, then the power receiving shoe is determined to have a notch, and a defect alarm is triggered. Then, based on the notch anomaly detection results, the defect color and contour features are determined to classify the notch as a scratch, burn, or chip.

[0077] It is worth mentioning that the embodiments of this application detect the status of the electric shock boot during vehicle entry and exit or operation. Non-contact online detection is performed using machine vision and intelligent image recognition methods, which can automatically calculate the wear value, cracks, gaps and other defects of the electric shock boot. This enables 24-hour unattended operation, improves the automation level of electric shock boot detection, and reduces the labor intensity of manual detection.

[0078] As can be seen from the above, the embodiments of this application first require the use of the magnet in the electric shock boot acquisition unit to generate a pulse signal corresponding to the vehicle when it passes by, and then sending the pulse signal to the area array camera in the electric shock boot acquisition unit to acquire the image of the electric shock boot, obtaining an initial electric shock boot image. The initial electric shock boot image is then enhanced using the supplementary lighting device in the electric shock boot acquisition unit to obtain the target electric shock boot image. Secondly, the target electric shock boot image is transmitted and stored in the electric shock boot image storage and recognition unit using a data transmission device connected between the electric shock boot acquisition unit and the electric shock boot image storage and recognition unit. Finally, the deep neural network classifier in the electric shock boot image storage and recognition unit performs component localization on the target electric shock boot image, obtaining the component localization result. An anomaly detection algorithm is then used to sequentially detect the geometric dimensions, tilt angle, and surface defects of the component localization result, obtaining the corresponding wear value detection result, crack detection result, and notch anomaly detection result of the electric shock boot, respectively. In this way, the efficiency of detecting the vehicle's electric shock boot is improved during the vehicle electric shock boot detection process, thereby enhancing the user experience.

[0079] Accordingly, see Figure 6 As shown, this application also provides a vehicle electric shock boot detection device, applied to a vehicle electric shock boot detection device including an electric shock boot acquisition unit, an electric shock boot image storage and recognition unit, and a data transmission device, comprising:

[0080] The image acquisition module 11 for receiving boots is used to generate a pulse signal corresponding to the vehicle when the magnet in the receiving boot acquisition unit passes by, and send the pulse signal to the area array camera in the receiving boot acquisition unit, so as to acquire the image of the receiving boot by using the area array camera and the supplementary lighting device in the receiving boot acquisition unit to obtain the target image of the receiving boot.

[0081] The image storage module 12 for receiving boots is used to transmit and store the target image of the receiving boots to the image storage and recognition unit using the data transmission device connected between the receiving boots acquisition unit and the receiving boots image storage and recognition unit.

[0082] The detection result determination module 13 is used to locate the target electric boot image using the deep neural network classifier in the electric boot image storage and recognition unit, obtain the component location result, and use the anomaly detection algorithm to detect the geometric size, tilt angle and surface defects of the component location result in turn, so as to obtain the corresponding wear value detection result, crack detection result and notch anomaly detection result of the electric boot respectively.

[0083] In some specific embodiments, the power receiving boot image acquisition module 11 may specifically include:

[0084] A pulse signal generation unit is used to configure the magnet on both sides of the wheel track, and trigger the magnet to generate a pulse signal corresponding to the vehicle when the vehicle passes over the track, and then send the pulse signal to the area array camera in the power receiving shoe acquisition unit; the area array camera is a camera that meets preset high-definition conditions;

[0085] The image capture unit is used to adjust the image brightness under preset strong backlight conditions using a dimming algorithm, a lens aperture control algorithm and a supplementary lighting device in the power receiving boot acquisition unit to obtain the target image brightness. Then, the area array camera is called and the image is captured from the top and side angles of the power receiving boot based on the target image brightness and a preset trigger and high-speed capture method to obtain the power receiving boot image to be processed.

[0086] The image analysis result determination unit is used to perform image analysis on the key areas of the power receiving boot in the image to be processed using ROI technology, and obtain image analysis results. The image analysis results are then used to adjust the image to be processed to obtain a target power receiving boot image. The target power receiving boot image includes geometric features and surface conditions corresponding to the power receiving boot.

