Rail profile loss detection method, device and system

Abstract

The present invention provides a method, device and system for detecting the loss of rail profile, which are applied to a carrier platform. The method comprises: emitting structural lasers symmetrically relative to both sides of the rail and collecting return data; splicing and vibration compensation correction of the return data to obtain the actual profile of the rail; matching the actual profile with a pre-acquired standard profile by iterative closest point technology to obtain a matching result; and determining the loss value of the rail profile according to the matching result. The present invention calculates the loss value of the rail profile more accurately by three technical solutions: emitting symmetrical structural lasers, performing vibration compensation correction on the return lasers, and matching the profile by iterative closest point technology.

Description

Technical Field

[0001] The present invention relates to the technical field of loss testing, and in particular to a method, device and system for detecting loss of rail profile. Background Art

[0002] Rails are the main components of railway tracks. Their function is to guide the wheels of locomotives and vehicles forward, bear the huge pressure of the wheels, and transmit it to the sleepers. Rails must provide a continuous, smooth and least resistant rolling surface for the wheels. In electrified railways or automatic block sections, rails can also serve as track circuits.

[0003] With the rapid development of rail transit technologies such as railways and subways in my country, the increase in axle weight of locomotives and vehicles, and the increase in vehicle speed, higher requirements have been placed on the quality of railway lines, and the treatment of rail surface diseases is imminent. The service life of rails commonly used in railways and subways is controlled by fatigue cracks and wear caused by rolling contact. Fatigue cracks affect rail flaw detection, wheel-rail running status, increase maintenance costs, shorten the service life of rails, and even cause rail breakage; wear, especially side wear on curves, causes rails to go down the track prematurely. In order to delay the development of rolling contact fatigue damage and wear of rails and extend the service life of rails, modern rail maintenance technology must be adopted. In order to achieve the ideal maintenance effect, the rails must first be tested. Only when the current loss of the rails is understood can accurate and effective maintenance be carried out according to different situations. Summary of the invention

[0004] Based on this, an object of the present invention is to provide a method, device and system for detecting the loss of rail profile, so as to accurately detect the loss of rail profile.

[0005] In a first aspect, the present invention provides a method for detecting loss of a rail profile, which is applied to a carrier platform, and comprises: emitting a structural laser symmetrically relative to both sides of the rail, and collecting return data; splicing and vibration compensation correction of the return data to obtain the actual profile of the rail; matching the actual profile with a pre-acquired standard profile through an iterative closest point technique to obtain a matching result; and determining the loss value of the rail profile based on the matching result.

[0006] In an optional embodiment, the carrier platform is provided with a laser sensor; the steps of emitting structural lasers symmetrically relative to both sides of the rails and collecting return data include: according to a preset first time interval or a preset first spatial interval, the laser sensor emits structural lasers symmetrically relative to both sides of the rails; according to a preset second time interval or a preset second spatial interval, the laser sensor collects return data; wherein the return data includes: return laser data, current time data corresponding to the return laser data, position data, attitude data of the carrier platform, speed data of the carrier platform, distance data traveled by the carrier platform within the target time, and the attitude data includes roll angle, yaw angle and pitch angle; the return data is saved to a preset synchronization database.

[0007] In an optional embodiment, the step of splicing and vibration compensating the return data to obtain the actual contour of the rail includes: converting the return laser data into a return laser point cloud; performing angle correction on multiple return laser point clouds according to a preset calibration angle; physically splicing the angle-corrected return laser point clouds to obtain a first complete rail point cloud; performing filtering correction on the first complete rail point cloud to obtain a second complete rail point cloud; and performing vibration compensation correction on the second complete rail point cloud to obtain the actual contour of the rail.

[0008] In an optional embodiment, the step of performing vibration compensation correction on the second complete rail point cloud to obtain the actual contour of the rail includes: performing carrier coordinate system conversion on the second complete rail point cloud to obtain a third complete rail point cloud; performing vibration compensation correction on the third complete rail point cloud according to the roll angle, yaw angle and pitch angle in the synchronization database to obtain the actual contour of the rail.

