Panoramic detection device and method for railway vehicle and railway vehicle

Abstract

The invention provides a panoramic detection device and method for a railway vehicle and the railway vehicle. The panoramic detection device comprises a top support, a bottom support and two oppositely-arranged side supports which form a detection space in a surrounding mode. The plurality of image acquisition modules are respectively arranged on the top bracket, the bottom bracket and the two side brackets and are used for carrying out panoramic image acquisition on the railway vehicle passing through the detection space; and the posture adjusting modules are arranged in one-to-one correspondence with the image acquisition modules and are used for adjusting the acquisition angles of the image acquisition modules. By arranging the top support, the side supports and the bottom support, a relatively closed image acquisition space for performing panoramic detection on the railway vehicle is formed, image information of the railway vehicle can be better focused and captured, meanwhile, automatic adjustment of acquisition of the image acquisition module is achieved through the posture adjustment module, and the image acquisition efficiency is improved. Manual operation is not needed in the whole process, and dependence on manpower is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of rail vehicles, and in particular to a panoramic detection device and method for a rail vehicle and a rail vehicle. Background Art

[0002] Train operation and maintenance is a key link in the entire life cycle of a train. It can identify and deal with abnormal conditions such as component fatigue defects, deformation, missing parts and foreign objects generated during the service of rail vehicles. Traditional maintenance operations are mainly based on manual visual inspections. This operation mode has the disadvantages of long time, low efficiency, strong dependence on manual labor, and the possibility of missed inspections, which hinders the development of intelligent train operation and maintenance. Summary of the invention

[0003] The present invention provides a panoramic detection device and method for a rail vehicle and a rail vehicle, which are used to solve the problem that in the operation and maintenance process of the rail vehicle in the prior art, the rail vehicle is highly dependent on manual labor and cannot be manually collected for the panoramic image of the rail vehicle under extreme working conditions.

[0004] According to a first aspect of the present invention, a panoramic detection device for a rail vehicle is provided, comprising: a top bracket, a bottom bracket and two oppositely arranged side brackets surrounding a detection space; a plurality of image acquisition modules, respectively arranged on the top bracket, the bottom bracket and the two side brackets, for performing panoramic image acquisition on the rail vehicle passing through the detection space; and a posture adjustment module, arranged in a one-to-one correspondence with the image acquisition module, for adjusting the acquisition angle of the image acquisition module.

[0005] According to one embodiment of the present invention, the top bracket includes: a mounting beam, which is used to be installed in a mounting position; and a suspension beam, which can be movably installed on the mounting beam; wherein the image acquisition module for acquiring the top image of the rail vehicle is installed on the suspension beam.

[0006] Specifically, this embodiment provides an implementation of a top bracket.

[0007] According to one embodiment of the present invention, the side bracket includes: a base; a roller installed on the base; a vertical beam installed on the base, the vertical beam includes a plurality of vertical segments, and the plurality of vertical segments are plugged into each other to adjust the height of the vertical beam; a support beam hinged between the vertical beam and the base; wherein the image acquisition module for acquiring the side image of the rail vehicle is installed on the vertical beam.

[0008] Specifically, this embodiment provides an implementation of a side bracket.

[0009] According to one embodiment of the present invention, the image acquisition module includes: a two-dimensional image acquisition element, a three-dimensional image acquisition element, a speed measuring radar, a vehicle number recognition component and a fill light; the two-dimensional image acquisition element is used to acquire a two-dimensional image of the rail vehicle; the three-dimensional image acquisition element is used to acquire a three-dimensional image of the rail vehicle; the speed measuring radar and the vehicle number recognition component are installed on the vertical beam; the fill light is installed on the side of the vertical beam facing the bottom bracket.

[0010] Specifically, this embodiment provides an implementation of an image acquisition module.

[0011] According to one embodiment of the present invention, the posture adjustment module includes: a fixed plate; a first movable plate, which is rotatably arranged on the fixed plate around a first axis, and the first axis is perpendicular to the first movable plate; a first driving mechanism, which is suitable for driving the first movable plate to rotate relative to the fixed plate; a second movable plate, which is suitable for installing an image acquisition module, the second movable plate is connected to the first movable plate, and the second movable plate can at least rotate around a second axis and a third axis respectively relative to the first movable plate, the second axis is perpendicular to the third axis, and the second axis and the third axis are both perpendicular to the first axis; a second driving mechanism, which is suitable for driving the second movable plate to rotate around the second axis relative to the first movable plate; and a third driving mechanism, which is suitable for driving the second movable plate to rotate around the third axis relative to the first movable plate.

[0012] Specifically, this embodiment provides an implementation of a posture adjustment module.

[0013] According to one embodiment of the present invention, the first driving mechanism includes: two groups of first temperature-dependent deformation rods, which are respectively connected to a power source; the first ends of the first temperature-dependent deformation rods are connected to the fixed plate, and the second ends of the first temperature-dependent deformation rods are connected to the first movable plate; wherein the axes of the two groups of the first temperature-dependent deformation rods are parallel to each other and perpendicular to the first axis, and the two groups of the first temperature-dependent deformation rods are respectively located on both sides of the first axis.

[0014] Specifically, this embodiment provides an implementation of a first driving mechanism.

[0015] According to one embodiment of the present invention, the second driving mechanism includes: two groups of second temperature-dependent deformation rods, which are respectively connected to the power supply; the first end of the second temperature-dependent deformation rod is connected to the first movable plate, and the second end of the second temperature-dependent deformation rod is connected to the second movable plate; wherein the axes of the two groups of second temperature-dependent deformation rods are parallel to the first axis, and the two groups of second temperature-dependent deformation rods are respectively located on both sides of the second axis.

[0016] Specifically, this embodiment provides an implementation of a second driving mechanism.

[0017] According to one embodiment of the present invention, the third driving mechanism includes: two groups of third temperature-dependent deformation rods, which are respectively connected to the power supply; the first end of the third temperature-dependent deformation rod is connected to the first movable plate, and the second end of the third temperature-dependent deformation rod is connected to the second movable plate; wherein the axes of the two groups of the third temperature-dependent deformation rods are parallel to the first axis, and the two groups of the third temperature-dependent deformation rods are respectively located on both sides of the third axis.

[0018] Specifically, this embodiment provides an implementation of a third driving mechanism.

[0019] According to an embodiment of the present invention, the posture adjustment module further includes: a posture sensor, which is arranged on the second movable plate and is suitable for detecting the three-dimensional posture of the second movable plate.

[0020] Specifically, this embodiment provides an implementation of a gesture sensor.

[0021] According to one embodiment of the present invention, the first temperature-dependent deformation rod, the second temperature-dependent deformation rod and the third temperature-dependent deformation rod all include: a temperature-dependent deformation rod body, the length of which changes with temperature; an insulating member disposed at both ends of the temperature-dependent deformation rod body; and a temperature sensor disposed inside the temperature-dependent deformation rod body, the temperature sensor being suitable for detecting the temperature of the temperature-dependent deformation rod body.

[0022] Specifically, this embodiment provides an implementation of a first temperature-dependent deformation rod, a second temperature-dependent deformation rod, and a third temperature-dependent deformation rod.

[0023] According to one embodiment of the present invention, the posture adjustment module also includes: a connecting rod, the axis of the connecting rod coincides with the first axis; a universal connector, arranged between the first movable plate and the second movable plate; wherein the first end of the universal connector is connected to the first movable plate through the connecting rod, and the second end of the universal connector is fixedly connected to the second movable plate.

[0024] Specifically, this embodiment provides an implementation of a connecting rod and a universal connector.

[0025] A panoramic detection method for a rail vehicle according to a second aspect of the present invention is applied to the panoramic detection device for a rail vehicle described above; The method comprises: In response to a start signal of the panoramic detection of the rail vehicle, image features of the rail vehicle in the detection space are acquired, the image features at least comprising image information collected from the top, side and bottom of the rail vehicle; Based on the image features, a panoramic detection is performed on the rail vehicle.

[0026] According to an embodiment of the present invention, the image features include at least a standard image of a rail vehicle and an image to be detected, the standard image is an image of a rail vehicle captured by an image acquisition module when the rail vehicle moves at a standard speed, and the image to be detected is an image of a rail vehicle captured by an image acquisition module when the rail vehicle moves at an actual speed; The step of performing panoramic detection on the rail vehicle based on the image features specifically includes: Determining a line frequency difference according to the line frequency of the standard image and the line frequency of the image to be detected; Establishing a line frequency difference timing prediction model data set based on the line frequency difference, wherein the line frequency difference timing prediction model data set includes a plurality of data pairs, wherein the data pairs include a standard speed of a historical image and the line frequency difference; The line frequency difference timing prediction model is trained according to the line frequency difference timing prediction model data set to obtain a trained line frequency difference timing prediction model; Based on the trained line frequency difference timing prediction model, the line frequency of the image acquisition module is controlled by adopting a line frequency compensation method.

[0027] Specifically, this embodiment provides an implementation method for performing panoramic detection on the rail vehicle.

[0028] According to an embodiment of the present invention, determining the line frequency difference according to the line frequency of the standard image and the line frequency of the image to be detected includes: Acquire a standard image and an image to be detected; Performing feature extraction and feature matching on the standard image and the image to be detected to obtain a spatial distribution map of feature pairs, wherein the feature pairs include feature points on the standard image and matching feature points on the image to be detected; Determining the line frequency of the standard image and the line frequency of the image to be detected according to the spatial distribution diagram of the feature pairs; Converting the line frequency of the image to be detected into the actual line frequency at the standard speed; The line frequency difference is obtained by subtracting the line frequency of the standard image from the actual line frequency at the standard speed.

[0029] Specifically, this embodiment provides an implementation method for determining the line frequency difference according to the line frequency of the standard image and the line frequency of the image to be detected.

[0030] According to an embodiment of the present invention, the step of extracting and matching features of the standard image and the image to be detected to obtain a spatial distribution map of feature pairs includes: Using a threshold segmentation algorithm to align the standard image with one end of the image to be detected; Using the SFIT feature extraction algorithm to extract feature points from the standard image and the image to be detected, respectively, to obtain feature points on the standard image and feature points on the image to be detected; The DMatch operator is used to pair the feature points on the standard image with the features on the image to be detected to obtain a spatial distribution map of the feature pairs.

[0031] Specifically, this embodiment provides an implementation method for performing feature extraction and feature matching on the standard image and the image to be detected to obtain a spatial distribution map of feature pairs.

[0032] According to an embodiment of the present invention, determining the line frequency of the standard image and the line frequency of the image to be detected according to the spatial distribution diagram of the feature pair includes: Determining the line frequency of the standard image according to the distance and time difference between any two feature points in the standard image; The line frequency of the image to be detected is determined according to the distance and time difference between any two feature points in the image to be detected.

[0033] Specifically, this embodiment provides an implementation method for determining the line frequency of the standard image and the line frequency of the image to be detected according to the spatial distribution diagram of the feature pairs.

[0034] According to an embodiment of the present invention, the step of establishing a line frequency difference timing prediction model data set based on the line frequency difference includes: Read the speed data of the historical image in segments to determine an actual speed data sequence; Normalizing the actual speed data sequence to obtain a standard speed data sequence; The correlation relationship between the standard speed data sequence and the line frequency difference is established by using time distribution, and the line frequency difference timing prediction model data set is established. The line frequency difference timing prediction model data set includes a training set, a validation set and a test set.

