Multi-dimensional adjustable vehicle side and vehicle bottom collecting device and light supplementing cooperation system
By using a multi-dimensional adjustable side and undercarriage data acquisition device and a supplementary lighting system, the problem of balancing efficiency and detail in rail train inspection has been solved. This has enabled comprehensive and efficient data acquisition of components on the side and undercarriage, improving the overall efficiency and accuracy of the inspection.
Patent Information
- Application Number
- CN202511217032.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-08-28
- Publication Date
- 2025-11-11
Smart Images
Figure CN120935336A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of rail train inspection technology, and in particular to a multi-dimensional adjustable side and undercarriage acquisition device and a supplementary lighting system. Background Technology
[0002] In the daily maintenance and inspection of railcars, the inspection and maintenance of the sides and undercarriage are one of the important aspects.
[0003] In related technologies, when inspecting the sides and bottom of railcars, traditional inspections often involve maintenance personnel using handheld metal flaw detectors to sequentially inspect the metal components of the railcar to find any damaged parts, or using inspection robots to perform image inspections on components in obvious areas to find components with obvious damage or abnormalities.
[0004] Regarding the aforementioned technologies, when inspecting rail trains, personnel using instruments can produce relatively detailed results and perform multi-angle inspections. However, the inspection efficiency is not as good as that of robots using image acquisition and analysis. In turn, robots using image acquisition and analysis are not as comprehensive as personnel using multi-angle inspections. This makes it impossible to balance efficiency and inspection detail in the daily maintenance and inspection work of rail trains. Summary of the Invention
[0005] To improve the efficiency and detail of daily maintenance and inspection of rail trains, this application provides a multi-dimensional adjustable side and undercarriage acquisition device and a supplementary lighting system.
[0006] Firstly, this application provides a multi-dimensional adjustable vehicle side and undercarriage data acquisition device: A multi-dimensional adjustable vehicle side and undercarriage data acquisition device, comprising: include: The main body of the acquisition device is arranged along the preset track vehicle running path and is equipped with an inspection robot, multiple types of camera modules and a lifting adjustment module. The inspection robot carries the multiple types of camera modules and moves along the preset acquisition path to acquire component images of different areas on the side and under the vehicle. The lifting adjustment module is used to adjust the acquisition height of the multiple types of camera modules. An adjustment and control module, connected to the main body of the acquisition device, is configured with a multi-dimensional adjustment strategy to adjust the image acquisition parameters of the various types of camera modules and the acquisition path of the inspection robot, so as to adapt to the acquisition needs of different components on the side and under the vehicle. The collaborative control module is connected to the main body of the acquisition device and the adjustment control module respectively, and is configured with a device collaboration strategy to coordinate the working sequence of the inspection robot, the multi-type camera module and the lifting and adjusting module.
[0007] By adopting the above technical solution, the main body of the acquisition device, the adjustment and control module, and the collaborative control module work together to complete collaborative tasks such as mobile acquisition of components on the side and undercarriage of the train, adjustment of multi-dimensional parameters, and coordination of equipment working sequence. Combined with the mobile acquisition capability of the inspection robot, the differentiated image acquisition of multiple types of camera modules, and the height adaptation of the lifting and adjustment module, it helps to achieve comprehensive and detailed acquisition of components in different areas on the side and undercarriage of the rail train, adapt to the acquisition needs of different components, avoid the problems of incomplete coverage and low efficiency of traditional manual acquisition, and at the same time, through the time-series coordination between modules, ensure the orderliness and stability of the acquisition process, and improve the overall efficiency of image acquisition of rail train components.
[0008] Optionally, the multi-dimensional adjustment strategy includes: The height adjustment sub-strategy analyzes the height differences of vehicle side and undercarriage components based on the acquisition path to generate the target component shooting height, and generates the optimal shooting distance adapted to the shooting of the multiple types of camera modules based on the target component shooting height; The path adjustment sub-strategy generates a full-area topographic map based on the detection site and generates backup paths for data collection using a preset path analysis model for personnel to select and switch between. The range adjustment sub-strategy performs image analysis based on a preset rail vehicle template library to determine the detection type of the vehicle to be inspected. When the detection type is full inspection, the full inspection path is generated by integrating the acquisition path and the additional path. When the detection type is special inspection, the special inspection path is generated by segmenting the acquisition path and the special inspection path.
[0009] By adopting the above technical solutions, the height adjustment sub-strategy generates the optimal shooting distance by analyzing the height differences of components, ensuring clear imaging of components at different heights; the path adjustment sub-strategy improves the flexibility of the acquisition path and the ability to handle faults by generating a full-area topographic map and backup paths; and the range adjustment sub-strategy determines the detection type and generates the corresponding path by matching the vehicle template library, adapting to the different needs of full vehicle inspection and special inspection. The collaboration of the three helps to improve the targeting and adaptability of the acquisition device for the acquisition of railcar components, avoid the limitations of a single acquisition mode, reduce invalid acquisition operations, and improve acquisition accuracy and efficiency.
[0010] Optionally, the path adjustment sub-policy is further configured with a path switching verification sub-policy, including: The optimal path width pixel range and traffic flow threshold are determined based on the track size parameters of the depot, and the track size parameters correspond to the actual layout of the rail vehicle depot. Based on the current image of the inspection robot along the current acquisition path, extract the actual path width pixel value and the robot's travel speed value. The actual values are compared with the standard parameters to calculate the width deviation and speed deviation. If both are within the preset allowable deviation range, the current path is deemed to meet the requirements. If they exceed the range, the path selection is fine-tuned according to the deviation values and the path is re-verified until the path parameters meet the standard requirements.
[0011] By adopting the above technical solution, the path switching verification sub-strategy can accurately determine whether the current acquisition path meets the requirements by determining the standard parameters of the corresponding section track, extracting the actual parameters of the robot passage, and comparing and verifying them. If it exceeds the deviation range, it will be fine-tuned and re-verified. This helps to avoid problems such as robot passage jamming and acquisition position offset caused by inconsistent path parameters, ensures the smoothness of passage and acquisition stability after path switching, improves the adaptability and reliability of the acquisition device in different path scenarios, and reduces acquisition interruption or error caused by path problems.
[0012] Optionally, the adjustment and control module is further configured with a strategy to enhance acquisition accuracy, including: The vehicle number identification and positioning steps involve analyzing the image of the front of the rail vehicle and extracting and correcting the vehicle number text to determine the vehicle number information. The task image selection step involves extracting feature points from continuously acquired component images using a preset feature point extraction algorithm, and then searching for the task image with the smallest average distance in the template database corresponding to the vehicle number information as the current inspection task image for image acquisition.