[0009] In an optional implementation, the vibration compensation correction is performed on the third complete rail point cloud according to the roll angle, yaw angle and pitch angle by the following formula: Among them, x s ,y s is the point coordinate of the actual contour, x m ,y m is the point coordinate of the laser sensor device in the third complete rail point cloud; θ R is the roll angle, θ H is the yaw angle, θ P is the pitch angle.

[0010] In an optional embodiment, the actual contour is matched with a pre-acquired standard contour by an iterative nearest point technique to obtain a matching result, including: determining a source point cloud based on the actual contour; determining a target point cloud based on the pre-acquired standard contour; determining corresponding point pairs based on the source point cloud and the target point cloud; constructing a rotation matrix and a translation matrix based on the corresponding point pairs; transforming the source point cloud to the coordinate system of the target point cloud based on the rotation matrix and the translation matrix; determining an error function based on the target point cloud and the transformed source point cloud; judging whether the error function satisfies a preset function condition, and if the error function satisfies the preset function condition, generating a final contour of the rail; matching the final contour with the pre-acquired standard contour to obtain a matching result.

[0011] In an optional embodiment, the step of determining corresponding point pairs based on the source point cloud and the target point cloud includes: selecting a source point cloud that meets a preset curvature rule as the first source point cloud, or selecting a source point cloud whose stability meets a preset stability rule as the first source point cloud; determining corresponding point pairs based on the first source point cloud and the target point cloud.

[0012] In an optional embodiment, the step of determining the loss value of the rail profile based on the matching result includes: comparing the difference between the actual coordinates of the final profile of a preset target position and the standard coordinates of the standard profile based on the matching result to obtain the loss value of the rail profile, wherein the loss value includes at least one of the following: a side loss value, a vertical loss value, and a total loss value.

[0013] In a second aspect, the present invention provides a loss detection device for a rail profile, the loss detection device comprising: a structural laser processing module for emitting a structural laser symmetrically relative to both sides of the rail and collecting return data; a vibration compensation correction module for splicing and vibration compensation correction of the return data to obtain an actual profile; an iteration module for matching the actual profile with a pre-acquired standard profile through an iterative closest point technique to obtain a matching result; and a loss calculation module for obtaining a loss value of the rail profile based on the matching result.

[0014] In a third aspect, the present invention provides a rail profile loss detection system, which is used to execute the loss detection method of the aforementioned embodiment; the loss detection system includes a carrier platform, a server, a laser sensor, an interactive device, a photoelectric encoding device, a positioning device, a posture acquisition device, a clock device, a synchronization device and a storage device; the server, the laser sensor, the interactive device, the photoelectric encoding device, the positioning device, the posture acquisition device, the clock device, the synchronization device and the storage device are all arranged on the carrier platform; the interactive device is used to receive instructions; the synchronization device is used to control the laser sensor, the photoelectric encoding device, the positioning device, the posture acquisition device and the clock device to work synchronously according to the instructions; the laser sensor is used to emit a structure symmetrical relative to both sides of the rail a laser for collecting return data; a photoelectric encoding device for collecting speed data of the carrier platform and the distance data traveled by the carrier platform within the target time; a positioning device for collecting position data of the carrier platform; a posture collection device for collecting posture data of the carrier platform; a clock device for collecting time data; a synchronization device for storing return data, speed data, distance data, position data, posture data and time data in a synchronization database; a storage device for storing the synchronization database; a server for splicing and vibration compensation correction of the return data to obtain the actual profile, and matching the actual profile with the pre-acquired standard profile through the iterative closest point technology to obtain a matching result; and according to the matching result, the loss value of the rail profile is obtained.

[0015] The beneficial effects of the present invention are as follows:

[0016] The present invention provides a method, device and system for detecting the loss of rail profile, which are applied to a carrier platform. The method comprises: emitting structural lasers symmetrically relative to both sides of the rail and collecting return data; splicing and vibration compensation correction of the return data to obtain the actual profile of the rail; matching the actual profile with a pre-acquired standard profile by iterative closest point technology to obtain a matching result; and determining the loss value of the rail profile according to the matching result. The present invention calculates the loss value of the rail profile more accurately by three technical solutions: emitting symmetrical structural lasers, performing vibration compensation correction on the return lasers, and matching the profile by iterative closest point technology.