[0035] Specifically, this embodiment provides an implementation method for establishing a line frequency difference timing prediction model data set based on the line frequency difference.

[0036] According to an embodiment of the present invention, the training of the line frequency difference timing prediction model according to the line frequency difference timing prediction model data set to obtain the trained line frequency difference timing prediction model includes: The LSTM model is trained according to the standard speed data sequence and the line frequency difference, and a cross entropy loss function is selected to obtain a trained line frequency difference time series prediction model.

[0037] Specifically, this embodiment provides an implementation method for training a line frequency difference timing prediction model according to the line frequency difference timing prediction model data set to obtain a trained line frequency difference timing prediction model.

[0038] According to an embodiment of the present invention, the line frequency control of the image acquisition module is performed by line frequency compensation based on the trained line frequency difference timing prediction model, including: Based on the trained line frequency difference timing prediction model, the line frequency compensation formula is used to control the line frequency of the image acquisition module. The line frequency compensation formula is: F't=Vt / Pa+ΔFs(t) / k Among them, F't is the line frequency controlled by the image acquisition module, Vt is the speed of the rail vehicle at time t, Pa is the lateral pixel accuracy, ΔFs(t) is the trained line frequency difference timing prediction model, and k is the conversion coefficient between the line frequency of the image to be detected and the line frequency of the standard image.

[0039] Specifically, this embodiment provides an implementation method for controlling the line frequency of the image acquisition module by using line frequency compensation based on the trained line frequency difference timing prediction model.

[0040] According to a third aspect of the present invention, a rail vehicle is provided, comprising the above-mentioned panoramic detection device for a rail vehicle; or, when performing panoramic detection of a rail vehicle, the above-mentioned panoramic detection method for a rail vehicle is adopted.

[0041] The above one or more technical solutions in the present invention have at least one of the following technical effects: a panoramic detection device for a rail vehicle provided by the present invention forms a relatively closed image acquisition space for panoramic detection of the rail vehicle by setting a top bracket, a side bracket and a bottom bracket, which helps to better focus and capture the image information of the rail vehicle. At the same time, the image acquisition module can be automatically adjusted through the posture adjustment module. The whole process does not require manual operation, which reduces the dependence on manual operation. Moreover, under extreme working conditions, there is no need to manually touch the posture adjustment module itself, which removes the restrictions of the working environment on the posture adjustment process, and solves the problem in the prior art that when adjusting the posture of the image acquisition module, it is highly dependent on manual operation and cannot be adjusted manually under extreme working conditions.

[0042] It should be noted that this structured design also enhances the stability and reliability of the system. In addition, the image data collected from different angles by multiple image acquisition modules can be complementary and compared, so as to more accurately identify and judge the status and problems of rail vehicles. The detection device can add more image acquisition modules or adjust the position and angle of existing modules as needed to meet the detection needs of rail vehicles of different types and specifications. This modular and scalable design makes the detection device more adaptable and versatile.

[0043] The present invention provides a panoramic detection method for a rail vehicle, which realizes fully automated detection of the panoramic view of the rail vehicle by collecting images of the top, side and bottom of the rail vehicle and extracting corresponding image features, thereby solving the problem that previous detection required manual assistance.

[0044] The present invention provides a rail vehicle, which has the above-mentioned beneficial effects and solves the corresponding technical problems. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0046] Figure 1 It is a schematic structural diagram of the panoramic detection device for rail vehicles provided by the present invention.

[0047] Figure 2 It is a schematic front view of a side bracket in the panoramic detection device for a rail vehicle provided by the present invention.

[0048] Figure 3 It is a schematic side view of a side bracket in the panoramic detection device for a rail vehicle provided by the present invention.

[0049] Figure 4 It is a structural schematic diagram of a posture adjustment module at a certain viewing angle in the panoramic detection device for a rail vehicle provided by the present invention.

[0050] Figure 5 It is a structural schematic diagram of a posture adjustment module in the panoramic detection device for a rail vehicle provided by the present invention at another viewing angle.

[0051] Figure 6 It is a schematic diagram of the connection structure between the first movable plate and the fixed plate in the panoramic detection device for a rail vehicle provided by the present invention.

[0052] Figure 7It is a schematic structural diagram between a first movable plate and a second movable plate in the panoramic detection device for a rail vehicle provided by the present invention.

[0053] Figure 8 It is a schematic structural diagram of a first temperature-dependent deformation rod, a second temperature-dependent deformation rod and a third temperature-dependent deformation rod in the panoramic detection device for a rail vehicle provided by the present invention.

[0054] Fig. 9 It is a flow chart of a posture adjustment method provided by an embodiment of the present invention.

[0055] Fig.10 It is a flow chart of the panoramic detection method of the rail vehicle of the present invention.

[0056] Fig.11 It is a schematic diagram of the principle of the panoramic detection method of the rail vehicle of the present invention.

[0057] Fig.12 It is a schematic diagram of constructing a first light source parameter matching table in an embodiment provided by the present invention.

[0058] Fig.13 It is a schematic diagram of constructing a second light source parameter matching table in an embodiment provided by the present invention.

[0059] Fig.14 It is a schematic diagram of constructing a light source parameter control model in an embodiment provided by the present invention.

[0060] Reference numerals: 10. Top bracket; 101. Mounting beam; 102. Overhanging beam; 20, side bracket; 201, base; 202, roller; 203, vertical beam; 204, support beam; 30. Bottom bracket; 40. Image acquisition module; 401. Two-dimensional image acquisition element; 402. Three-dimensional image acquisition element; 403. Speed ​​radar; 404. Vehicle number recognition element; 405. Fill light element; 50. Posture adjustment module; 501. Fixed plate; 502. First movable plate; 503. Second movable plate; 504. First temperature-dependent deformation rod; 505. Second temperature-dependent deformation rod; 506. Third temperature-dependent deformation rod; 507. Posture sensor; 508. Temperature-dependent deformation rod body; 509. Insulator; 510. Temperature sensor; 511. Universal joint; 512. Connecting rod; 513. Coil spring; 514. Thrust bearing. DETAILED DESCRIPTION

[0061] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0062] The present invention is described in detail below in conjunction with the accompanying drawings of the specification. The specific operating method in the method embodiment can also be applied to the device embodiment or the system embodiment. In the description of the present invention, unless otherwise specified, "at least one" includes one or more. "Multiple" refers to two or more. For example, at least one of A, B and C includes: A exists alone, B exists alone, A and B exist at the same time, A and C exist at the same time, B and C exist at the same time, and A, B and C exist at the same time. In the present invention, " / " means or, for example, A / B can mean A or B; "and / or" in this article is only a kind of association relationship that describes the associated objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone.

[0063] The present invention will be described in detail below in conjunction with specific implementation modes.

[0064] In some specific embodiments of the present invention, Figures 1 to 9 As shown, the present solution provides a panoramic detection device for a rail vehicle, comprising: a top bracket 10, a bottom bracket 30 and two oppositely arranged side brackets 20 that surround and form a detection space; a plurality of image acquisition modules 40, respectively arranged on the top bracket 10, the bottom bracket 30 and the two side brackets 20, for performing panoramic image acquisition on the rail vehicle passing through the detection space; a posture adjustment module 50, arranged one-to-one correspondingly to the image acquisition module 40, for adjusting the acquisition angle of the image acquisition module 40.

[0065] It should be noted that the panoramic detection of rail vehicle operation and maintenance, that is, the 360° panoramic detection of rail vehicles, the 360° panoramic detection device provided by the present invention can collect the appearance image information of the monitored object through the image acquisition module, and perform fault detection and troubleshooting through image recognition technology or manual. The use of 360° panoramic detection can realize intelligent operation and maintenance of trains, improve efficiency, reduce dependence on manual labor, and eliminate the problem of missed detection.

[0066] Furthermore, if Figure 1 and Figure 2 As shown, the setting positions and corresponding relationship of the image acquisition module 40 and the posture adjustment module 50, Figure 1 and Figure 2 Only the corresponding relationship between the image acquisition module 40 and the posture adjustment module 50 is shown, and it does not represent the structural shape, size, and size of the image acquisition module 40 and the posture adjustment module 50 .

[0067] Furthermore, the present invention has image acquisition modules 40 installed on the top bracket 10, the side bracket 20 and the bottom bracket 30 respectively. The image acquisition module 40 includes a two-dimensional image acquisition element 401, a three-dimensional image acquisition element 402, a speed measurement radar 403, a vehicle number recognition element 404 and a fill light 405, which can realize all-round image acquisition of rail vehicles. This multi-angle and multi-position acquisition method can ensure a detailed and comprehensive inspection of rail vehicles, and improve the detection efficiency and accuracy. A roller 202 is provided at the bottom of the side bracket 20, so that the entire detection device can be easily moved and adjusted at different positions to adapt to different environments and detection requirements. The design of the roller 202 also facilitates the rapid deployment and disassembly of the detection device, and improves the convenience of use. The top bracket 10, the side bracket 20 and the bottom bracket 30 are jointly enclosed to form a relatively closed image acquisition space, which helps to better focus and capture the image information of the rail vehicle. At the same time, this structured design also enhances the stability and reliability of the system. In addition, the image data collected from different angles by multiple image acquisition modules 40 can be complementary and compared, so as to more accurately identify and judge the status and problems of the rail vehicle. The detection device can add more image acquisition modules 40 or adjust the position and angle of the existing modules as needed to meet the detection needs of rail vehicles of different types and specifications. This modular and scalable design makes the detection device more adaptable and versatile.

[0068] In a possible embodiment, the panoramic detection device provided by the present invention is mainly composed of three parts: a top bracket 10, a side bracket 20 and a bottom bracket 30. A corresponding image acquisition module 40 is installed on each bracket, which work together to achieve comprehensive detection of rail vehicles, and the posture adjustment of the image acquisition module 40 is achieved through a posture adjustment module connected to the image acquisition module 40.

[0069] In a possible embodiment, the top bracket 10 is located at the top of the entire detection device, and an image acquisition module 40 is installed thereon, which is mainly used to acquire images of the top and upper part of the rail vehicle, including the roof, windows, antennas and other parts.

[0070] In a possible embodiment, the side bracket 20 is located below the top bracket 10, and a roller 202 is provided at the bottom thereof to facilitate the movement and positioning of the entire detection device. An image acquisition module 40 is also installed on the side bracket 20, which is mainly used to acquire images of the side of the rail vehicle, including the running gear, body, door, wheel and other parts.

[0071] In a possible embodiment, the bottom bracket 30 is located at the bottom of the side bracket 20, and an image acquisition module 40 is also installed thereon. It is mainly used to acquire images of the bottom and lower part of the rail vehicle, including key parts such as the bottom of the vehicle and the suspension system.

[0072] It can be understood that a closed image acquisition space is formed by the enclosure of the top bracket 10, the side bracket 20 and the bottom bracket 30. When the rail vehicle enters this space, multiple image acquisition modules 40 will be started at the same time to perform all-round image acquisition of the rail vehicle from multiple angles.