[0013] By adopting the above technical solution, the vehicle number identification and positioning step and the task map selection step in the acquisition accuracy enhancement strategy work together. The vehicle number identification and positioning step achieves accurate vehicle number identification through image analysis and text correction, providing an initial positioning benchmark for acquisition. The task map selection step selects the optimal task map through feature point extraction and template comparison, making up for the robot's positioning accuracy limitations. The combination of the two helps to improve the matching accuracy of the acquisition device for the acquisition items of rail train components, avoids the omission or incorrect acquisition of components due to positioning deviation, provides accurate acquired images for subsequent component image analysis, and reduces detection errors caused by insufficient acquisition accuracy.
[0014] Optionally, the device collaboration strategy includes: In the collaborative shooting step, feature recognition is performed based on the component images along the acquisition path. The lifting adjustment module triggers the lifting adjustment command and adjusts the lifting according to the optimal shooting distance, while simultaneously triggering the image shooting command. In the obstacle avoidance coordination step, when the inspection robot moves along the collection path, it identifies obstacles on the path based on the obstacle recognition model, generates an obstacle avoidance adjustment distance in the height direction of the obstacle, and triggers the lifting module to adjust the obstacle avoidance.
[0015] By adopting the above technical solutions, the shooting coordination step triggers the lifting adjustment and image shooting synchronously through feature recognition, ensuring that the parts are imaged at the optimal shooting distance and improving image quality. The obstacle avoidance coordination step generates an avoidance distance through obstacle recognition and triggers the lifting adjustment to avoid collisions between the robot and obstacles in the path. The collaboration of the two helps to improve the safety of the acquisition device while ensuring the quality of the acquired images, avoiding image blurring caused by improper shooting distance or equipment damage caused by obstacle collisions, and ensuring the continuity and reliability of the acquisition process.
[0016] Optionally, the main body of the acquisition device is also equipped with a camera position pre-calibration strategy, including: Based on the direction of movement of the rail vehicle, a standard coordinate axis orientation is determined, and coordinate analysis is performed on the panoramic cameras located at both ends of the multi-type camera modules carried by the inspection robot to generate a baseline line connecting the coordinates of the two cameras. Based on the coordinates of the baseline and the corresponding coordinates of the 3D camera at the end of the robotic arm located in the middle, a point-to-line distance analysis is performed to determine the error distance between the 3D camera and the baseline. Based on the error distance and a preset position compensation model, a position compensation coefficient is generated. The position value of the component detected by the 3D camera is corrected according to the position compensation coefficient to ensure that the acquisition position deviation is within a preset small error range.
[0017] By adopting the above technical solution, the camera position pre-calibration strategy can accurately calibrate the acquisition positions of multiple types of camera modules by establishing standard coordinate axes, generating panoramic camera baselines, calculating 3D camera error distances, and correcting positions, thereby reducing positional deviations between the intermediate 3D camera and the panoramic cameras at both ends. This strategy helps ensure the consistency of positional acquisition by multiple cameras, avoids component positioning deviations caused by camera position offsets, improves the positional accuracy of image acquisition of rail train components, provides an accurate positional reference for subsequent image feature extraction and fault detection, and reduces detection misjudgments caused by camera position errors.
[0018] Optionally, the collaborative control module is further configured with a light source optimization strategy, including: Based on the feature analysis of the vehicle side and underside edge images acquired by the multi-type camera modules, the image shooting area is determined, and the optimal light intensity of the corresponding camera model is matched with the preset light database. The light source in the shooting area is adjusted according to the optimal light intensity and the preset parallel illumination angle. The light source adjustment gain value is calculated using a preset light source gain analysis model, and the gain value is kept within the preset reference gain value range.
[0019] By adopting the above technical solution, the light source optimization strategy analyzes the image features of the vehicle side and underside edges, matches the optimal illumination intensity, and adjusts the light source. Combined with a light source gain analysis model to calculate the gain value, it can accurately adapt to the acquisition illumination requirements of various types of camera modules. This strategy helps avoid image overexposure, underexposure, or edge blurring caused by insufficient or excessive illumination, ensuring grayscale uniformity and edge clarity in the acquired images. This provides a high-quality image foundation for subsequent component image analysis, improves the accuracy of fault feature identification, and reduces false negatives or missed detections due to illumination issues.
[0020] Optionally, the light source gain analysis model is calculated using the following formula:
[0021] Wherein, G is the light source adjustment gain value, which comprehensively reflects the enhancement coefficient of train components. The grayscale standard deviation of the component image after light source adjustment is used to quantify grayscale uniformity. The grayscale standard deviation of the component image before light source adjustment. This refers to the actual light transmittance of the camera lens. Set a baseline transmittance for the lens. This represents the actual reflection coefficient of the component. A baseline reflection coefficient is preset for the factory inspection values of the components. This refers to the actual exposure time of the camera. Set a baseline exposure time for the camera. This refers to the actual distance between the lens and the components. Preset a baseline acquisition distance for the lens and components. This is the preset ambient light interference correction factor. The actual ambient light intensity of the set segment field, Indicates the preset reference ambient light intensity. Indicates the contrast-edge co-correction coefficient. This indicates the amount of image contrast improvement after adjusting the light source. This is a preset baseline contrast, representing the minimum requirement for recognizing component details. This indicates the average grayscale gradient at the edge of the component after adjustment. The preset baseline grayscale gradient represents the threshold for edge sharpness. To compensate for the edge detection error of the parts, To preset the baseline edge detection error, This is the preset baseline edge detection error.
[0022] By adopting the above technical solution, the light source adjustment gain value is calculated by integrating multi-dimensional parameters such as image grayscale standard deviation, lens transmittance, component reflectance coefficient, exposure time, acquisition distance, and ambient light intensity. This allows for quantitative evaluation of the light source adjustment effect and ensures that the gain value is within the benchmark range. This helps to accurately control the illumination adjustment parameters, adapt to the acquisition needs of different material components and different ambient light scenarios of rail trains, reduce the subjectivity and instability of traditional experience-based illumination adjustment, and ensure the consistency of component image acquisition quality.