[0087] In some specific embodiments, the power receiving boot image storage module 12 may specifically include:

[0088] A data transmission device construction unit is used to construct a data transmission device based on several fiber optic switching devices, and to connect the data transmission device to the power receiving shoe acquisition unit and the power receiving shoe image storage and recognition unit; the data transmission device includes a gigabit switch and fiber optic terminal equipment;

[0089] An image transmission unit is used to transmit the target electric boot image acquired by the electric boot acquisition unit to the electric boot image storage and recognition unit implemented by an industrial control computer using the data transmission device; the electric boot image storage and recognition unit includes a data storage management server, which is used to run acquisition software, recognition software and management software, and to perform functions such as vehicle arrival and departure judgment, image acquisition, vehicle number matching, image calculation, data storage, data forwarding and alarm setting.

[0090] In some specific embodiments, the detection result determination module 13 may specifically include:

[0091] The weight parameter adjustment unit is used to adjust the network weight parameters corresponding to the initial deep neural network classifier using a preset neural network algorithm and based on historical sample images to obtain the target deep neural network classifier; the historical sample images are historical sample images corresponding to the electric boot.

[0092] The component positioning result determination unit is used to perform component positioning on the target power receiving boot image using the target deep neural network classifier, and obtain component positioning results including category and location information corresponding to each component of the power receiving boot. Then, based on the component positioning results, an anomaly detection algorithm is used to sequentially detect the geometric dimensions, tilt angle, and surface defects of the power receiving boot, and obtain the corresponding wear value detection results, crack detection results, and notch anomaly detection results of the power receiving boot, respectively. Among them, the geometric dimensions include the length and thickness of the power receiving boot; the tilt angle is the angle between the boundary contour line of the lower surface of the carbon slide plate and the upper edge line of the rail; the surface defects include scratches, burns, and chips on the power receiving boot.

[0093] In some specific embodiments, the detection result determination module 13 may specifically include:

[0094] The geometric dimension detection result determination unit is used to locate the position of the electric boot corresponding to the image of the electric boot using the target deep neural network classifier, and to perform geometric dimension detection using the anomaly detection algorithm based on the position of the electric boot and the component positioning result to obtain the geometric dimension detection result;

[0095] The pixel count determination unit is used to determine the first boundary contour feature value corresponding to the upper surface of the carbon slide plate in the electric boot and the second boundary contour feature value corresponding to the lower surface of the carbon slide plate based on the geometric size detection result using edge contour extraction technology, so as to determine the first boundary contour line based on the first boundary contour feature value and the second boundary contour feature value, and then determine the number of pixels on the first boundary contour line.

[0096] The wear value detection result determination unit is used to determine the thickness and length of the power receiving boot corresponding to the power receiving boot based on the number of pixels and the first boundary contour line, and to determine the wear value corresponding to the power receiving boot based on the thickness and length of the power receiving boot, and then determine whether the wear value is greater than a preset wear threshold to obtain the wear value detection result;

[0097] The crack detection result determination unit is used to determine the edge contour feature value corresponding to the edge of the power receiving shoe on the rail, and construct a second boundary contour line based on the second boundary contour feature value. Then, it constructs an edge contour line based on the edge contour feature value to determine the angle between the second boundary contour line and the edge contour line, and determines whether the angle is greater than a preset angle threshold to obtain the crack detection result.

[0098] The notch anomaly detection result determination unit is used to perform normal state interval stripping operation on the component positioning result using dimensional stripping technology to obtain the stripped result, and to determine whether the area of ​​the unstripped interval corresponding to the stripped result is greater than a preset area threshold, thereby obtaining the notch anomaly detection result.

[0099] The contour feature calibration unit is used to issue a defect alarm if the area of ​​the unpeeled interval corresponding to the peeled result is greater than the preset area threshold, and to calibrate the defect as a scratch, burn or chip based on the defect color and contour features.

[0100] In some specific embodiments, the vehicle electric shock shoe detection device may further include:

[0101] The first power receiving shoe determination unit is used to determine that the power receiving shoe is worn if the wear value detection result indicates that the wear value is greater than the preset wear threshold.

[0102] The second power receiving shoe determination unit is used to determine that there is a crack in the power receiving shoe if the crack detection result indicates that the included angle is greater than the preset included angle threshold.