[0073] In addition, each image acquisition module 40 can be independently controlled, and its shooting angle, focal length and other parameters can be adjusted as needed through the posture adjustment module to obtain the clearest and most accurate image information. At the same time, these image data can also be transmitted to the background system in real time for analysis and processing, further improving the detection efficiency and accuracy.

[0074] The detection device also has a high degree of automation and intelligence. Each group of image acquisition modules 40 can automatically start and stop, automatically adjust shooting parameters, and automatically transmit image data. The background system can also automatically analyze and process the image data, generate detection reports and early warning information, greatly reducing the burden of manual operation and improving detection efficiency and accuracy.

[0075] In some possible embodiments of the present invention, the top bracket 10 includes: a mounting beam 101, which is used to be installed in the mounting position; a suspension beam 102, which can be movably installed on the mounting beam 101; wherein the image acquisition module 40 for capturing the top image of the rail vehicle is installed on the suspension beam 102.

[0076] Specifically, this embodiment provides an implementation of a top bracket 10, wherein the mounting beam 101 is the main part of the top bracket 10, and is used to be stably mounted at a specific mounting position. The mounting position may be a roof, a gantry, etc. The mounting beam 101 has sufficient strength and rigidity to ensure that it remains stable during the detection process and is not disturbed by the outside world.

[0077] The suspension beam 102 is a part that is movably mounted on the mounting beam 101. The suspension beam 102 has certain flexibility and adjustability, and its position and angle can be adjusted as needed. This design enables the suspension beam 102 to drive the two-dimensional image acquisition element 401 and the three-dimensional image acquisition element 402 and position them to the best shooting position to obtain the clearest and most comprehensive image information.

[0078] By combining the mounting beam 101 with the suspension beam 102, the entire support structure is more stable and reliable, ensuring stability during the detection process without being affected by vibration or displacement. The mobility of the suspension beam 102 allows the acquisition piece to flexibly adjust its position and angle to adapt to the size and shape of different rail vehicles, ensuring the accuracy and integrity of image acquisition. This design also improves the operating convenience of the detection device, allowing operators to easily install, adjust and replace the acquisition piece.

[0079] The two-dimensional image acquisition element 401 and the three-dimensional image acquisition element 402 are key components that are movably mounted on the suspension beam 102. The image acquisition module 40 can be used to achieve fine-tuning, translation, rotation or tilt, etc., to ensure that the two-dimensional image acquisition element 401 and the three-dimensional image acquisition element 402 can accurately align with the target area and capture high-quality image data.

[0080] In some possible embodiments of the present invention, the side bracket 20 includes: a base 201; a roller 202, installed on the base 201; a vertical beam 203, installed on the base 201, the vertical beam 203 includes a plurality of vertical segments, and the plurality of vertical segments are plugged into each other to adjust the height of the vertical beam 203; a support beam 204, hinged between the vertical beam 203 and the base 201; wherein, an image acquisition module 40 for acquiring a side image of the rail vehicle is installed on the vertical beam 203.

[0081] Specifically, this embodiment provides an implementation of a side bracket 20, which mainly includes a base 201, a vertical beam 203 and a support beam 204. This design makes the side bracket 20 not only stable and reliable, but also rich in functions and can meet various detection requirements. The roller 202 meets the movement requirements of the detection device.

[0082] The base 201 is the basic part of the side bracket 20. The base 201 bears the weight of the entire side bracket 20 and ensures the stability and safety of the side bracket 20. The roller 202 is installed at the lower part of the base 201, which enables the entire detection device to be easily moved and positioned to adapt to different detection environments and requirements.

[0083] The vertical beam 203 is the main structure of the side bracket 20. The vertical beam 203 is installed on the base 201 to provide a solid support for the image acquisition module 40. The height and position of the vertical beam 203 can be adjusted as needed to ensure that the image acquisition module 40, the speed radar 403 and the vehicle number recognition element 404 can be accurately aligned with the target area.

[0084] Furthermore, the design of the vertical segments allows the height of the vertical beam 203 to be adjusted according to actual needs. Each vertical segment has a standard size and interface to ensure that they can be tightly plugged into each other to form a stable support structure. By increasing or decreasing the number of vertical segments, the height of the vertical beam 203 can be flexibly adjusted to meet the requirements of different detection scenarios and vehicle heights.

[0085] This design not only improves the adaptability of the detection device, but also simplifies the adjustment process. The operator can easily add or remove vertical segments as needed to quickly adjust the height of the vertical beam 203. This flexibility enables the detection device to adapt to more types of rail vehicles, improving its practicality and work efficiency.

[0086] At the same time, the plug-in connection mode of the vertical segments also ensures the stability and safety of the vertical beam 203. The plug-in interface is carefully designed to ensure the tight connection between the vertical segments to prevent loosening or falling off during the detection process. In addition, the vertical segments are also made of high-strength materials, which can withstand large weight and impact force, ensuring the stable operation of the detection device.

[0087] The support beam 204 is hinged between the vertical beam 203 and the base 201 to form a movable connection structure. This design enables the support beam 204 to adjust the angle and position within a certain range, thereby further enhancing the flexibility and adaptability of the side bracket 20.

[0088] In some possible embodiments of the present invention, the image acquisition module 40 includes: a two-dimensional image acquisition element 401, a three-dimensional image acquisition element 402, a speed radar 403, a vehicle license plate recognition component 404 and a fill light component 405; the speed radar 403 and the vehicle license plate recognition component 404 are installed on the vertical beam 203; the fill light component 405 is installed on the side of the vertical beam 203 facing the bottom bracket 30.

[0089] Specifically, this embodiment provides an implementation of an image acquisition module 40, in which a two-dimensional image acquisition element 401 captures two-dimensional image information of a rail vehicle. This acquisition element can generate a high-definition plane image, showing in detail the appearance, details and possible problems of the side of the vehicle. It is very helpful for detecting the integrity of the vehicle body, the state of the windows, and the working conditions of components such as doors. The three-dimensional image acquisition element 402 can collect three-dimensional image data of the side of the rail vehicle, providing more comprehensive and in-depth detection information. The three-dimensional image not only shows the appearance of the object, but also reveals the shape, structure and depth information of the object. This enables the detection device to more accurately identify tiny depressions, protrusions or other irregular shapes on the side of the vehicle, further improving the accuracy and reliability of the detection.

[0090] In a possible embodiment, the two-dimensional image acquisition element 401 and the three-dimensional image acquisition element 402 are cameras for acquiring image information of the rail vehicle.

[0091] Furthermore, the speed radar 403 is mounted on the side bracket 20 so as to accurately measure the speed of the rail vehicle when it passes by. This design enables the detection device to obtain the speed information of the rail vehicle in real time, providing an important basis for subsequent data analysis and processing.

[0092] By measuring the speed of rail vehicles in real time, the detection device can more accurately assess the operating status of the vehicle and promptly detect abnormal conditions such as speeding.

[0093] Furthermore, the vehicle number recognition element 404 is a device that can automatically recognize the vehicle number. It usually uses image processing technology to obtain the vehicle identity information by photographing, recognizing and analyzing the vehicle number on the vehicle. In the present invention, the vehicle number recognition element 404 is also installed on the side bracket 20 to automatically capture and recognize the vehicle number of the rail vehicle during the detection process. This function not only improves the detection efficiency, but also reduces the possibility of manual input errors.

[0094] The function of automatic identification of vehicle numbers enables the detection device to automatically record the identity information of each detected vehicle, providing convenience for subsequent data management and tracing.

[0095] By integrating the speed radar 403 and the vehicle number recognition element 404 on the side bracket 20, the detection device can realize the synchronous detection and recognition of the speed and vehicle number of the rail vehicle. This design not only improves the automation of the detection device, but also makes the entire detection process more efficient, accurate and reliable.

[0096] In addition, the fill light 405 is installed on the side of the vertical beam 203 facing the bottom bracket 30, that is, near the two-dimensional image acquisition component 401, the three-dimensional image acquisition component 402, the speed radar 403 and the vehicle number recognition component 404. This design enables the fill light 405 to directly illuminate the side of the detected vehicle, effectively eliminating shadows and dark areas, and improving the clarity and contrast of image acquisition.

[0097] The specific type and specification of the fill light 405 can be selected according to actual needs. For example, an LED fill light, a halogen fill light or other suitable light sources can be used. At the same time, the brightness and irradiation angle of the fill light 405 can also be adjusted to meet the needs of different detection environments and vehicle types.

[0098] By adding the fill light 405, the detection device can maintain stable image acquisition quality in an environment with insufficient light or large light changes. Whether at night, on a rainy day or in other dark conditions, the fill light 405 can provide sufficient light support for image acquisition to ensure that the acquired image is clear and accurate.

[0099] In some possible embodiments of the present invention, the posture adjustment module 50 includes: a fixed plate 501; a first movable plate 502, which is rotatably arranged on the fixed plate 501 around a first axis, and the first axis is perpendicular to the first movable plate 502; a first driving mechanism, suitable for driving the first movable plate 502 to rotate relative to the fixed plate 501; a second movable plate 503, suitable for installing the image acquisition module 40, the second movable plate 503 is connected to the first movable plate 502, and the second movable plate 503 can at least rotate around the second axis and the third axis respectively relative to the first movable plate 502, the second axis is perpendicular to the third axis, and the second axis and the third axis are both perpendicular to the first axis; a second driving mechanism, suitable for driving the second movable plate 503 to rotate around the second axis relative to the first movable plate 502; a third driving mechanism, suitable for driving the second movable plate 503 to rotate around the third axis relative to the first movable plate 502.

[0100] Specifically, this embodiment provides an implementation of a posture adjustment module 50, wherein a first movable plate 502 is rotatably disposed on a fixed plate 501 around a first axis, and the first axis is perpendicular to the first movable plate 502. The first driving mechanism is used to drive the first movable plate 502 to rotate relative to the fixed plate 501.

[0101] The second movable plate 503 is connected to the first movable plate 502 . The image acquisition module 40 is installed on the second movable plate 503 . When the first movable plate 502 rotates relative to the fixed plate 501 , the first movable plate 502 and the image acquisition module 40 are driven to rotate around the first axis.

[0102] The second movable plate 503 can at least rotate about the second axis and the third axis relative to the first movable plate 502, the second axis is perpendicular to the third axis, and the second axis and the third axis are both perpendicular to the first axis. Figure 4 The straight line indicated by z in the figure, the second axis is as follows Figure 4 The straight line indicated by x in the figure, the third axis is as follows Figure 4 The straight line indicated by y in .

[0103] The second driving mechanism is used to drive the second movable plate 503 to rotate around the second axis relative to the first movable plate 502, and the third driving mechanism is used to drive the second movable plate 503 to rotate around the third axis relative to the first movable plate 502. When the second movable plate 503 rotates around the second axis and the third axis, the image acquisition module 40 rotates along with the second movable plate 503 around the second axis and the third axis.