[0023] Secondly, this application provides a supplementary lighting system, which adopts the following technical solution: A supplementary lighting system is applied to a multi-dimensional adjustable vehicle side and under-vehicle data acquisition device, comprising: The supplementary lighting execution module is located next to the main body of the acquisition device of the multi-dimensional adjustable vehicle side and undercarriage acquisition device, and is arranged correspondingly to the multi-type camera modules of the acquisition device main body. It is used to output illumination adapted to the acquisition of images of vehicle side and undercarriage components. The supplementary lighting execution module can adjust the light intensity and illumination angle to adapt to the field of view of the multi-type camera modules. The supplementary lighting control module is connected to the supplementary lighting execution module and the collaborative control module of the main body of the acquisition device, respectively. It is configured with the light source optimization strategy preset by the collaborative control module. The supplementary lighting control module is used to receive the vehicle side and vehicle bottom edge images acquired by the multi-type camera modules, perform feature analysis on the images to determine the image shooting area, and match the optimal light intensity corresponding to the model of the multi-type camera module in the preset lighting database. The gain calculation submodule, integrated into the supplementary lighting control module, calculates the light source adjustment gain value according to the preset light source gain analysis model, and determines whether the gain value is within the preset reference gain value range. The collaborative adaptation module is connected to the supplementary lighting control module, the adjustment control module of the main body of the acquisition device, and the lifting adjustment module, respectively. It is used to receive the multi-dimensional adjustment strategy and acquisition accuracy enhancement strategy results issued by the adjustment control module, synchronously send the illumination adjustment trigger signal to the supplementary lighting control module, and receive the lifting adjustment in place signal from the lifting adjustment module. After the lifting adjustment module adjusts to the optimal shooting distance, it triggers the supplementary lighting execution module to output the adapted illumination. The parameter storage module, connected to the supplementary lighting control module, is used to store the preset illumination database, light source gain analysis model parameters, and historical illumination adjustment data, which are then called by the supplementary lighting control module to optimize the illumination adjustment accuracy.
[0024] By adopting the above technical solution, the supplementary lighting execution module, supplementary lighting control module, gain calculation submodule, collaborative adaptation module, and parameter storage module work together. The supplementary lighting execution module provides adapted lighting, the supplementary lighting control module matches the optimal light intensity, the gain calculation submodule quantifies the light source adjustment effect, the collaborative adaptation module achieves timing coordination with the acquisition device, and the parameter storage module provides data support. This multi-module collaboration helps provide accurate and adapted acquisition lighting for multi-dimensional adjustable vehicle side and undercarriage acquisition devices, avoiding poor image quality due to lighting issues, improving edge clarity and grayscale uniformity of vehicle side and undercarriage component images, and adapting to the lighting requirements of different acquisition paths, detection types, and component materials. This reduces detection errors caused by insufficient or improper lighting, improving the overall reliability of rail train component image acquisition and analysis.
[0025] In summary, this application includes at least one of the following beneficial technical effects: 1. The device integrates the main body, adjustment and control module, and collaborative control module to work together to complete collaborative tasks such as mobile acquisition of components on the side and undercarriage of the train, multi-dimensional parameter adjustment, and equipment working sequence coordination. Combining the mobile acquisition capabilities of the inspection robot, the differentiated image acquisition of multiple types of camera modules, and the height adaptation of the lifting adjustment module, it helps to achieve comprehensive and detailed acquisition of components in different areas of the side and undercarriage of the rail train. This adapts to the acquisition needs of different components, avoiding the problems of incomplete coverage and low efficiency of traditional manual acquisition. At the same time, through the time-series coordination between modules, the orderly and stable acquisition process is ensured, improving the overall efficiency of image acquisition of rail train components. 2. The height adjustment sub-strategy generates the optimal shooting distance by analyzing the height differences of components, ensuring clear imaging of components at different heights. The path adjustment sub-strategy improves the flexibility of the acquisition path and the ability to handle faults by generating a full-area topographic map and backup paths. The range adjustment sub-strategy determines the detection type and generates the corresponding path by matching the vehicle template library, adapting to the different needs of full vehicle inspection and special inspection. 3. In the strategy to enhance the accuracy of data acquisition, the vehicle number recognition and localization step and the task map selection step work together. The vehicle number recognition and localization step achieves accurate vehicle number recognition through image analysis and text correction, providing an initial localization benchmark for data acquisition. The task map selection step selects the optimal task map through feature point extraction and template comparison, compensating for the limitations of robot localization accuracy. Attached Figure Description
[0026] Figure 1 This is a schematic diagram of the overall structure of the vehicle side and undercarriage data acquisition device in this application.
[0027] Figure 2 This is a flowchart of steps S100 to S300 in this application.
[0028] Figure 3 This is a flowchart of steps S301 to S303 in this application.
[0029] Figure 4 This is a flowchart of steps S400 to S401 in this application.
[0030] Figure 5 This is a flowchart of steps S500 to S501 in this application.
[0031] Figure 6 This is a flowchart of steps S600 to S602 in this application.
[0032] Figure 7 This is a flowchart of steps S700 to S701 in this application. Detailed Implementation
[0033] To make the purpose, technical solution, and advantages of this application clearer, the following description is provided in conjunction with the appendix. Figures 1-7 The present application will be further described in detail below with reference to embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the application.
[0034] The embodiments of the present invention will now be described in further detail with reference to the accompanying drawings.
[0035] This application discloses a multi-dimensional adjustable vehicle side and undercarriage data acquisition device. The acquisition device body, adjustment control module, and collaborative control module work together to complete collaborative tasks such as mobile acquisition of vehicle side and undercarriage components, multi-dimensional parameter adjustment, and equipment working sequence coordination. Combining the mobile acquisition capabilities of the inspection robot, the differentiated image acquisition of multiple types of camera modules, and the height adaptation of the lifting adjustment module, it helps to achieve comprehensive and detailed acquisition of components in different areas of the side and undercarriage of rail trains, adapting to the acquisition needs of different components, avoiding the problems of incomplete coverage and low efficiency of traditional manual acquisition. At the same time, through the time-series coordination between modules, the orderly and stable acquisition process is ensured, improving the overall efficiency of image acquisition of rail train components.
[0036] Reference Figure 1 A multi-dimensional adjustable vehicle side and undercarriage data acquisition device includes: The main body of the data acquisition device is arranged along the preset track vehicle running path. It is equipped with an inspection robot, multiple types of camera modules and a lifting adjustment module. The inspection robot, carrying multiple types of camera modules, moves along the preset acquisition path and acquires images of components in different areas of the vehicle side and under the vehicle. The lifting adjustment module is used to adjust the acquisition height of the multiple types of camera modules. A pre-set rail vehicle operating path refers to a fixed installation path planned according to the standard operating trajectory of rail vehicles, such as those used for subways and light rail. This path is typically parallel to the direction of track extension and 0.8-1.2 meters from the edge of the track, ensuring full coverage of the area to the side of the vehicle.
[0037] The data acquisition path refers to the movement trajectory set by the inspection robot for different areas on the side and under the vehicle. The movement trajectory for the side is a straight line parallel to the vehicle body, while the movement trajectory for the underside is a serpentine trajectory along the longitudinal direction of the vehicle. These trajectories can cover critical areas such as the underbody bogies and braking components. Multi-type camera modules refer to combinations of industrial cameras integrating different functions. For example, a 20-megapixel high-definition color camera can be used to capture details of the component's appearance, while an infrared thermal imaging camera can be used to detect abnormal temperatures in components. The lens focal length of the cameras can be adapted within the range of 8-25mm depending on the acquisition distance.