[0017] Other features and advantages of the present invention will be set forth in the following description, or some features and advantages may be inferred or unambiguously determined from the description, or may be learned by implementing the above-mentioned technology of the present invention.

[0018] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are specifically listed below and described in detail with reference to the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0020] Figure 1 A flow chart of a rail profile loss detection method provided by an embodiment of the present invention;

[0021] Figure 2 A flow chart of a second rail profile loss detection method provided by an embodiment of the present invention;

[0022] Figure 3 A schematic diagram of a fragmented point cloud provided by an embodiment of the present invention;

[0023] Figure 4 A schematic diagram of a returned laser point cloud after angle correction provided by an embodiment of the present invention;

[0024] Figure 5 A schematic diagram of a full-section rail profile provided in an embodiment of the present invention;

[0025] Figure 6 A schematic diagram of matching results provided by an embodiment of the present invention;

[0026] Figure 7 A schematic diagram of the signal-data-information translation process provided by an embodiment of the present invention;

[0027] Figure 8 A schematic diagram of a rail profile loss detection device provided by an embodiment of the present invention;

[0028] Fig. 9 A schematic diagram of a rail profile loss detection system provided by an embodiment of the present invention;

[0029] Fig.10 A schematic diagram of a system provided in accordance with an embodiment of the present invention. DETAILED DESCRIPTION

[0030] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions 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. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0031] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. 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.

[0032] As a key technology for track inspection, dynamic measurement of rail profile (or contour) wear plays an important role in track inspection. The test results can be directly used for the calculation of track geometry parameters, and provide basic data for the detection of rail surface block drop and the study of rail light band anomalies. At present, dynamic profile wear detection technology is mainly divided into contact detection and non-contact detection.

[0033] Contact measurement is also called static measurement. Contact measurement can be divided into two categories: (1) Contact measurement using mechanical equipment, generally measured manually with a caliper. This measuring device is a special measuring tool used to measure horizontal and vertical wear of rails. It uses the rail jaw on the non-wear side of the rail as the reference point, and uses a vernier caliper to measure the wear values ​​of the two wear points. There are different corresponding models of measuring instruments depending on the rail type. (2) Electronic testing equipment developed by combining mechanics, electronics, programmable technology, etc. During testing, the contact detection probe scans the rail shape to obtain the rail shape curve, and the wear value (i.e. loss value) is obtained by comparing it with the standard rail shape.

[0034] Non-contact measurement refers to the detection of the rail state by a line structure laser installed on a track inspection vehicle without contacting the surface of the object being measured. The main principle of line structured light is triangulation. The laser will emit a beam of laser to the target object. Since the outer surface of the target object is rough and uneven, diffuse reflection occurs. Then, the imaging system collects part of the laser beam reflected from the outer surface of the target object at another angle, and the light spot will generate a corresponding image on the camera. When the outer surface of the target object changes position along the laser axis, its reflection angle will also change accordingly, and the corresponding image presented by the light spot on the linear array will also change position accordingly. Then, the position of the presented image and the position of the laser axis are mutually corresponding. Finally, according to the corresponding depth coordinates, the final rail profile can be calculated according to the position relationship, and then the extracted profile can be matched with the standard rail profile to obtain the wear value.

[0035] The limitations of the contact measurement method are: the method of measuring with mechanical equipment is completely dependent on manual labor, which is labor-intensive, inefficient, and has certain safety hazards. Long-term use will cause tool wear and reduce the performance of the measuring equipment. The contact measurement method using electronic equipment also requires manual participation, and also has problems such as heavy workload and harsh environment.

[0036] The limitation of non-contact measurement method is that this method will be affected by ambient light, diffuse reflection, and oil residue on the rails in actual detection. At the same time, the roughness of the rails themselves and the errors of their measuring devices, such as the optical distortion error of the camera lens and the laser displacement error caused by equipment vibration, may affect the final rail profile, which also complicates the extraction of rail profile.