[0104] With such arrangement, the first drive mechanism, the second drive mechanism and the third drive mechanism can drive the second movable plate 503 to rotate around the first axis, the second axis and the third axis respectively relative to the fixed plate 501, thereby adjusting the posture of the second movable plate 503, and further adjusting the posture of the image acquisition module 40 located on the second movable plate 503. No manual operation is required in the process of adjusting the posture of the image acquisition module 40, which reduces the dependence on manual operation. Moreover, under extreme working conditions, there is no need to manually touch the posture adjustment device itself, and the first drive mechanism, the second drive mechanism and the third drive mechanism can be remotely controlled, which removes the limitation of the working environment on the posture adjustment process, and solves the problem in the prior art that the posture adjustment of the camera is highly dependent on manual operation and cannot be completed manually under extreme working conditions.

[0105] In some possible embodiments of the present invention, the first driving mechanism includes: two groups of first temperature-dependent deformation rods 504, which are respectively connected to the power supply; the first ends of the first temperature-dependent deformation rods 504 are connected to the fixed plate 501, and the second ends of the first temperature-dependent deformation rods 504 are connected to the first movable plate 502; wherein the axes of the two groups of first temperature-dependent deformation rods 504 are parallel to each other and perpendicular to the first axis, and the two groups of first temperature-dependent deformation rods 504 are respectively located on both sides of the first axis.

[0106] Specifically, this embodiment provides an implementation of a first driving mechanism, the size of the fixed plate 501 is larger than the size of the first movable plate 502, the two groups of first temperature-dependent deformation rods 504 are respectively located on opposite sides of the first movable plate 502, and the first axis is located between the straight lines where the axes of the two groups of first temperature-dependent deformation rods 504 are located. Referring to the figure, each group of first temperature-dependent deformation rods 504 includes only one first temperature-dependent deformation rod 504.

[0107] Each first temperature-dependent deformation rod 504 is connected to a power source, and current is passed through the power source. Different currents passed through the first temperature-dependent deformation rod 504 will generate different amounts of heat and temperatures, and thus the first temperature-dependent deformation rod 504 will have different lengths.

[0108] By passing different current values ​​through the two groups of first temperature-dependent deformation rods 504 of the first driving mechanism, the two groups of first temperature-dependent deformation rods 504 can be made to have different lengths, thereby generating torque on the first movable plate 502 to drive the first movable plate 502 to rotate relative to the fixed plate 501 .

[0109] In some possible embodiments of the present invention, the second driving mechanism includes: two groups of second temperature-dependent deformation rods 505, which are respectively connected to the power supply; the first end of the second temperature-dependent deformation rod 505 is connected to the first movable plate 502, and the second end of the second temperature-dependent deformation rod 505 is connected to the second movable plate 503; wherein the axes of the two groups of second temperature-dependent deformation rods 505 are parallel to the first axis, and the two groups of second temperature-dependent deformation rods 505 are respectively located on both sides of the second axis.

[0110] Specifically, this embodiment provides an implementation of a second driving mechanism, wherein a gap is provided between the second movable plate 503 and the first movable plate 502, two groups of second temperature-dependent deformation rods 505 are disposed between the second movable plate 503 and the first movable plate 502, and both ends of the second temperature-dependent deformation rods 505 are respectively connected to the side of the second movable plate 503 close to the first movable plate 502. Referring to the figure, each group of second temperature-dependent deformation rods 505 includes only one second temperature-dependent deformation rod 505, and the two second temperature-dependent deformation rods 505 are distributed along the third axis, that is, the distribution direction of the two second temperature-dependent deformation rods 505 is perpendicular to the second axis.

[0111] Each second temperature-dependent deformation rod 505 is connected to a power source, and current is passed through the power source. Different currents passed through the second temperature-dependent deformation rod 505 will generate different amounts of heat and temperatures, and thus the second temperature-dependent deformation rod 505 will have different lengths.

[0112] By passing different current values ​​through the two groups of second temperature-dependent deformation rods 505 of the second driving mechanism, the two groups of second temperature-dependent deformation rods 505 can be made to have different lengths, thereby generating torque on the second movable plate 503 to drive the second movable plate 503 to rotate relative to the first movable plate 502 .

[0113] In some possible embodiments of the present invention, the third driving mechanism includes: two groups of third temperature-dependent deformation rods 506, which are respectively connected to the power supply; the first end of the third temperature-dependent deformation rod 506 is connected to the first movable plate 502, and the second end of the third temperature-dependent deformation rod 506 is connected to the second movable plate 503; wherein the axes of the two groups of third temperature-dependent deformation rods 506 are parallel to the first axis, and the two groups of third temperature-dependent deformation rods 506 are respectively located on both sides of the third axis.

[0114] Specifically, this embodiment provides an implementation of a third driving mechanism, wherein two groups of third temperature-dependent deformation rods 506 are disposed between the second movable plate 503 and the first movable plate 502, and both ends of the third temperature-dependent deformation rods 506 are respectively connected to the side of the second movable plate 503 close to the first movable plate 502. Referring to the figure, each group of third temperature-dependent deformation rods 506 includes only one third temperature-dependent deformation rod 506, and the two third temperature-dependent deformation rods 506 are distributed along the second axis, that is, the distribution direction of the two third temperature-dependent deformation rods 506 is perpendicular to the third axis.

[0115] Each third temperature-dependent deformation rod 506 is connected to a power source, and current is passed through the power source. Different currents passed through the third temperature-dependent deformation rod 506 will generate different amounts of heat and temperatures, and thus the third temperature-dependent deformation rod 506 will have different lengths.

[0116] By passing different current values ​​through the two groups of third temperature-dependent deformation rods 506 of the third driving mechanism, the two groups of third temperature-dependent deformation rods 506 can be made to have different lengths, thereby generating torque on the second movable plate 503 to drive the second movable plate 503 to rotate relative to the first movable plate 502.

[0117] It should be noted that each first temperature-dependent deformable rod 504 , each second temperature-dependent deformable rod 505 , and each third temperature-dependent deformable rod 506 can be individually configured with a power supply to facilitate control and eliminate the influence between the currents of the first temperature-dependent deformable rod 504 , the second temperature-dependent deformable rod 505 , and the third temperature-dependent deformable rod 506 .

[0118] In some possible implementations of the present invention, the posture adjustment module 50 further includes: a posture sensor 507 , which is disposed on the second movable plate 503 and is suitable for detecting the three-dimensional posture of the second movable plate 503 .

[0119] Specifically, this embodiment provides an implementation of a posture sensor 507, the posture sensor 507 and the power supply are electrically connected to the control system, the controller can obtain the three-dimensional posture information of the second movable plate 503 detected by the posture sensor 507, and control the output current of the power supply according to the three-dimensional posture information of the second movable plate 503.

[0120] In some possible embodiments of the present invention, the first temperature-dependent deformation rod 504, the second temperature-dependent deformation rod 505 and the third temperature-dependent deformation rod 506 all include: a temperature-dependent deformation rod body 508, the length of which changes with temperature; an insulating member 509, which is arranged at both ends of the temperature-dependent deformation rod body 508; and a temperature sensor 510, which is arranged inside the temperature-dependent deformation rod body 508, and the temperature sensor 510 is suitable for detecting the temperature of the temperature-dependent deformation rod body 508.

[0121] Specifically, this embodiment provides an implementation of a first temperature-dependent deformation rod 504, a second temperature-dependent deformation rod 505, and a third temperature-dependent deformation rod 506, wherein the length of the temperature-dependent deformation rod body 508 changes with the change of its own temperature, and an insulating member 509 is disposed at both ends of the temperature-dependent deformation rod body 508, and the ends of the temperature-dependent deformation rod body 508 are connected to the fixed plate 501, the first movable plate 502, or the second movable plate 503 through the insulating member 509, which can ensure the insulation between the temperature-dependent deformation rod body 508 and the fixed plate 501, the insulation between the temperature-dependent deformation rod body 508 and the first movable plate 502, and the insulation between the temperature-dependent deformation rod body 508 and the second movable plate 503. In specific applications, only current is passed through the temperature-dependent deformation rod body 508.

[0122] The temperature sensor 510 is disposed inside the temperature-dependent deformation rod body 508, and the temperature sensor 510 is used to detect the temperature of the temperature-dependent deformation rod body 508. The temperature sensor 510 and the power supply are both electrically connected to the control system, and the control system can obtain the temperature data detected by each temperature sensor 510 to control the output current of the power supply.

[0123] When the temperature-dependent deformation rod body 508 is connected to the first movable plate 502 and the second movable plate 503 through the insulating member 509, the insulating member 509 at at least one end of the temperature-dependent deformation rod body 508 can be connected to the temperature-dependent deformation rod body 508 through a universal hinge, or the insulating member 509 at at least one end of the temperature-dependent deformation rod body 508 can be connected to the fixed plate 501, the first movable plate 502 or the second movable plate 503 through a universal hinge. When adjusting the posture of the image acquisition module 40 and the posture of the second movable plate 503, the connection structure between the second temperature-dependent deformation rod 505 and the first movable plate 502, the connection structure between the second temperature-dependent deformation rod 505 and the second movable plate 503, the connection structure between the third temperature-dependent deformation rod 506 and the first movable plate 502, and the connection structure between the third temperature-dependent deformation rod 506 and the second movable plate 503 can adapt to the posture change of the second movable plate 503, avoiding the problem that the posture adjustment device is damaged due to stress.

[0124] A coil spring 513 may also be sleeved outside the temperature-dependent deformation rod body 508 , and two ends of the coil spring 513 are respectively connected to two ends of the temperature-dependent deformation rod body 508 to assist the temperature-dependent deformation rod body 508 in deformation.

[0125] The insulating member 509 may be, but is not limited to, ceramics which are thermally insulating materials.

[0126] The material of the temperature-dependent deformation rod body 508 can be a memory alloy, and the length of the memory alloy rod changes with the temperature. In the specific real-time process, different formulas can be selected to print the memory alloy according to the change rate of the length of the temperature-dependent deformation rod body 508 with temperature.

[0127] In an optional embodiment, a rod with adjustable length such as a cylinder or a hydraulic cylinder may be used to replace the first temperature-dependent deformation rod 504, the second temperature-dependent deformation rod 505 and the third temperature-dependent deformation rod 506. The structural principle of the telescopic adjustment of the rod itself is different, but the layout position of the rod is consistent with the layout position of the first temperature-dependent deformation rod 504, the second temperature-dependent deformation rod 505 and the third temperature-dependent deformation rod 506, and the connection relationship between the rod and other components is completely consistent with the connection relationship between the first temperature-dependent deformation rod 504, the second temperature-dependent deformation rod 505 and the third temperature-dependent deformation rod 506 and other connecting components, which will not be elaborated in detail.

[0128] In some possible embodiments of the present invention, the posture adjustment module 50 also includes: a connecting rod 512, the axis of the connecting rod 512 coincides with the first axis; a universal connector 511, arranged between the first movable plate 502 and the second movable plate 503; wherein the first end of the universal connector 511 is connected to the first movable plate 502 through the connecting rod 512, and the second end of the universal connector 511 is fixedly connected to the second movable plate 503.

[0129] Specifically, this embodiment provides an implementation of a connecting rod 512 and a universal connector 511 . The connecting rod 512 can support the second movable plate 503 to improve the stability of the second movable plate 503 .

[0130] When the first movable plate 502 is rotatably connected to the fixed plate 501, a thrust bearing 514 can be provided between the first movable plate 502 and the fixed plate 501. The thrust bearing 514 can limit the displacement of the first movable plate 502 relative to the fixed plate 501 along the first axis and can reduce friction during posture adjustment.