[0038] The acquisition height refers to the vertical distance between the camera module and the component to be acquired. This distance needs to be adjusted according to the position of the component. For example, the acquisition height of the side door area of the vehicle needs to be controlled at 1.2-1.8 meters, and the acquisition height of the undercarriage bogie needs to be controlled at 0.3-0.8 meters.
[0039] In practice, the first step is to determine the preset rail vehicle operating path based on the model of the rail vehicle. The B-type subway car is a common example. After determining the path, the fixed bracket of the main body of the data acquisition device is installed along that path. The inspection robot, equipped with multiple types of camera modules, starts moving according to the preset acquisition path. When acquiring data from the side of the vehicle, the robot moves parallel to the vehicle at a speed of 0.5 m / s along the guide rail beside the track, while simultaneously activating the high-definition color camera. When acquiring data from under the vehicle, the robot moves in a serpentine pattern at a speed of 0.3 m / s along the hidden guide rail under the vehicle, while simultaneously activating the high-definition color camera and the infrared thermal imaging camera. When acquiring data from components at different heights, such as side windows and undercarriage suspension components, the lifting adjustment module drives the camera module up and down via an electric push rod to adjust the acquisition height to the appropriate value. For example, when acquiring data from windows, the height is adjusted to 1.5 meters, and when acquiring suspended components, the height is adjusted to 0.5 meters, thus ensuring that the camera module and the component to be acquired are at the optimal imaging distance.
[0040] By adapting the mobile coverage of the inspection robot to the height of the lifting and adjustment module, and combining the complementary functions of multiple types of camera modules, it is possible to acquire multi-dimensional component images of the entire area of the vehicle side and underside, providing a complete and clear image data source for subsequent component inspection and avoiding the loss of image information due to incomplete acquisition range or unsuitable distance.
[0041] The adjustment and control module is connected to the main body of the acquisition device and is equipped with a multi-dimensional adjustment strategy to adjust the image acquisition parameters of various types of camera modules and the acquisition path of the inspection robot to adapt to the acquisition needs of different parts on the side and under the vehicle. The adjustment and control module is the adjustment unit for the collected parameters and paths. It is connected to the main body of the acquisition device via an industrial Ethernet. It adapts to the acquisition needs of different components through a multi-dimensional adjustment strategy. The specific multi-dimensional adjustment strategy will be further explained in subsequent steps.
[0042] The collaborative control module is connected to the main body of the acquisition device and the adjustment control module respectively. It is configured with a device collaboration strategy to coordinate the working sequence of the inspection robot, the multi-type camera module and the lifting adjustment module.
[0043] The collaborative control module is the unit that coordinates the timing of the equipment. It is connected to the main body of the acquisition device and the adjustment control module via a CAN bus. Its core configuration is the equipment collaboration strategy, which is used to ensure that each device works in an orderly manner.
[0044] The equipment coordination strategy refers to the rules governing the sequence and time intervals of operation for inspection robots, various camera modules, and lifting and adjustment modules. This includes basic timing sequences such as movement-adjustment-acquisition and anomaly compensation timing sequences. The operational sequence refers to the time connection between the actions of each device. Taking the coordination of the inspection robot, lifting and adjustment module, and camera module as an example, after the inspection robot moves into position, the lifting and adjustment module will start after a 0.5-second delay. After the lifting and adjustment is completed, the camera module will start acquiring data after a further 0.3-second delay. This time interval setting avoids the impact of vibrations caused by equipment movement on image quality. The anomaly compensation timing sequence means that when a device experiences a delay, such as a 1-second delay in the lifting and adjustment module due to mechanical resistance, the system will automatically extend the start interval of subsequent devices. Specifically, the camera module's start delay can be increased from 0.3 seconds to 1.3 seconds to ensure smooth coordination between the actions of each device.
[0045] Reference Figure 2 The multi-dimensional adjustment strategy is implemented using the following methods: Step S100: Height adjustment sub-strategy, based on the acquisition path analysis of the height difference of the vehicle side and vehicle bottom components to generate the target component shooting height, and based on the target component shooting height, generate the optimal shooting distance adapted to shooting multiple types of camera modules; Step S200: Path adjustment sub-strategy, generate a full-area detection topographic map based on the detection site, and generate backup paths for the acquisition path using a preset path analysis model for personnel to select and switch. Step S300: Range adjustment sub-strategy. Based on the preset rail vehicle template library, perform image analysis to determine the detection type of the vehicle to be inspected. When the detection type is full inspection, integrate the acquisition path and additional path to generate a full inspection path. When the detection type is special inspection, divide the acquisition path and special inspection path into segments to generate a special inspection path.
[0046] Reference Figure 3 The path adjustment sub-policy is also configured with a path switching verification sub-policy, which includes the following steps: Step S301: Determine the width pixel range and traffic flow threshold of the optimal path based on the track size parameters of the depot. The track size parameters correspond to the actual layout of the rail vehicle depot. Data acquisition path analysis refers to the system's data processing of the component list and location coordinates covered by the inspection robot's preset acquisition path, clarifying the distribution and height relationship of each component along the path. Height difference between vehicle side and undercarriage components refers to the vertical height difference between components on the vehicle side and undercarriage of the same vehicle. For example, the typical height of a vehicle side door is approximately 1.2-1.8 meters, while the typical height of a vehicle undercarriage bogie is approximately 0.3-0.8 meters, resulting in a height difference of 0.9-1.5 meters. Target component shooting height refers to the vertical height of the camera module set for a single component to be acquired, ensuring a complete view of the component's appearance and details. This needs to be determined in conjunction with the component's size and installation location. Optimal shooting distance refers to the horizontal distance between the camera module lens and the surface of the component to be acquired. This distance needs to be matched with parameters such as camera focal length and pixel resolution to ensure the component image occupies an appropriate proportion in the frame. In practice, the control module first calls the component association data along the acquisition path to perform height statistics on the vehicle side and undercarriage components along the path. For example, it identifies four types of components covered by the path: vehicle side windows, vehicle side armrests, undercarriage suspension components, and undercarriage brake pads, clarifying the height differences of each component. Then, the system generates the target component shooting height based on the component size. For the large vehicle side window, the target shooting height is set to 1.5 meters; for the smaller undercarriage suspension components, the target shooting height is set to 0.6 meters. Finally, combining the current parameters of various camera modules, such as a high-definition color camera with a 12mm focal length and 20 megapixels, the optimal shooting distance is generated using the imaging ratio calculation formula: Shooting distance = Focal length × Actual component size × Horizontal pixels of image. The optimal shooting distance for the vehicle side window is 0.8 meters, and for the undercarriage suspension components, it is 0.5 meters. The lifting adjustment module then synchronously adjusts the vertical height and horizontal position of the camera module based on the target shooting height and the optimal shooting distance.