[0103] The defect alarm unit is used to determine that there is a gap in the power receiving shoe and to issue a defect alarm if the area of ​​the unpeeled section corresponding to the peeled result is greater than the preset area threshold, based on the gap anomaly detection result. Then, based on the gap anomaly detection result, the defect color and contour features are determined to label the gap as a scratch, burn or chip.

[0104] Furthermore, embodiments of this application also disclose an electronic device, Figure 7 This is a structural diagram of an electronic device 20 according to an exemplary embodiment. The content of the diagram should not be construed as limiting the scope of this application. The electronic device 20 may specifically include: at least one processor 21, at least one memory 22, a power supply 23, a communication interface 24, an input / output interface 25, and a communication bus 26. The memory 22 stores a computer program, which is loaded and executed by the processor 21 to implement the relevant steps in the vehicle electric shoe detection method disclosed in any of the foregoing embodiments. Furthermore, the electronic device 20 in this embodiment may specifically be a computer.

[0105] In this embodiment, the power supply 23 is used to provide operating voltage for each hardware device on the electronic device 20; the communication interface 24 can create a data transmission channel between the electronic device 20 and external devices, and the communication protocol it follows can be any communication protocol applicable to the technical solution of this application, and is not specifically limited here; the input / output interface 25 is used to acquire external input data or output data to the outside world, and its specific interface type can be selected according to specific application needs, and is not specifically limited here.

[0106] In addition, the memory 22, as a carrier for resource storage, can be a read-only memory, random access memory, disk or optical disk, etc. The resources stored thereon can include operating system 221, computer program 222, etc., and the storage method can be temporary storage or permanent storage.

[0107] The operating system 221 is used to manage and control the various hardware devices on the electronic device 20 and the computer program 222, which may be Windows Server, Netware, Unix, Linux, etc. In addition to including a computer program capable of performing the vehicle electric shoe detection method executed by the electronic device 20 as disclosed in any of the foregoing embodiments, the computer program 222 may further include a computer program capable of performing other specific tasks.

[0108] Furthermore, this application also discloses a computer-readable storage medium for storing a computer program; wherein, when the computer program is executed by a processor, it implements the aforementioned vehicle electric shock shoe detection method. Specific steps of this method can be found in the corresponding content disclosed in the foregoing embodiments, and will not be repeated here.

[0109] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to in the method section.

[0110] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0111] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly by hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0112] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0113] The technical solutions provided in this application have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A vehicle pantograph detection method characterized by comprising: The application is applied to a vehicle shoe detection device including a shoe collecting unit, a shoe image storage and recognition unit and a data transmission device, and comprises: The magnetic steel in the shoe collecting unit generates a pulse signal corresponding to the vehicle when the vehicle passes, and sends the pulse signal to the area array camera in the shoe collecting unit to collect the image of the shoe by using the area array camera and the light supplement device in the shoe collecting unit, and obtain the target shoe image; The data transmission device connected between the shoe collecting unit and the shoe image storage and recognition unit is used to transmit and store the target shoe image to the shoe image storage and recognition unit; The deep neural network classifier in the shoe image storage and recognition unit is used to locate the components of the target shoe image, and obtain the component positioning result, and the abnormal detection algorithm is used to detect the geometric size, the inclination angle and the surface defect of the component positioning result in sequence, and obtain the wear value detection result, the crack detection result and the gap abnormality detection result of the shoe respectively.

2. The vehicle pantograph detection method according to claim 1, characterized by, The magnetic steel in the shoe collecting unit generates a pulse signal corresponding to the vehicle when the vehicle passes, and sends the pulse signal to the area array camera in the shoe collecting unit to collect the image of the shoe by using the area array camera and the light supplement device in the shoe collecting unit, and obtain the target shoe image; The magnetic steel is arranged on both sides of the wheel track, and the magnetic steel generates a pulse signal corresponding to the vehicle when the vehicle passes the track, and then the pulse signal is sent to the area array camera in the shoe collecting unit; the area array camera is a camera meeting the preset high-definition condition; The light adjustment algorithm, the lens aperture control algorithm and the light supplement device in the shoe collecting unit are used to adjust the image brightness under the preset strong backlight condition, and obtain the target image brightness, then the area array camera is called and the image is captured from the top and side of the shoe based on the target image brightness and the preset trigger and high-speed snapshot mode, and the to-be-processed shoe image is obtained; The ROI technology is used to analyze the image of the key area of the shoe in the to-be-processed shoe image, and obtain the image analysis result, and the to-be-processed shoe image is adjusted by using the image analysis result, and the target shoe image is obtained; the target shoe image includes the geometric features and the surface state corresponding to the shoe.