[0131] In an application scenario, such as Fig. 9 As shown, the second embodiment of the present invention provides a posture adjustment method, which is suitable for adjusting the posture adjustment device described above. The posture adjustment method described below and the posture adjustment device described above can be referred to each other.

[0132] When the posture adjustment device includes a posture sensor 507, and the posture adjustment device is disposed on the second movable plate 503, the posture adjustment method provided by the second aspect of the present invention includes the following steps: Step S110, determining the relationship between the coordinate system of the posture sensor 507 and the coordinate system of the image acquisition module 40; Step S120, obtaining the target posture of the image acquisition module 40; Step S130, determining the target posture of the posture sensor 507 corresponding to the target posture of the image acquisition module 40 according to the relationship between the coordinate system of the posture sensor 507 and the coordinate system of the image acquisition module 40; Step S140, determining a target current of each first temperature-dependent deformable rod 504, a target current of each second temperature-dependent deformable rod 505, and a target current of each third temperature-dependent deformable rod 506 according to a target posture of the posture sensor 507 and a database, wherein the database includes a mapping relationship between the posture of the posture sensor 507, the current of each first temperature-dependent deformable rod 504, the current of each second temperature-dependent deformable rod 505, and the current of each third temperature-dependent deformable rod 506; Step S150 , controlling the output current of the power supply according to the target current of each first temperature-dependent deformable rod 504 , the target current of each second temperature-dependent deformable rod 505 , and the target current of each third temperature-dependent deformable rod 506 .

[0133] The image acquisition module 40 and the posture sensor 507 are installed at different positions on the second movable plate 503, the reference points of the coordinate system of the image acquisition module 40 and the coordinate system of the posture sensor 507 are different, the posture of the second movable plate 503 in the coordinate system of the image acquisition module 40 and the posture of the second movable plate 503 in the coordinate system of the posture sensor 507 are different, that is, the three-dimensional posture of the second movable plate 503 detected by the posture sensor 507 is completely different from the posture of the second movable plate 503 in the coordinate system of the image acquisition module 40, and accordingly, the posture of the image acquisition module 40 is also inconsistent with the three-dimensional posture detected by the posture sensor 507.

[0134] In an embodiment of the present invention, the relationship between the coordinate system of the posture sensor 507 and the coordinate system of the image acquisition module 40 is first determined to unify the reference point, so as to facilitate the establishment of a corresponding relationship between the posture of the image acquisition module 40 and its posture in the coordinate system of the posture sensor 507, thereby determining the target detection posture, which is the target posture of the image acquisition module 40 in the coordinate system of the posture sensor 507.

[0135] Then, according to the mapping relationship between the target detection posture and the database, the target current of each first temperature-dependent deformable rod 504 , the target current of each second temperature-dependent deformable rod 505 , and the target current of each third temperature-dependent deformable rod 506 can be determined and controlled.

[0136] The derivation process of the beneficial effects of the posture adjustment method in the embodiment of the present invention is substantially similar to the derivation process of the beneficial effects of the posture adjustment device described above, and thus will not be described in detail herein.

[0137] In a specific embodiment, the steps of the posture adjustment method are as follows: a. Establishing the relationship between the coordinate system of the posture sensor 507 and the coordinate system of the image acquisition module 40, and calibrating the initial posture of the image acquisition module 40 to provide a reference for subsequent posture adjustment; b. The processor obtains the target posture of the image acquisition module 40 and calculates the difference between the target posture and the current posture: ∆α, ∆β, ∆γ; c. The processor calculates the temperature difference ∆T between each group of temperature-dependent deformation rod bodies 508 required to meet the target posture, and then calculates the current difference ∆I required to meet the above temperature difference; d. Control the current size of each group of temperature-dependent deformation rod body 508; e. Real-time detection of the temperature information of each group of temperature-dependent deformation stem bodies; f. Real-time detection of the current posture of the image acquisition module 40 to determine whether the target posture is reached. If the target posture is not reached, steps b to f are re-executed. If the target posture is reached, the posture adjustment is completed.

[0138] The mapping relationship in the above database is generally discrete data, which is difficult to achieve continuity. In the specific implementation process, it is inevitable that the target detection posture is not located in the above database.

[0139] In a further embodiment of the present invention, when the target detection posture is not in the above database, the applicable posture is first searched and determined in the database based on the target detection posture and the database, and the applicable posture is the posture in the database that is closest to the target detection posture.

[0140] Furthermore, according to the applicable posture and the database, the applicable current of each first temperature-dependent deformable rod 504, the applicable current of each second temperature-dependent deformable rod 505, and the applicable current of each third temperature-dependent deformable rod 506 corresponding to the applicable state can be determined, and then the output current of the power supply can be controlled according to the applicable current of each first temperature-dependent deformable rod 504, the applicable current of each second temperature-dependent deformable rod 505, and the applicable current of each third temperature-dependent deformable rod 506.

[0141] After the current of each first temperature-dependent deformation rod 504, the current of each second temperature-dependent deformation rod 505, and the current of each third temperature-dependent deformation rod 506 are adjusted to the right position, the current is fine-tuned. During the fine-tuning process, the current posture of the image acquisition module 40 in the coordinate system of the posture sensor 507 is obtained in real time until the current posture of the image acquisition module 40 in the coordinate system of the posture sensor 507 is consistent with the target detection posture.

[0142] At this time, it is also necessary to update the mapping relationship between the target detection posture and the current of each first temperature-dependent deformation rod 504, the current of each second temperature-dependent deformation rod 505 and the current of each third temperature-dependent deformation rod 506 to the database to achieve a self-learning effect to facilitate subsequent posture adjustment.

[0143] In some specific embodiments of the present invention, Figures 1 to 14 As shown, this solution provides a panoramic detection method for a rail vehicle, which is applied to the panoramic detection device for the rail vehicle mentioned above; Methods include: In response to a start signal of the panoramic detection of the rail vehicle, image features of the rail vehicle in the detection space are acquired, the image features at least including image information collected from the top, side and bottom of the rail vehicle; Based on image features, panoramic detection of rail vehicles is performed.

[0144] It should be noted that the camera line frequency is an important parameter of industrial cameras. In practical applications, most inspections require imaging of moving products. At this time, the camera line frequency must meet the specific use requirements to clearly and accurately image rail vehicles. Taking the application of cameras in the vehicle field as an example, in the daily maintenance of vehicles, line scan cameras are used to collect vehicle images. The image collection of line scan cameras is controlled by the camera line frequency parameters. In actual application scenarios, the vehicle speed is collected in real time by radar, and then the speed is converted into the camera line frequency according to the speed-line frequency conversion formula. However, when the vehicle speed changes, due to the delay in the conversion of radar-vehicle speed-line frequency information, the camera line frequency will be inconsistent with the actual speed of the vehicle, resulting in distortion of the collected vehicle image.

[0145] The current method to solve distortion is mainly to use the coordinate difference between the image to be detected and the standard image to correct the image to be detected. The distortion problem caused by the time delay from speed acquisition to image acquisition has not been solved from the source of image acquisition. There are problems such as large amount of calculation, slow calculation speed, and high hardware requirements. When the system is transferred from the current model to other models, the image correction algorithm needs to be redeveloped, and the translation cost is high and the cycle is long.

[0146] In some possible implementations of the present invention, the image features include at least a standard image of the rail vehicle and an image to be detected, the standard image is an image of the rail vehicle captured by the image acquisition module when the rail vehicle moves at a standard speed, and the image to be detected is an image of the rail vehicle captured by the image acquisition module when the rail vehicle moves at an actual speed; The steps of performing panoramic detection of rail vehicles based on image features specifically include: Determining a line frequency difference according to the line frequency of the standard image and the line frequency of the image to be detected; Establishing a line frequency difference time series prediction model data set based on the line frequency difference, the line frequency difference time series prediction model data set includes a plurality of data pairs, the data pairs include the standard speed and line frequency difference of the historical image; The line frequency difference timing prediction model is trained according to the line frequency difference timing prediction model data set to obtain a trained line frequency difference timing prediction model; Based on the trained line frequency difference timing prediction model, the line frequency of the image acquisition module is controlled by line frequency compensation.

[0147] Specifically, this embodiment provides an implementation method for panoramic detection of rail vehicles. First, a standard image is established in a test environment, and the vehicle obtains the image to be detected under normal operating conditions. Then, the line frequency of the standard image and the line frequency of the image to be detected are used to calculate the line frequency difference at each position of the image. After that, based on multiple acquisitions of historical image data, a line frequency difference timing prediction model data set is established. The line frequency difference timing prediction model is trained according to the line frequency difference timing prediction model data set to obtain a trained line frequency difference timing prediction model. The line frequency of the image acquisition module is compensated based on the trained line frequency difference timing prediction model to achieve precise control of the line frequency of the image acquisition module. The present invention solves the distortion problem caused by the delay of the speed signal from the source of image acquisition, reduces the workload of subsequent image processing, and thus improves the operating efficiency of the system. The calculated data is numerical, and the number of parameters is small, the calculation speed is fast, the example requirements can be reduced, and the system cost can be reduced. Based on the principle of line scanning image acquisition module acquisition, a line frequency compensation method is proposed, which has a low correlation with the specific application scenario, has universality, and can be promoted and applied in industrial scenarios.

[0148] It can be understood that the present invention determines the line frequency difference according to the line frequency of the standard image and the line frequency of the image to be detected; the line frequency of the standard image is the line frequency of the rail vehicle image collected by the image acquisition module when the rail vehicle moves at the standard speed, and the line frequency of the image to be detected is the line frequency of the rail vehicle image collected by the image acquisition module when the rail vehicle moves at the actual speed; a line frequency difference timing prediction model data set is established based on the line frequency difference; the line frequency difference timing prediction model data set includes several groups of data pairs; the data pair includes the standard speed and line frequency difference of the historical image; the line frequency difference timing prediction model is trained according to the line frequency difference timing prediction model data set to obtain a trained line frequency difference timing prediction model; based on the trained line frequency difference timing prediction model, the line frequency of the image acquisition module is controlled by line frequency compensation. The present invention can solve the image acquisition distortion problem from the source of image acquisition, thereby reducing the workload of subsequent image processing. The control method is extensible and can be promoted and applied in industrial scenarios.

[0149] In some possible implementations of the present invention, determining the line frequency difference according to the line frequency of the standard image and the line frequency of the image to be detected includes: Acquire a standard image and an image to be detected; Perform feature extraction and feature matching on the standard image and the image to be detected to obtain a spatial distribution map of feature pairs, where the feature pairs include feature points on the standard image and feature points on the matching image to be detected; Determine the line frequency of the standard image and the line frequency of the image to be detected according to the spatial distribution diagram of the feature pair; Convert the line frequency of the image to be detected into the actual line frequency at the standard speed; The line frequency difference is obtained by subtracting the line frequency of the standard image from the actual line frequency at the standard speed.

[0150] Specifically, this embodiment provides an implementation method for determining the line frequency difference according to the line frequency of the standard image and the line frequency of the image to be detected.