[0047] By accurately analyzing the height differences of components and matching the optimal shooting height and distance, the problem of components being too small due to excessive camera height or overflowing due to excessive distance is avoided. This ensures that components on the side and under the vehicle at different heights can be clearly presented in detail in the images, providing a high-quality image foundation for subsequent component defect identification. Step S302: Based on the passing image of the inspection robot under the current acquisition path, extract the actual path width pixel value and the robot's passing speed value of the image; The testing site refers to the physical area where vehicle side and undercarriage data collection is conducted. Common scenarios include rail vehicle maintenance garages and temporary outdoor testing stations. The site may contain fixed equipment, temporary obstacles, and other factors affecting the path. The full-area testing topographic map is a 3D map of the testing site generated using laser scanning and image modeling technology, including geographical information such as site boundaries, track positions, equipment coordinates, and obstacle distribution. The preset path analysis model is a path planning algorithm model developed based on Dijkstra's algorithm, which can generate the optimal path based on obstacle information and data collection efficiency requirements in the topographic map. Backup paths refer to 2-3 additional alternative paths generated by the system to meet the data collection requirements, in addition to the default main path. Each backup path must cover all data collection areas of the main path while avoiding obstacles. Personnel selection and switching refers to the process where staff can manually select and switch to a backup path based on the actual site conditions by adjusting the human-computer interaction interface of the control module, viewing the route map, obstacle prompts, and estimated data collection time for each path.
[0048] Step S303: Compare the actual values with the standard parameters to calculate the width deviation and speed deviation. If both are within the preset allowable deviation range, the current path is deemed to meet the requirements. If they exceed the range, the path selection is fine-tuned according to the deviation values and the path is re-verified until the path parameters meet the standard requirements.
[0049] The pre-set rail vehicle template library is a database storing structural parameters, component distribution, and inspection standards for different types of rail vehicles. It includes 3D vehicle models, coordinates of each component, a full inspection list, and a list of specialized inspection items, such as those for the braking system and bogies. Image analysis refers to the adjustment and control module extracting features from the acquired vehicle exterior images and comparing them with the vehicle models in the template library to determine the model of the vehicle to be inspected and the corresponding inspection scope. Inspection types include full inspection and specialized inspection: full inspection involves comprehensive collection and inspection of all key components on the vehicle's sides and undercarriage, such as doors, windows, bogies, braking systems, and suspension components; specialized inspection involves collection and inspection of only a specific system or component of the vehicle. The full inspection path is the complete path formed by integrating the inspection robot's default collection path with additional paths. Additional paths are supplementary paths covering full inspection components not included in the default path. The specialized testing path refers to a targeted path formed by segmenting and splitting the default data collection path and the specialized testing path. Segmentation and splitting means deleting road segments in the default path that are not related to the specialized testing, retaining and extending the road segments that are related to the specialized testing, and ensuring coverage of all testing points of the specialized components.
[0050] The adjustment and control module is also equipped with a strategy to enhance acquisition accuracy, including: Step S400: Vehicle number identification and positioning step, based on the image of the front of the rail vehicle, the vehicle number text is extracted and corrected to determine the vehicle number information; This step aims to improve the accuracy and reliability of vehicle number information during inspections, thereby providing accurate positioning data for subsequent component image acquisition. The core of vehicle number recognition and positioning lies in high-precision analysis of images of the front area of the rail vehicle. This includes utilizing advanced image processing and machine learning techniques, such as convolutional neural networks (CNNs), to extract features and recognize patterns from the acquired front images, thereby accurately detecting the front area.
[0051] After identifying the front area of the train, the extraction and correction of the train number text begins. During operation, the front image of the rail vehicle may be affected by factors such as changes in lighting, shooting angle, and dirt, leading to blurred, distorted, or even partially missing train number text. To address these challenges, this step employs multimodal image analysis and optical character recognition (OCR) technology. First, image enhancement algorithms, such as contrast stretching, noise reduction, and brightness adjustment, are used to restore the clarity of the train number text to the maximum extent possible. Then, an advanced OCR engine is used to locate the text region and recognize characters in the enhanced image.
[0052] Building upon text extraction, further correction of vehicle registration numbers is performed. This involves logical verification and error correction of the identified vehicle registration numbers. For example, based on industry standards or a pre-set vehicle model database, the format and content of the identified vehicle registration numbers are matched to eliminate obvious recognition errors. Furthermore, contextual information, such as the numbering patterns of adjacent trains, can be utilized to assist in correcting errors in individual vehicle registration numbers. Ultimately, through refined identification, extraction, and correction, the accuracy of the acquired vehicle registration number information is ensured, providing crucial identification for subsequent asset management and maintenance.
[0053] Step S401: Task image selection step. Based on the preset feature point extraction algorithm, feature points are extracted from the continuously acquired component images, and the task image with the smallest average distance in the template database corresponding to the vehicle number information is selected as the current inspection task image for image acquisition.
[0054] In a continuously acquired sequence of component images, feature points are extracted and matched using a pre-defined feature point extraction algorithm to obtain feature descriptions related to the component objects. Specifically, stable feature point detection and descriptors (such as a combination of SIFT / ORB algorithms) are used to consistently describe images of the same component under different angles and lighting conditions. The extracted feature descriptors are then compared one by one with template images in a pre-established task image template database, calculating the average distance or matching probability. The template image with the closest match and highest matching degree to the current sequence is selected as the current inspection task image, serving as the benchmark template for subsequent image acquisition and comparative analysis. If the matching degree is lower than a set threshold, resampling is triggered or the preliminary positioning results are returned for manual verification. This step ensures high consistency in object recognition for the inspection task, providing a reliable target reference for subsequent image acquisition and parameter alignment.
[0055] The configuration of device collaboration strategies involves the following steps: Step S500: Shooting and Coordination Step. Feature recognition is performed based on the component images on the acquisition path. The lifting and adjusting module triggers the lifting and adjusting command and adjusts the height according to the optimal shooting distance. At the same time, the image shooting command is triggered. Based on feature recognition of component images along the acquisition path, the lifting adjustment module triggers a lifting adjustment command according to the current recognition results and target shooting requirements, adjusting the height to the optimal shooting distance while simultaneously issuing an image capture command. Specifically, the system calculates a target shooting height and attitude parameters by integrating the current path position, component size, distance sensor information, and lens focal length and depth-of-field requirements. Subsequently, the lifting mechanism is controlled to smoothly perform fine-tuning of the height, ensuring the camera lens is at the optimal shooting distance, and dynamically optimizing focal length and exposure parameters as needed. After adjustment, an image capture command is sent, completing the acquisition of a set of high-quality images. The core objective of this step is to ensure that key components can obtain clear and comparable image data under different postures and environments, improving the reliability of subsequent recognition and measurement.