3. The vehicle pantograph detection method according to claim 1, characterized by, The data transmission device connected between the shoe collecting unit and the shoe image storage and recognition unit is used to transmit and store the target shoe image to the shoe image storage and recognition unit, comprising: Based on a plurality of optical fiber exchange devices, a data transmission device is constructed, and the data transmission device is connected with the shoe collecting unit and the shoe image storage and recognition unit; the data transmission device includes a gigabit switch and an optical fiber terminal device; The data transmission device is used to transmit the target shoe image collected by the shoe collecting unit to the shoe image storage and recognition unit realized by an industrial computer; the shoe image storage and recognition unit includes a data storage management server, which is used to run collection software, recognition software and management software, and is used to perform functions of judging arrival and departure of a train, image collection, correspondence between a train number and an image, image calculation, data storage, data forwarding and alarm setting.

4. The vehicle pantograph detection method according to any one of claims 1 to 3, characterized by, The target shoe image is subjected to component positioning by using a deep neural network classifier in the shoe image storage and recognition unit, component positioning results are obtained, and the component positioning results are subjected to detection of geometric dimensions, inclination angles and surface defects in sequence by using an anomaly detection algorithm, to obtain respectively corresponding wear value detection results, crack detection results and gap anomaly detection results of the shoe, including: The network weight parameters of an initial deep neural network classifier are adjusted by using a preset neural network algorithm and based on historical sample images, to obtain a target deep neural network classifier; the historical sample images are historical sample images corresponding to the shoe; The target shoe image is subjected to component positioning by using the target deep neural network classifier, component positioning results including categories and position information corresponding to each component of the shoe are obtained, and then the geometric dimensions, inclination angles and surface defects of the shoe are detected in sequence based on the component positioning results and by using the anomaly detection algorithm, to obtain respectively corresponding wear value detection results, crack detection results and gap anomaly detection results of the shoe; wherein the geometric dimensions include a shoe length and a shoe thickness; the inclination angle is an included angle between a lower surface boundary profile line of a carbon slide plate and an upper edge line of a rail; and the surface defects include scratches, burns and chipping of the shoe.

5. The vehicle pantograph detection method according to claim 4, characterized by, The target shoe image is subjected to component positioning by using the target deep neural network classifier, component positioning results including categories and position information corresponding to each component of the shoe are obtained, and then the geometric dimensions, inclination angles and surface defects of the shoe are detected in sequence based on the component positioning results and by using the anomaly detection algorithm, to obtain respectively corresponding wear value detection results, crack detection results and gap anomaly detection results of the shoe; wherein the geometric dimensions include a shoe length and a shoe thickness; the inclination angle is an included angle between a lower surface boundary profile line of a carbon slide plate and an upper edge line of a rail; and the surface defects include scratches, burns and chipping of the shoe. The shoe position corresponding to the shoe image is positioned by using the target deep neural network classifier, to perform geometric dimension detection based on the shoe position and the component positioning results by using the anomaly detection algorithm, to obtain a geometric dimension detection result; First boundary profile feature values corresponding to an upper surface of the carbon slide plate and second boundary profile feature values corresponding to a lower surface of the carbon slide plate in the shoe are determined based on the geometric dimension detection result by using an edge profile extraction technique, to determine a first boundary profile line based on the first boundary profile feature values and the second boundary profile feature values, and then determine a number of pixel points on the first boundary profile line; The shoe thickness and the shoe length corresponding to the shoe are determined based on the number of pixel points and the first boundary profile line, and the wear value corresponding to the shoe is determined based on the shoe thickness and the shoe length, and then it is judged whether the wear value is greater than a preset wear threshold, to obtain a wear value detection result; determining an edge profile feature value corresponding to an edge of the power shoe on the rail, constructing a second boundary profile line based on the second boundary profile feature value, and then constructing an edge profile line based on the edge profile feature value to determine an included angle between the second boundary profile line and the edge profile line, and judging whether the included angle is greater than a preset included angle threshold to obtain a crack detection result; performing a normal state interval stripping operation on the component positioning result by using a dimension stripping technology to obtain a stripped result, and judging whether an unstripped interval area corresponding to the stripped result is greater than a preset area threshold to obtain a notch abnormality detection result; if the unstripped interval area corresponding to the stripped result is greater than the preset area threshold, performing defect alarm, and labeling the defect as a scratch, a burn, or a block according to defect color and profile features.