[0151] In some possible implementations of the present invention, feature extraction and feature matching are performed on the standard image and the image to be detected to obtain a spatial distribution map of feature pairs, including: A threshold segmentation algorithm is used to align the standard image with one end of the image to be detected; The SFIT feature extraction algorithm is used to extract feature points of the standard image and the image to be detected, respectively, to obtain feature points on the standard image and feature points on the image to be detected; The DMatche operator is used to pair the feature points on the standard image with the features on the image to be detected to obtain the spatial distribution map of the feature pairs.

[0152] Specifically, this embodiment provides an implementation method for performing feature extraction and feature matching on a standard image and an image to be detected to obtain a spatial distribution map of feature pairs.

[0153] In some possible implementations of the present invention, determining the line frequency of the standard image and the line frequency of the image to be detected according to the spatial distribution diagram of the feature pair includes: Determine the line frequency of the standard image according to the distance and time difference between any two feature points in the standard image; The line frequency of the image to be detected is determined according to the distance and time difference between any two feature points in the image to be detected.

[0154] Specifically, this embodiment provides an implementation method for determining the line frequency of the standard image and the line frequency of the image to be detected according to the spatial distribution diagram of the feature pairs.

[0155] It should be noted that if Fig.11 As shown, Fig.11 A schematic diagram of the principle of a line frequency control method for an image acquisition module provided by the present invention.

[0156] In this embodiment, by extracting and matching the features of the standard image and the image to be inspected, the information space distribution difference between the image to be inspected and the standard image is calculated, thereby calculating the line frequency difference. Specifically, when a vehicle passes through the panoramic detection device, the detection device uses radar to collect the real-time speed of the vehicle, the speed is m / s, and uses the speed information to control the image acquisition module to change the line frequency to complete the vehicle information acquisition. The conversion formula between the original speed and the line frequency of the image acquisition module is: Ft=Vt / Pa, Ft is the line frequency of the image acquisition module at time t, Vt is the speed of the vehicle at time t, and Pa is the horizontal pixel accuracy.

[0157] Select a vehicle to pass through the panoramic detection device at a standard speed Vs, set the corresponding image acquisition module line frequency Fs, and complete standard image acquisition.

[0158] The vehicle passes through the panoramic detection device under normal operating conditions, forming a collection of images to be detected and corresponding speed data.

[0159] The threshold segmentation method is used to align the leftmost end of the image to be detected with the standard image. The SFIT (Scale-invariant feature transform) feature extraction algorithm is used to extract feature points of the standard image and the image to be detected, mainly extracting corner and line feature points. 2000-3000 feature points can be extracted within the spatial range. The DMarke operator is used to pair the feature points of the standard image with the feature points of the image to be detected, and a spatial distribution map of the feature pairs can be obtained.

[0160] Feature extraction and feature matching will generate feature pairs on the image. By aligning the left ends of the images, the feature points of the standard image and the feature points of the image to be detected will form a time distribution under the same coordinate system. Ps1-n is the feature point of the standard image, ts1-n is the time of the corresponding feature point relative to the initial position of the standard image, Pd1-n is the feature point of the image to be detected, td1-n is the time of the corresponding feature point relative to the initial position of the image to be detected.

[0161] Calculate the distance between two feature points of the standard image, such as Ss1=Ps2-Ps1, and the time difference between the two feature points Δt1=ts2-ts1. 2000-3000 feature points can be extracted within the spatial range, and the density of feature points is large enough, so the standard line frequency of the standard image at this feature point can be obtained as Fs=Ss1 / Δt1. Similarly, the line frequency of each position of the standard image can be calculated as Fs(t)=Sst / Δtt.

[0162] Calculate the distance between two feature points of the image to be detected, such as Sd1=Pd2-Pd1, the time difference between the two feature points Δt1=td2-td1, because the image to be detected is 2000-3000 feature points can be extracted within the spatial range, and the density of feature points is large enough, so the standard line frequency of the image to be detected at this feature point can be obtained as Fd=Sd1 / Δt1. Similarly, the line frequency Fd(t)=Sdt / Δtt at each position of the image to be detected can be calculated.

[0163] Find the conversion coefficient between the line frequency of the image to be detected and the line frequency of the standard image, , where Vd(t) is the actual speed data of the vehicle.

[0164] Therefore, the actual line frequency of the image to be detected at time t when the speed is converted into the standard speed can be obtained as follows: .

[0165] Calculate the line frequency difference ΔFs(t)=Fs(t)-F's(t) of the image to be detected at the standard speed.

[0166] In some possible implementations of the present invention, establishing a line frequency difference timing prediction model data set based on the line frequency difference includes: Read the speed data of the historical image in segments to determine the actual speed data sequence; Normalize the actual speed data sequence to obtain a standard speed data sequence; The correlation between the standard speed data sequence and the line frequency difference is established by using the time distribution, and a line frequency difference timing prediction model data set is established. The line frequency difference timing prediction model data set includes a training set, a validation set and a test set.

[0167] Specifically, this embodiment provides an implementation method for establishing a line frequency difference timing prediction model data set based on the line frequency difference. By extracting data by segmenting the speed information and using time distribution to establish a correlation between the standard speed data sequence and the line frequency difference, a line frequency difference timing prediction model data set is established.

[0168] Specifically, based on the multiple acquisitions of historical image data, a line frequency difference time series prediction model data set is established. The data includes two parts of data. The first part is the actual speed data before time t. The actual speed data is read in three sections within 30 seconds before time t. In the first section, 30 actual speed data are evenly taken from 10 seconds before time t (3 data per second), the second section is 10 seconds before time t to 20 seconds before time t, and 10 actual speed data are evenly taken from 10 seconds before time t to 20 seconds before time t. The third section is 5 actual speed data are evenly taken from 20 seconds before time t to 30 seconds before time t, forming an actual speed data sequence. The actual speed value needs to be normalized to the standard speed, that is, the speed sequence speed .

[0169] The second part is the line frequency difference data ΔFs(t) at time t. The above two parts of data form a data pair {V}-ΔFs(t), and a total of 6000 data pairs are established to form a line frequency difference time series prediction model data set, of which 70% is a training set, 15% is a validation set, and 15% is a test set. The present invention normalizes all actual data (speed data, line frequency data) to standard data, thereby reducing the difficulty of model training and data requirements.

[0170] In some possible implementations of the present invention, a line frequency difference timing prediction model is trained according to a line frequency difference timing prediction model data set to obtain a trained line frequency difference timing prediction model, including: The LSTM model is trained according to the standard speed data sequence and line frequency difference, and the cross entropy loss function is selected to obtain the trained line frequency difference time series prediction model.

[0171] Specifically, this embodiment provides an implementation method for training a line frequency difference timing prediction model according to a line frequency difference timing prediction model data set to obtain a trained line frequency difference timing prediction model, using LSTM (Long ShortTerm Memory) for supervised learning, speed sequence data as LSTM training data, and the corresponding line frequency difference data ΔFs(t) as training labeled data, and selecting a cross entropy loss function. After training, a line frequency difference timing prediction model ΔFs(t)=L({V}) will be formed.

[0172] In some possible implementations of the present invention, based on the trained line frequency difference timing prediction model, the line frequency of the image acquisition module is controlled by line frequency compensation, including: Based on the trained line frequency difference timing prediction model, the line frequency compensation formula is used to control the line frequency of the image acquisition module. The line frequency compensation formula is: F't=Vt / Pa+ΔFs(t) / k Among them, F't is the line frequency controlled by the image acquisition module, Vt is the speed of the rail vehicle at time t, Pa is the lateral pixel accuracy, ΔFs(t) is the trained line frequency difference time series prediction model, and k is the conversion coefficient between the line frequency of the image to be detected and the line frequency of the standard image.

[0173] Specifically, this embodiment provides an implementation method for controlling the line frequency of an image acquisition module by means of line frequency compensation based on a trained line frequency difference timing prediction model. When the detection device is running, when calculating the line frequency at time t, the speed data 30 seconds before time t is first read and the standard speed data {Vt} is screened out according to the training set establishment method. The trained line frequency difference timing prediction model ΔFs(t)=L({Vt}) is used, and this value is used to perform real-time compensation for the standard line frequency at time t (Ft=Vt / Pa), and the line frequency F't=Vt / Pa+ΔFs(t) / k controlled by the image acquisition module is obtained.

[0174] Of course, the method of the present invention is not limited to the vehicle field, but can also be promoted and applied in other industrial scenarios.

[0175] In addition, it should be specially pointed out that the panoramic detection device can realize the intelligent, accurate and rapid identification and fault alarm of the fault items of the appearance status during the operation and maintenance process, and the high-quality image acquisition under different application conditions is an important basis for improving the application ability of the panoramic detection device. In order to improve the environmental adaptability of the image acquisition device of the panoramic detection device and ensure the quality of image acquisition, a laser light source is required as an active light source to fill the image acquisition device. The light source parameters of traditional laser light sources are fixed or can only achieve parameter adjustment within a small range. Its response timeliness and accuracy are difficult to meet the fill light requirements of the image acquisition device of the panoramic detection device. It can only achieve high-quality acquisition of images of certain specific vehicle conditions under certain specific environments. This greatly restricts the application and popularization of panoramic detection devices.

[0176] The present invention also provides a method for adjusting the light source parameters of a laser light source, which is used to solve the defect in the prior art that the fill light of the laser light source is difficult to meet the fill light requirements of the image acquisition device under different working conditions, and achieves the technical effect of adjusting the light source parameters of the laser light source to meet the fill light requirements of the image acquisition device under different working conditions.

[0177] In an exemplary embodiment, the present invention provides a method for adjusting light source parameters of a laser light source, including steps 210 to 240, which are described in detail as follows.

[0178] Step S210, obtaining current vehicle condition information of the rail vehicle and current ambient light information of the environment in which the rail vehicle is located.

[0179] In the embodiment of the present invention, a photosensitive sensor or other device is arranged at a designated position of a rail vehicle to pre-perceive the current vehicle condition information of the rail vehicle and the current ambient light information of the environment in which the rail vehicle is located.

[0180] Step S220 , determining a first light source parameter interval based on current ambient light information, and determining a second light source parameter interval based on current vehicle condition information.

[0181] In an embodiment of the present invention, a first light source parameter interval is determined based on current ambient light information, and the determined first light source parameter interval is a light source parameter set that can improve the image imaging effect after fill light is performed based on the light source parameters within the first light source parameter interval. A second light source parameter interval is determined based on current vehicle condition information, and similarly, the determined second light source parameter interval is a light source parameter set that can improve the image imaging effect after fill light is performed based on the light source parameters within the second light source parameter interval.

[0182] Step S230: determining a target light source parameter interval based on the first light source parameter interval and the second light source parameter interval.

[0183] In the embodiment of the present invention, a target light source parameter interval is determined based on the determined first light source parameter interval and the second light source parameter interval.

[0184] In an exemplary embodiment of the present invention, determining a target light source parameter interval based on a first light source parameter interval and a second light source parameter interval includes the following steps, which are described in detail as follows.

[0185] A union of the first light source parameter interval and the second light source parameter interval is determined as a target light source parameter interval.