[0056] Step S501: Obstacle avoidance coordination step. When the inspection robot moves along the collection path, it identifies obstacles on the path based on the obstacle recognition model, generates an obstacle avoidance adjustment distance in the height direction of the obstacle, and triggers the lifting module to adjust the obstacle avoidance.
[0057] As the inspection robot moves along the data collection path, it identifies obstacles ahead using an obstacle recognition model and generates avoidance strategies. These strategies are then implemented through a lifting module to avoid obstacles in the vertical direction. Specifically, the system first uses sensors and a computer vision model to perceive the area ahead in real time, identifying movable and static obstacles and their height information. Based on the obstacle's height, distance, and movement trend, the minimum safe avoidance distance in the vertical direction is calculated. Combined with the robot's current posture and workload, the lifting adjustment amount and timing are determined. The lifting module is then triggered to adjust the height, and if necessary, movement control is activated to ensure safe distances and continuous image capture. This collaborative strategy ensures that the robot maintains a stable shooting posture and unobstructed path even in complex scenarios, reducing data quality degradation caused by collisions or abnormal postures.
[0058] The main body of the acquisition device is also equipped with a camera position pre-calibration strategy, which adopts the following steps and methods: Step S600: Determine the standard coordinate axis orientation based on the direction of movement of the rail vehicle, perform coordinate analysis on the panoramic cameras located at both ends of the multi-type camera modules carried by the inspection robot, and generate a baseline connecting the coordinates of the two cameras. By establishing a unified coordinate system based on the movement direction of the rail vehicle, the coordinates of the panoramic cameras at both ends of the inspection robot are analyzed, and a baseline for subsequent calibration is generated, providing coordinate reference for the detection of 3D camera position errors. The core principle of standard coordinate axis orientation is to maintain consistency with the direction of movement of the rail vehicle, ensuring that the coordinate system accurately maps the relative positions of the vehicle and the camera. Typically, the direction of the rail vehicle's movement is taken as the positive X-axis, the direction perpendicular to the track extension is taken as the positive Y-axis, and the vertically upward direction is taken as the positive Z-axis. This coordinate system needs to be associated with fixed reference points at the testing site, such as positioning stakes beside the track, to ensure the uniqueness of the coordinate origin and scale. The panoramic cameras at both ends refer to the cameras installed at both ends of the inspection robot's body along the X-axis. Their main function is to provide coordinate references. Because they are mechanically fixed in preset positions during installation, their coordinate stability is high. Typically, the front panoramic camera is located at the front of the robot, and the rear panoramic camera is located at the rear of the robot.
[0059] Coordinate analysis refers to obtaining the three-dimensional coordinate values of the two panoramic cameras in a standard coordinate system using a laser positioning instrument or preset installation coordinate parameters.
[0060] Step S601: Analyze the point-to-line distance based on the coordinates of the baseline and the corresponding coordinates of the 3D camera at the end of the robotic arm located in the middle, and determine the error distance between the 3D camera and the baseline. Point-to-line distance analysis refers to calculating the vertical distance from the center point of a 3D camera lens to a baseline based on the mathematical formula for the distance from a spatial point to a straight line. This distance directly reflects the positional deviation of the 3D camera relative to the baseline. If the distance is 0, it means that the 3D camera is perfectly aligned with the baseline and there is no positional deviation. If the distance is greater than 0, the deviation needs to be corrected through subsequent compensation. The 3D camera at the end of the robotic arm located in the middle refers to the camera installed at the end of the robotic arm in the middle of the inspection robot's body. It is mainly used for three-dimensional position detection of the components on the side and under the vehicle. Because there may be slight offsets in the installation of the robotic arm or deformation after long-term use, the actual position of the camera is prone to deviate from the baseline, which needs to be quantified by error distance.
[0061] Error distance refers to the vertical distance from the center point of the 3D camera lens to the baseline. To calculate it, you need to first obtain the actual coordinates of the 3D camera in the standard coordinate system and then substitute them into the formula for the distance from a point in space to a line. The result is the error distance.
[0062] Step S602: Generate position compensation coefficients based on the error distance and a preset position compensation model, and correct the position values of the components detected by the 3D camera according to the position compensation coefficients to ensure that the acquisition position deviation is within a preset small error range.
[0063] The position compensation model refers to the mathematical model preset by the system to convert the error distance into the position compensation coefficient. Common models include linear compensation models (compensation coefficient = 1 - error distance / maximum permissible error) or nonlinear fitting models (establishing a mapping relationship between error distance and compensation coefficient based on historical calibration data). The model needs to be calibrated and verified multiple times to ensure compensation accuracy.
[0064] The position compensation coefficient is a parameter used to correct the position detection values of 3D camera components. Its value range is usually between 0.9 and 1.1. The larger the error distance, the greater the deviation of the compensation coefficient from 1. When the error distance is 0, the compensation coefficient is 1.
[0065] The preset small error range refers to the maximum deviation of the acquisition position allowed by the system. It is set according to the accuracy requirements of the detection of rail vehicle components, and is usually ±0.5 mm. The corrected component position value must fall within this range to ensure that subsequent image acquisition can accurately align with the component.
[0066] The collaborative control module is also equipped with a light source optimization strategy, including: Step S700: Perform feature analysis on the vehicle side and undercarriage edge images acquired by multiple types of camera modules to determine the image shooting area and match the optimal illumination intensity of the corresponding camera model in the preset illumination database. Feature analysis refers to the analysis of pixel distribution, grayscale differences, and contour continuity of edge images. By identifying the concentrated areas of edge pixels, the specific location and extension range of the component edge can be determined, providing a basis for dividing the shooting area. The image capture area refers to the image range that includes the complete edge of the component and its surrounding related areas. It is necessary to ensure that all details of the edge features are covered. It is usually a strip-shaped area extending a certain number of pixels on both sides of the edge. The specific range is adjusted according to the size of the component. For example, the image capture area of the car door edge should cover 40-60 pixels on each side of the edge to avoid missing edge details due to the area being too small or introducing irrelevant background interference due to the area being too large.
[0067] The preset lighting database refers to a data set that stores the correspondence between different camera models and their corresponding lighting intensities. The database is classified according to the models of various types of camera modules and records the effective lighting intensity range of each camera when shooting the edges of different components, providing data support for matching the optimal value.
[0068] Optimal illumination intensity refers to the illumination intensity value that enables the edges of components within the shooting area to form a clear grayscale difference with the background, without obvious reflections or shadows. This value needs to be adapted to the light-sensing characteristics of the corresponding camera model to ensure rich grayscale levels in the image.