6. The vehicle pantograph detection method according to claim 5, characterized by, After the utilization of the anomaly detection algorithm on the component positioning result to sequentially perform the detection of geometric size, inclination angle, and surface defect to obtain the wear value detection result, the crack detection result, and the notch abnormality detection result of the power shoe respectively, the method further includes: if the wear value detection result represents that the wear value is greater than the preset wear threshold, determining that the power shoe is worn out; if the crack detection result represents that the included angle is greater than the preset included angle threshold, determining that there is a crack in the power shoe; if the notch abnormality detection result represents that the unstripped interval area corresponding to the stripped result is greater than the preset area threshold, determining that there is a notch in the power shoe, and performing defect alarm, and then determining defect color and profile features based on the notch abnormality detection result to label the notch as a scratch, a burn, or a block.

7. A vehicle pantograph detection device characterized by comprising: The vehicle power shoe detection device is applied to a vehicle power shoe detection device including a power shoe collection unit, a power shoe image storage and recognition unit, and a data transmission device, and includes: a power shoe image collection module configured to generate a pulse signal corresponding to the vehicle by using a magnetic steel in the power shoe collection unit when the vehicle is driving, and transmit the pulse signal to a face array camera in the power shoe collection unit to collect an image of a power shoe by using the face array camera and a light supplementing device in the power shoe collection unit to obtain a target power shoe image; a power shoe image storage module configured to transmit and store the target power shoe image to the power shoe image storage and recognition unit by using the data transmission device connected between the power shoe collection unit and the power shoe image storage and recognition unit; a detection result determination module configured to perform component positioning on the target power shoe image by using a deep neural network classifier in the power shoe image storage and recognition unit to obtain a component positioning result, and sequentially perform the detection of geometric size, inclination angle, and surface defect on the component positioning result by using an anomaly detection algorithm to obtain a wear value detection result, a crack detection result, and a notch abnormality detection result of the power shoe respectively.

8. The vehicle pantograph detection device according to claim 7, characterized by The power shoe image collection module includes: The pulse signal generation unit is configured to arrange the magnetic steel on both sides of the wheel track, trigger the magnetic steel to generate a pulse signal corresponding to the vehicle when the vehicle drives through the track, and then send the pulse signal to the area array camera in the power shoe collection unit; the area array camera is a camera meeting preset high-definition conditions; The image capture unit is configured to adjust image brightness under preset strong backlight conditions by using a dimming algorithm, a lens aperture control algorithm, and the light supplementing device in the power shoe collection unit, to obtain target image brightness, then call the area array camera and capture images from the top and side of the power shoe based on the target image brightness and a preset trigger and high-speed snapshot mode, to obtain a to-be-processed power shoe image; The image analysis result determination unit is configured to perform image analysis on a power shoe key area in the to-be-processed power shoe image by using ROI technology, to obtain an image analysis result, and to adjust the to-be-processed power shoe image by using the image analysis result, to obtain a target power shoe image; the target power shoe image includes geometric features and surface states corresponding to the power shoe.

9. An electronic device, comprising: Comprise: A memory for saving a computer program; A processor for executing the computer program to implement the vehicle power shoe detection method of any one of claims 1 to 6.

10. A computer-readable storage medium, characterized in that, A computer program is saved, wherein the computer program is executed by a processor to implement the vehicle power shoe detection method of any one of claims 1 to 6.

Images