[0186] In the embodiment of the present invention, the union of the first light source parameter interval and the second light source parameter interval is directly used as the target light source parameter interval. In other exemplary embodiments of the present invention, the intersection of the first light source parameter interval and the second light source parameter interval can also be used as the target light source parameter interval.

[0187] Step S240 , determining target light source parameters in a target light source parameter interval through a preset light source parameter control model, and adjusting light source parameters of the laser light source based on the target light source parameters.

[0188] In the embodiment of the present invention, the target light source parameters are determined in the target light source parameter interval through a preset light source parameter control model, and the light source parameters of the laser light source are adjusted based on the target light source parameters. The light source parameter control model is aimed at optimizing the image acquisition quality. Therefore, fill light based on the target light source parameters determined by the light source parameter control model can make the image acquired by the image acquisition device have a higher quality, thereby meeting the fill light requirements of the image acquisition device under different working conditions.

[0189] In an exemplary embodiment of the invention, determining the first light source parameter interval based on current ambient light information includes the following steps, which are described in detail as follows.

[0190] Extract current ambient light features from current ambient light information.

[0191] Based on the current ambient light characteristics, a first light source parameter interval is determined in a first light source parameter matching table; wherein the first light source parameter matching table is obtained by a preset first association relationship model, and the first association relationship model is a model constructed using a neural network, which is a model of the association relationship between ambient light characteristics and light source parameters.

[0192] In the embodiment of the present invention, the current ambient light feature is extracted from the current ambient light information. Fig.12 As shown, the light intensity in the ambient light information is measured by a light sensor to extract the brightness information, the color of the light source is measured by a color temperature meter to extract the light color temperature in the ambient light information, and the wavelength distribution of the light is measured by a spectrometer to extract the spectral characteristics of the ambient light, that is, the wavelength distribution. The extracted light intensity, light color temperature and wavelength distribution are used as the current ambient light characteristics.

[0193] A first light source parameter matching table is pre-established, which records the correspondence between ambient light features and light source parameters. The first light source parameter range can be obtained by matching the extracted current ambient light features in the first light source parameter matching table.

[0194] When establishing the first light source parameter matching table, multiple groups of experimental data are collected, each group of experimental data includes corresponding light source parameters and ambient light information, and the ambient light features in the experimental data are extracted through the aforementioned various devices. According to the diffraction theory, absorption theory and interference theory of light, as well as the signal processing principle, the influence of ambient light on the light source parameters of the laser light source is theoretically analyzed, and ray tracing software is used for numerical simulation, and the theoretical model is verified and optimized. The deviation between the actual data and the theoretical model is analyzed through the theoretical model, so that the theoretical model is adjusted to better match the actual situation. According to the extracted light feature information and the light source parameters recorded in the experiment, the correlation between the ambient light features and the light source parameters is analyzed, and the first correlation relationship model between the ambient light features and the light source parameters is constructed by using a neural network, and the performance of the first correlation relationship model is evaluated by cross-validation, and then the first light source parameter matching table in which the ambient light features and the light source parameters match each other is obtained based on the first correlation relationship model, and the first light source parameter matching table is stored in the ambient light-light source parameter mapping database.

[0195] In an exemplary embodiment of the present invention, determining the second light source parameter interval based on the current vehicle condition information includes the following steps, which are described in detail as follows.

[0196] Extract current vehicle condition features from current vehicle condition information.

[0197] Based on the current vehicle condition characteristics, a second light source parameter interval is determined in a second light source parameter matching table; wherein the second light source parameter matching table is obtained by a preset second association relationship model, and the second association relationship model is a model constructed using a neural network, which is a model of the association relationship between vehicle condition characteristics and light source parameters.

[0198] In the embodiment of the present invention, the current vehicle condition features are extracted from the current vehicle condition information. The vehicle condition features include multiple types, specifically, using a laser sensor to detect the reflective properties of the rail vehicle surface, including parameters such as surface smoothness and reflectivity; using a camera to identify the shape and structure of the rail vehicle to obtain the appearance information of the rail vehicle; using a color sensor to detect the color information of the rail vehicle; and also considering the motion state of the rail vehicle, including speed, direction and other information. Fig.13 As shown, reflectivity, appearance information and color information are used as vehicle condition features.

[0199] Computer vision technology, optical measurement methods and sensor technology are used, combined with image processing and data analysis, to accurately extract the characteristic parameters of the vehicle condition. When establishing the second light source parameter matching table, multiple groups of experimental data are collected, each of which includes corresponding light source parameters and vehicle condition information, and the vehicle condition characteristics in the experimental data are extracted through the aforementioned various devices. According to the diffraction theory, absorption theory and interference theory of light, as well as the signal processing principle, the influence of vehicle conditions on the light source parameters of the laser light source is theoretically analyzed, and ray tracing software is used for numerical simulation, and the theoretical model is verified and optimized. The deviation between the actual data and the theoretical model is analyzed, so that the theoretical model is adjusted to better match the actual situation. According to the extracted vehicle condition characteristics and the light source parameters recorded in the experiment, the matching relationship between the vehicle condition characteristics and the light source parameters is analyzed, and the second association relationship model between the vehicle condition characteristics and the light source parameters is constructed by using a neural network, and the performance of the second association relationship model is evaluated by cross-validation, and then the second light source parameter matching table in which the vehicle condition characteristics and the light source parameters match each other is obtained based on the second association relationship model, and the second light source parameter matching table is stored in the vehicle condition-light source parameter mapping database.

[0200] In an exemplary embodiment of the present invention, a target light source parameter is determined in a target light source parameter interval by using a preset light source parameter control model, including the following steps, which are described in detail as follows.

[0201] Initialization light source parameters are determined in the target light source parameter interval, and light source parameters of the laser light source are initialized based on the initialization light source parameters.

[0202] Based on the initialized light source parameters, the target light source parameters are obtained by dynamically optimizing the target light source parameter range through the light source parameter control model.

[0203] In an embodiment of the present invention, a light source parameter is determined in the target light source parameter interval as the initialization light source parameter. Specifically, an intermediate value in the target light source parameter interval can be selected as the initialization light source parameter. After the initialization light source parameter is determined, the light source parameter of the laser light source is initialized based on the initialization light source parameter. After the light source parameter is initialized, a dynamic optimization is performed in the target light source parameter interval through a light source parameter control model, so that the light source parameter of the laser light source is finely and dynamically adjusted within the target light source parameter interval to ensure image acquisition quality. The target light source parameter that best represents the image acquisition quality is obtained in the target light source parameter interval through the light source parameter control model.

[0204] In the embodiment of the present invention, the ambient light information and vehicle condition information under different working conditions are selected as factors affecting the image acquisition quality, and the optimal image acquisition quality is taken as the goal, and a light source parameter feedforward macro-control strategy is proposed, that is, the corresponding light source parameter interval is determined based on the ambient light information and the vehicle condition information, and then the target light source parameter interval for initialization of the light source parameter is determined; further, in the actual image acquisition process, the light source parameters are dynamically fine-tuned within the target light source parameter interval based on the light source parameter control model, and the light source parameters are selected and dynamically fine-tuned, which ensures the image acquisition quality and improves the adaptability of the panoramic detection device to complex application conditions. At the same time, this combination of macro and micro adjustments realizes the accuracy of light source parameter adjustment and the rapidity of response. In addition, the technical solution proposed by the present invention can be extended to other similar parameter control industrial scenarios, and has a relatively broad application prospect.

[0205] In an exemplary embodiment of the present invention, the process of constructing a light source parameter control model includes the following steps, which are described in detail as follows.

[0206] Acquire images collected based on laser light sources, as well as multi-dimensional evaluation parameters and laser light source parameters corresponding to the images; wherein the multi-dimensional evaluation parameters include laser light source station layout information, camera line frequency information, ambient light information and vehicle condition information.

[0207] The multidimensional evaluation parameters are input into the multidimensional fitness function to calculate the image acquisition quality of the image.

[0208] The relationship between laser light source parameters and image acquisition quality is iteratively optimized through genetic algorithm to determine the optimal solution parameter combination.

[0209] Construct a light source parameter control model based on parameter combination.

[0210] In the embodiment of the present invention, a light source parameter control model is constructed with the goal of optimizing image acquisition quality. Fig.14As shown in the figure, the image acquisition quality is quantitatively characterized by comprehensively considering the parameters such as the layout information of the laser light source station, the camera line frequency information, and the constraints such as the ambient light information and the vehicle condition information. The multidimensional fitness function is defined to comprehensively consider these goals and quantify the quality of image acquisition. The genetic algorithm is used to iteratively optimize the relationship between the light source parameters and the image acquisition quality to find the parameter combination with the best image acquisition quality. For each parameter combination, the defined multidimensional fitness function is used to evaluate, and the fitness value of each parameter combination is calculated. According to the fitness value, the image acquisition quality is determined to be the best in the current environment. According to the multidimensional fitness function, the parameter combination with a higher fitness value is selected from the population obtained by the genetic algorithm as the basis of the next generation population. At the same time, some parameter combinations are randomly selected for crossover operations to generate new parameter combinations, which increases the diversity of the population. The parameter combinations in the population are mutated to introduce new changes and promote the diversity of the search process. The optimal parameter combination is obtained through multiple iterations. And the optimal parameter combination in each generation of population is recorded. According to the parameter combination of the optimal solution obtained based on the genetic algorithm, a light source parameter control model is constructed, and then on the basis of light source parameter initialization, parameter selection and dynamic fine-tuning are performed within the target light source parameter range.

[0211] In an exemplary embodiment of the present invention, after determining the target light source parameters in the target light source parameter range through a preset light source parameter control model, the method further includes the following steps, which are described in detail as follows.

[0212] The target light source parameters are stored in a first light source parameter matching table corresponding to the ambient light information and / or a second light source parameter matching table corresponding to the vehicle condition information.

[0213] In an embodiment of the present invention, target light source parameters involved in the laser light source fill light process are stored in an ambient light-light source parameter mapping database and a vehicle condition-light source parameter mapping database, thereby continuously improving the laser light source parameter feedforward macro adjustment effect.

[0214] In the embodiment of the present invention, based on the requirements of the panoramic detection device and the detection algorithm for image acquisition quality, the influence of ambient light and vehicle conditions on image acquisition quality under different working conditions is compared and analyzed, and basic evaluation indicators of imaging effects such as uniform background grayscale and high target contrast are integrated to construct a comprehensive characterization model of image acquisition quality for panoramic detection devices, forming an image acquisition quality evaluation system. The evaluation strategies and evaluation criteria involved in the image acquisition quality evaluation process are quantitatively characterized through various indicators and models, and finally an image acquisition quality evaluation database is developed to provide support for the scientific evaluation of images collected by panoramic detection devices.

[0215] The development of the above-mentioned ambient light-light source parameter mapping database and vehicle condition-light source parameter mapping database needs to be based on the image acquisition quality evaluation database. At the same time, the image acquisition quality evaluation database will provide support for the image quality evaluation of the laser light source fill lighting process of the panoramic detection device image acquisition.

[0216] In some specific embodiments of the present invention, the present solution provides a rail vehicle, including the above-mentioned panoramic detection device for rail vehicles; or, when performing panoramic detection of rail vehicles, the above-mentioned panoramic detection method for rail vehicles is adopted.