[0069] Step S701: Adjust the light source in the shooting area according to the optimal light intensity and the preset parallel illumination angle, calculate the light source adjustment gain value using the preset light source gain analysis model, and keep the gain value within the preset reference gain value range.
[0070] The preset parallel illumination angle refers to the fixed angle formed between the direction of the light source and the surface of the vehicle side and bottom components. This angle needs to be set in combination with the material of the component, and is usually set to 40°-50°. The purpose is to reduce the specular reflection produced when the light source is vertically illuminating, such as the strong light reflection on the surface of the metal bogie, and at the same time avoid the angle being too small, which would cause obvious shadows in the shooting area, such as the shadow under the car door obscuring the details of the gap. The light source adjustment gain value is a parameter used to fine-tune the actual output intensity of the light source. This parameter compensates for the influence of ambient light fluctuations or differences in surface reflection of components on the light intensity, ensuring that the actual light intensity is stable and close to the optimal value. The preset reference gain value range refers to the normal operating range of the gain value set by the system. This range is determined according to the output capability of the light source device and the light sensitivity requirements of the camera, and is usually 0.8-1.2. If the gain value exceeds the range, it will affect the image quality when the light source output is too strong or too weak.
[0071] The light source gain analysis model is calculated using the following formula:
[0072] Wherein, G is the light source adjustment gain value, which comprehensively reflects the enhancement coefficient of train components. The grayscale standard deviation of the component image after light source adjustment is used to quantify grayscale uniformity. The grayscale standard deviation of the component image before light source adjustment. This refers to the actual light transmittance of the camera lens. Set a baseline transmittance for the lens. This represents the actual reflection coefficient of the component. A baseline reflection coefficient is preset for the factory inspection values of the components. This refers to the actual exposure time of the camera. Set a baseline exposure time for the camera. This refers to the actual distance between the lens and the components. Preset a baseline acquisition distance for the lens and components. This is the preset ambient light interference correction factor. The actual ambient light intensity of the set segment field, Indicates the preset reference ambient light intensity. Indicates the contrast-edge co-correction coefficient. This indicates the amount of image contrast improvement after adjusting the light source. This is a preset baseline contrast, representing the minimum requirement for recognizing component details. This indicates the average grayscale gradient at the edge of the component after adjustment. The preset baseline grayscale gradient represents the threshold for edge sharpness. To compensate for the edge detection error of the parts, To preset the baseline edge detection error, This is the preset baseline edge detection error.
[0073] Based on the same inventive concept, embodiments of the present invention provide a supplementary lighting system, comprising: The supplementary lighting execution module is located next to the main body of the acquisition device of the multi-dimensional adjustable vehicle side and undercarriage acquisition device, and is arranged correspondingly to the multi-type camera modules of the acquisition device main body. It is used to output illumination adapted to the acquisition of images of vehicle side and undercarriage components. The supplementary lighting execution module can adjust the light intensity and illumination angle to adapt to the field of view of the multi-type camera modules. The supplementary lighting control module is connected to the supplementary lighting execution module and the collaborative control module of the main body of the acquisition device, respectively. It is configured with the light source optimization strategy preset by the collaborative control module. The supplementary lighting control module is used to receive the vehicle side and vehicle bottom edge images acquired by the multi-type camera modules, perform feature analysis on the images to determine the image shooting area, and match the optimal light intensity corresponding to the model of the multi-type camera module in the preset lighting database. The gain calculation submodule, integrated into the supplementary lighting control module, calculates the light source adjustment gain value according to the preset light source gain analysis model, and determines whether the gain value is within the preset reference gain value range. The collaborative adaptation module is connected to the supplementary lighting control module, the adjustment control module of the main body of the acquisition device, and the lifting adjustment module, respectively. It is used to receive the multi-dimensional adjustment strategy and acquisition accuracy enhancement strategy results issued by the adjustment control module, synchronously send the illumination adjustment trigger signal to the supplementary lighting control module, and receive the lifting adjustment in place signal from the lifting adjustment module. After the lifting adjustment module adjusts to the optimal shooting distance, it triggers the supplementary lighting execution module to output the adapted illumination. The parameter storage module, connected to the supplementary lighting control module, is used to store the preset illumination database, light source gain analysis model parameters, and historical illumination adjustment data, which are then called by the supplementary lighting control module to optimize the illumination adjustment accuracy.
[0074] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional modules is used as an example. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. The specific working process of the system, device, and unit described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.
[0075] This invention provides a computer-readable storage medium storing a computer program that can be loaded and executed by a processor.
[0076] Computer storage media include, for example, USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks, and other media capable of storing program code.
[0077] Based on the same inventive concept, embodiments of the present invention provide a smart terminal, including a memory and a processor, wherein the memory stores a computer program that can be loaded by the processor and executed using the xxx method.
[0078] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional modules is used as an example. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. The specific working process of the system, device, and unit described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.
[0079] The above are all preferred embodiments of this application and are not intended to limit the scope of protection of this application. Any feature disclosed in this specification (including the abstract and drawings) may be replaced by other equivalent or similar features unless specifically stated otherwise. That is, unless specifically stated otherwise, each feature is only one example of a series of equivalent or similar features.
Claims
1. A multi-dimensional adjustable vehicle side and undercarriage data acquisition device, characterized in that, include: The main body of the acquisition device is arranged along the preset track vehicle running path and is equipped with an inspection robot, multiple types of camera modules and a lifting adjustment module. The inspection robot carries the multiple types of camera modules and moves along the preset acquisition path to acquire component images of different areas on the side and under the vehicle. The lifting adjustment module is used to adjust the acquisition height of the multiple types of camera modules. An adjustment and control module, connected to the main body of the acquisition device, is configured with a multi-dimensional adjustment strategy to adjust the image acquisition parameters of the various types of camera modules and the acquisition path of the inspection robot, so as to adapt to the acquisition needs of different components on the side and under the vehicle. The collaborative control module is connected to the main body of the acquisition device and the adjustment control module respectively, and is configured with a device collaboration strategy to coordinate the working sequence of the inspection robot, the multi-type camera module and the lifting and adjusting module.
2. The multi-dimensional adjustable vehicle side and undercarriage data acquisition device according to claim 1, characterized in that, The multi-dimensional adjustment strategy includes: The height adjustment sub-strategy analyzes the height differences of vehicle side and undercarriage components based on the acquisition path to generate the target component shooting height, and generates the optimal shooting distance adapted to the shooting of the multiple types of camera modules based on the target component shooting height; The path adjustment sub-strategy generates a full-area topographic map based on the detection site and generates backup paths for data collection using a preset path analysis model for personnel to select and switch between. The range adjustment sub-strategy performs image analysis based on a preset rail vehicle template library to determine the detection type of the vehicle to be inspected. When the detection type is full inspection, the full inspection path is generated by integrating the acquisition path and the additional path. When the detection type is special inspection, the special inspection path is generated by segmenting the acquisition path and the special inspection path.
3. The multi-dimensional adjustable vehicle side and undercarriage data acquisition device according to claim 2, characterized in that, The path adjustment sub-strategy is also configured with a path switching verification sub-strategy, including: The optimal path width pixel range and traffic flow threshold are determined based on the track size parameters of the depot, and the track size parameters correspond to the actual layout of the rail vehicle depot. Based on the current image of the inspection robot along the current acquisition path, extract the actual path width pixel value and the robot's travel speed value. The actual values are compared with the standard parameters to calculate the width deviation and speed deviation. If both are within the preset allowable deviation range, the current path is deemed to meet the requirements. If they exceed the range, the path selection is fine-tuned according to the deviation values and the path is re-verified until the path parameters meet the standard requirements.
4. The multi-dimensional adjustable vehicle side and undercarriage data acquisition device according to claim 2, characterized in that, The adjustment and control module is also equipped with a strategy to enhance the acquisition accuracy, including: The vehicle number identification and positioning steps involve identifying and analyzing the front image of the rail vehicle and extracting and correcting the vehicle number text to determine the vehicle number information. The task image selection step involves extracting feature points from continuously acquired component images using a preset feature point extraction algorithm, and then searching for the task image with the smallest average distance in the template database corresponding to the vehicle number information as the current inspection task image for image acquisition.
5. The multi-dimensional adjustable vehicle side and undercarriage data acquisition device according to claim 2, characterized in that, The device collaboration strategy includes: In the collaborative shooting step, feature recognition is performed based on the component images along the acquisition path. The lifting adjustment module triggers the lifting adjustment command and adjusts the lifting according to the optimal shooting distance, while simultaneously triggering the image shooting command. In the obstacle avoidance coordination step, when the inspection robot moves along the collection path, it identifies obstacles on the path based on the obstacle recognition model, generates an obstacle avoidance adjustment distance in the height direction of the obstacle, and triggers the lifting module to adjust the obstacle avoidance.
6. The multi-dimensional adjustable vehicle side and undercarriage data acquisition device according to claim 1, characterized in that, The main body of the acquisition device is also equipped with a camera position pre-calibration strategy, including: Based on the direction of movement of the rail vehicle, a standard coordinate axis orientation is determined, and coordinate analysis is performed on the panoramic cameras located at both ends of the multi-type camera modules carried by the inspection robot to generate a baseline line connecting the coordinates of the two cameras. Based on the coordinates of the baseline and the corresponding coordinates of the 3D camera at the end of the robotic arm located in the middle, a point-to-line distance analysis is performed to determine the error distance between the 3D camera and the baseline. Based on the error distance and a preset position compensation model, a position compensation coefficient is generated. The position value of the component detected by the 3D camera is corrected according to the position compensation coefficient to ensure that the acquisition position deviation is within a preset small error range.
7. The multi-dimensional adjustable vehicle side and undercarriage data acquisition device according to claim 1, characterized in that, The collaborative control module is also configured with a light source optimization strategy, including: Based on the feature analysis of the vehicle side and underside edge images acquired by the multi-type camera modules, the image shooting area is determined, and the optimal light intensity of the corresponding camera model is matched with the preset light database. The light source in the shooting area is adjusted according to the optimal light intensity and the preset parallel illumination angle. The light source adjustment gain value is calculated using a preset light source gain analysis model, and the gain value is kept within the preset reference gain value range.
8. The multi-dimensional adjustable vehicle side and undercarriage data acquisition device according to claim 7, characterized in that, The light source gain analysis model is calculated using the following formula: Wherein, G is the light source adjustment gain value, which comprehensively reflects the enhancement coefficient of train components. The grayscale standard deviation of the component image after light source adjustment is used to quantify grayscale uniformity. The grayscale standard deviation of the component image before light source adjustment. This refers to the actual light transmittance of the camera lens. Set a baseline transmittance for the lens. This represents the actual reflection coefficient of the component. A baseline reflection coefficient is preset for the factory inspection values of the components. This refers to the actual exposure time of the camera. Set a baseline exposure time for the camera. This refers to the actual distance between the lens and the components. Preset a baseline acquisition distance for the lens and components. This is the preset ambient light interference correction factor. The actual ambient light intensity of the set segment field, Indicates the preset reference ambient light intensity. Indicates the contrast-edge co-correction coefficient. This indicates the amount of image contrast improvement after adjusting the light source. This is a preset baseline contrast, representing the minimum requirement for recognizing component details. This indicates the average grayscale gradient at the edge of the component after adjustment. The preset baseline grayscale gradient represents the threshold for edge sharpness. To compensate for the edge detection error of the parts, To preset the baseline edge detection error, This is the preset baseline edge detection error.
9. A supplementary lighting system, applied to the multi-dimensional adjustable vehicle side and under-vehicle data acquisition device as described in any one of claims 1-8, characterized in that, include: The supplementary lighting execution module is located next to the main body of the acquisition device of the multi-dimensional adjustable vehicle side and undercarriage acquisition device, and is arranged correspondingly to the multi-type camera modules of the acquisition device main body. It is used to output illumination adapted to the acquisition of images of vehicle side and undercarriage components. The supplementary lighting execution module can adjust the light intensity and illumination angle to adapt to the field of view of the multi-type camera modules. The supplementary lighting control module is connected to the supplementary lighting execution module and the collaborative control module of the main body of the acquisition device, respectively. It is configured with the light source optimization strategy preset by the collaborative control module. The supplementary lighting control module is used to receive the vehicle side and vehicle bottom edge images acquired by the multi-type camera modules, perform feature analysis on the images to determine the image shooting area, and match the optimal light intensity corresponding to the model of the multi-type camera module in the preset lighting database. The gain calculation submodule, integrated into the supplementary lighting control module, calculates the light source adjustment gain value according to the preset light source gain analysis model, and determines whether the gain value is within the preset reference gain value range. The collaborative adaptation module is connected to the supplementary lighting control module, the adjustment control module of the main body of the acquisition device, and the lifting adjustment module, respectively. It is used to receive the multi-dimensional adjustment strategy and acquisition accuracy enhancement strategy results issued by the adjustment control module, synchronously send the illumination adjustment trigger signal to the supplementary lighting control module, and receive the lifting adjustment in place signal from the lifting adjustment module. After the lifting adjustment module adjusts to the optimal shooting distance, it triggers the supplementary lighting execution module to output the adapted illumination. The parameter storage module, connected to the supplementary lighting control module, is used to store the preset illumination database, light source gain analysis model parameters, and historical illumination adjustment data, which are then called by the supplementary lighting control module to optimize the illumination adjustment accuracy.