[0217] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, i.e., they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Those of ordinary skill in the art may understand and implement it without creative effort.

[0218] Through the description of the above implementation modes, those skilled in the art can clearly understand that each implementation mode can be implemented by means of software plus a necessary general hardware platform, or of course by hardware. Based on such an understanding, the above technical solution can essentially or in other words be embodied in the form of a software product that contributes to the prior art. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods of each embodiment or some parts of the embodiment.

[0219] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A panoramic detection device for a rail vehicle, characterized in that: include: A top bracket (10), a bottom bracket (30), and two oppositely disposed side brackets (20) that surround and form a detection space; A plurality of image acquisition modules (40), respectively arranged on the top bracket (10), the bottom bracket (30) and the two side brackets (20), and used for acquiring panoramic images of rail vehicles passing through the detection space; The posture adjustment module (50) is provided in a one-to-one correspondence with the image acquisition module (40) and is used to adjust the acquisition angle of the image acquisition module (40).

2. The panoramic detection device for a rail vehicle according to claim 1, characterized in that: The top bracket (10) comprises: A mounting beam (101), the mounting beam (101) being used for being mounted at a mounting position; A suspension beam (102) movably mounted on the mounting beam (101); Wherein, the image acquisition module (40) for acquiring the top image of the rail vehicle is installed on the suspension beam (102).

3. The panoramic detection device for a rail vehicle according to claim 1, characterized in that: The side bracket (20) comprises: Base (201); A roller (202) mounted on the base (201); A vertical beam (203) is mounted on the base (201), the vertical beam (203) comprising a plurality of vertical segments, the plurality of vertical segments being plugged into each other to adjust the height of the vertical beam (203); A support beam (204) hinged between the vertical beam (203) and the base (201); Wherein, the image acquisition module (40) for acquiring a side image of the rail vehicle is installed on the vertical beam (203).

4. The panoramic detection device for a rail vehicle according to claim 3, characterized in that: The image acquisition module (40) comprises: a two-dimensional image acquisition element (401), a three-dimensional image acquisition element (402), a speed measurement radar (403), a vehicle number recognition element (404) and a fill light element (405); The two-dimensional image acquisition element (401) is used to acquire a two-dimensional image of the rail vehicle; the three-dimensional image acquisition element (402) is used to acquire a three-dimensional image of the rail vehicle; The speed measuring radar (403) and the vehicle number recognition component (404) are installed on the vertical beam (203); The fill light component (405) is installed on a side of the vertical beam (203) facing the bottom bracket (30).

5. The panoramic detection device for a rail vehicle according to any one of claims 1 to 3, characterized in that: The posture adjustment module (50) comprises: Fixed plate (501); A first movable plate (502) is rotatably arranged on the fixed plate (501) around a first axis, wherein the first axis is perpendicular to the first movable plate (502); a first driving mechanism, adapted to drive the first movable plate (502) to rotate relative to the fixed plate (501); a second movable plate (503) adapted to mount the image acquisition module (40); the second movable plate (503) being connected to the first movable plate (502); the second movable plate (503) being capable of rotating relative to the first movable plate (502) at least about a second axis and a third axis, respectively; the second axis is perpendicular to the third axis, and the second axis and the third axis are both perpendicular to the first axis; a second driving mechanism, adapted to drive the second movable plate (503) to rotate around the second axis relative to the first movable plate (502); The third driving mechanism is suitable for driving the second movable plate (503) to rotate around the third axis relative to the first movable plate (502).

6. The panoramic detection device for a rail vehicle according to claim 5, characterized in that: The first driving mechanism comprises: Two groups of first temperature-dependent deformation rods (504) are respectively connected to a power source; The first end of the first temperature-dependent deformation rod (504) is connected to the fixed plate (501), and the second end of the first temperature-dependent deformation rod (504) is connected to the first movable plate (502); The axes of the two groups of the first temperature-dependent deformation rods (504) are parallel to each other and perpendicular to the first axis, and the two groups of the first temperature-dependent deformation rods (504) are respectively located on both sides of the first axis.

7. The panoramic detection device for a rail vehicle according to claim 6, characterized in that: The second driving mechanism comprises: Two groups of second temperature-dependent deformation rods (505) are respectively connected to a power source; The first end of the second temperature-dependent deformation rod (505) is connected to the first movable plate (502), and the second end of the second temperature-dependent deformation rod (505) is connected to the second movable plate (503); The axes of the two groups of the second temperature-dependent deformation rods (505) are both parallel to the first axis, and the two groups of the second temperature-dependent deformation rods (505) are respectively located on both sides of the second axis.

8. The panoramic detection device for a rail vehicle according to claim 7, characterized in that: The third driving mechanism comprises: Two groups of third temperature-dependent deformation rods (506) are respectively connected to a power source; The first end of the third temperature-dependent deformation rod (506) is connected to the first movable plate (502), and the second end of the third temperature-dependent deformation rod (506) is connected to the second movable plate (503); The axes of the two groups of third temperature-dependent deformation rods (506) are both parallel to the first axis, and the two groups of third temperature-dependent deformation rods (506) are respectively located on both sides of the third axis.

9. The panoramic detection device for a rail vehicle according to claim 8, characterized in that: The posture adjustment module (50) further includes: The posture sensor (507) is arranged on the second movable plate (503) and is suitable for detecting the three-dimensional posture of the second movable plate (503).

10. The panoramic detection device for a rail vehicle according to claim 8, characterized in that: The first temperature-dependent deformation rod (504), the second temperature-dependent deformation rod (505) and the third temperature-dependent deformation rod (506) all comprise: A temperature-dependent deformation rod body (508), wherein the length of the temperature-dependent deformation rod body (508) changes with temperature; Insulating members (509) are arranged at both ends of the temperature-dependent deformation rod body (508); A temperature sensor (510) is arranged inside the temperature-dependent deformation rod body (508), and the temperature sensor (510) is suitable for detecting the temperature of the temperature-dependent deformation rod body (508).

11. The panoramic detection device for a rail vehicle according to claim 5, characterized in that: The posture adjustment module (50) further includes: A connecting rod (512), wherein the axis of the connecting rod (512) coincides with the first axis; A universal connecting member (511) is arranged between the first movable plate (502) and the second movable plate (503); The first end of the universal connector (511) is connected to the first movable plate (502) via a connecting rod (512), and the second end of the universal connector (511) is fixedly connected to the second movable plate (503).

12. A panoramic detection method for a rail vehicle, characterized in that: A panoramic detection device for a rail vehicle as claimed in any one of claims 1 to 11 above; The method comprises: In response to a start signal of the panoramic detection of the rail vehicle, image features of the rail vehicle in the detection space are acquired, the image features at least comprising image information collected from the top, side and bottom of the rail vehicle; Based on the image features, a panoramic detection is performed on the rail vehicle.

13. The panoramic detection method of a rail vehicle according to claim 12, characterized in that: The image features at least include a standard image of the rail vehicle and an image to be detected, wherein the standard image is an image of the rail vehicle captured by the image acquisition module when the rail vehicle moves at a standard speed, and the image to be detected is an image of the rail vehicle captured by the image acquisition module when the rail vehicle moves at an actual speed; The step of performing panoramic detection on the rail vehicle based on the image features specifically includes: Determining a line frequency difference according to the line frequency of the standard image and the line frequency of the image to be detected; Establishing a line frequency difference timing prediction model data set based on the line frequency difference, wherein the line frequency difference timing prediction model data set includes a plurality of data pairs, wherein the data pairs include a standard speed of a historical image and the line frequency difference; The line frequency difference timing prediction model is trained according to the line frequency difference timing prediction model data set to obtain a trained line frequency difference timing prediction model; Based on the trained line frequency difference timing prediction model, the line frequency of the image acquisition module is controlled by adopting a line frequency compensation method.

14. The method for panoramic detection of a rail vehicle according to claim 13, characterized in that: The step of determining the line frequency difference according to the line frequency of the standard image and the line frequency of the image to be detected comprises: Obtain a standard image and an image to be detected; Performing feature extraction and feature matching on the standard image and the image to be detected to obtain a spatial distribution map of feature pairs, wherein the feature pairs include feature points on the standard image and matching feature points on the image to be detected; Determining the line frequency of the standard image and the line frequency of the image to be detected according to the spatial distribution diagram of the feature pairs; Converting the line frequency of the image to be detected into the actual line frequency at the standard speed; The line frequency difference is obtained by subtracting the line frequency of the standard image from the actual line frequency at the standard speed.

15. The panoramic detection method of a rail vehicle according to claim 14, characterized in that: The step of extracting and matching features of the standard image and the image to be detected to obtain a spatial distribution map of feature pairs includes: Using a threshold segmentation algorithm to align the standard image with one end of the image to be detected; Using the SFIT feature extraction algorithm to extract feature points from the standard image and the image to be detected, respectively, to obtain feature points on the standard image and feature points on the image to be detected; The DMatch operator is used to pair the feature points on the standard image with the features on the image to be detected to obtain a spatial distribution map of the feature pairs.

16. The method for panoramic detection of a rail vehicle according to claim 14, characterized in that: The determining the line frequency of the standard image and the line frequency of the image to be detected according to the spatial distribution diagram of the feature pair comprises: Determining the line frequency of the standard image according to the distance and time difference between any two feature points in the standard image; The line frequency of the image to be detected is determined according to the distance and time difference between any two feature points in the image to be detected.

17. The panoramic detection method of a rail vehicle according to any one of claims 14 to 16, characterized in that: The step of establishing a line frequency difference timing prediction model data set based on the line frequency difference includes: Read the speed data of the historical image in segments to determine an actual speed data sequence; Normalizing the actual speed data sequence to obtain a standard speed data sequence; The correlation relationship between the standard speed data sequence and the line frequency difference is established by using time distribution, and the line frequency difference timing prediction model data set is established. The line frequency difference timing prediction model data set includes a training set, a validation set and a test set.

18. The method for panoramic detection of a rail vehicle according to claim 17, characterized in that: The step of training the line frequency difference timing prediction model according to the line frequency difference timing prediction model data set to obtain a trained line frequency difference timing prediction model includes: The LSTM model is trained according to the standard speed data sequence and the line frequency difference, and a cross entropy loss function is selected to obtain a trained line frequency difference time series prediction model.

19. The panoramic detection method of a rail vehicle according to any one of claims 14 to 16, characterized in that: The line frequency control of the image acquisition module is performed by line frequency compensation based on the trained line frequency difference timing prediction model, including: Based on the trained line frequency difference timing prediction model, the line frequency compensation formula is used to control the line frequency of the image acquisition module. The line frequency compensation formula is: F't=Vt / Pa+ΔFs(t) / k Among them, F't is the line frequency controlled by the image acquisition module, Vt is the speed of the rail vehicle at time t, Pa is the lateral pixel accuracy, ΔFs(t) is the trained line frequency difference timing prediction model, and k is the conversion coefficient between the line frequency of the image to be detected and the line frequency of the standard image.

20. A rail vehicle, characterized in that: A panoramic detection device for a rail vehicle comprising any one of claims 1 to 11; Alternatively, when performing panoramic detection of a rail vehicle, the panoramic detection method of a rail vehicle described in any one of claims 12 to 19 is adopted.

Images