[0037] Based on this, the present invention proposes a rail profile wear detection method, device and system. This technology is an improvement on the non-contact measurement method. It is a rail profile wear detection method that is not affected by lighting conditions, has a complete set of detection items, and is dynamic, continuous, efficient and accurate. This technology belongs to the field of instrument science and technology, and involves a rail profile wear dynamic detection method and device based on multi-sensor integration technology and laser ranging technology. It can be widely used in daily inspections of rail surfaces in rail transit, rail grinding and other scenarios.

[0038] Embodiment 1

[0039] This embodiment provides a rail profile loss detection method, which is applied to a carrier platform.

[0040] Specifically, the above-mentioned carrier platform can be a cart that can be moved along the rails. The cart is equipped with multiple devices that emit structural lasers, namely laser sensors, and these sensors are symmetrical with respect to the two sides of the rails. By manually pushing the cart to move on the rails, the loss of the rail profile can be automatically calculated based on the collected return lasers.

[0041] like Figure 1 As shown, the method includes:

[0042] Step S102, emitting structural lasers symmetrically relative to both sides of the rail, and collecting return data.

[0043] Specifically, the laser sensor is used to emit laser light and collect return data. When collecting the return data, the laser sensor will emit laser light or collect return laser light at intervals or at a certain distance according to the user's instructions. Generally speaking, in this solution, the return data can be collected once every 6 mm (the distance can be set manually) when the cart moves. However, if the data is collected manually, the collection interval cannot be so accurate, and it also wastes manpower and material resources.

[0044] Specifically, the above-mentioned return data not only includes the return laser data, but also includes the time, position, speed, attitude angle and other information corresponding to the collection of the return laser data. All these periodically collected information are matched in space and time, and then packaged and combined as return data, and then stored in the database for subsequent calls.

[0045] Step S104, splicing and vibration compensation correction are performed on the returned data to obtain the actual profile of the rail.

[0046] Specifically, the returned data is the data corresponding to multiple rail positions at each moment (i.e., one instant), and is fragmented, so it needs to be spliced ​​into a preliminary complete outline.

[0047] Specifically, the return data is directly collected by the laser sensor, but the subsequent calculations are all based on the carrier platform. Therefore, the coordinate system of the above-mentioned preliminary complete contour must be converted, that is, from the laser sensor coordinate system to the carrier platform coordinate system.

[0048] Specifically, the carrier platform is a cart that moves along the rails. The rails are not smooth enough, or there are errors in manual operation, so it is inevitable that it will vibrate or shake. Therefore, it is necessary to correct the jitter of the contour based on the attitude angle information collected by the attitude acquisition device to obtain the actual contour of the rails.

[0049] Step S106, matching the actual contour with the pre-acquired standard contour by using an iterative closest point technique to obtain a matching result.

[0050] Specifically, the iterative closest point technology is to perform matrix construction, coordinate system transformation, error function establishment and other operations based on the point cloud corresponding to the actual contour and the point cloud corresponding to the standard contour, so as to obtain a more accurate contour, also known as the final contour. Then the final contour is matched with the edge information of the standard contour to obtain a matching result. Here, the standard contour refers to the contour data in the standard file issued by the Ministry of Railways.

[0051] Step S108, determining the loss value of the rail profile according to the matching result.

[0052] Specifically, the above matching results can be displayed in the form of pictures. The server can then use edge computing software to automatically calculate the difference between the final profile in the matching results and the standard profile, that is, to obtain the loss value of the rail profile.

[0053] The present invention provides a method for detecting the loss of rail profile, which is applied to a carrier platform. The method comprises: emitting a structural laser symmetrical to both sides of the rail and collecting return data; splicing and vibration compensation correction of the return data to obtain the actual profile of the rail; matching the actual profile with a pre-acquired standard profile by iterative closest point technology to obtain a matching result; and determining the loss value of the rail profile according to the matching result. The present invention calculates the loss value of the rail profile more accurately through three technical solutions: periodically emitting and collecting symmetrical structural lasers, performing vibration compensation correction on the return lasers, and matching the profile by iterative closest point technology.

[0054] Embodiment 2

[0055] In a second aspect, this embodiment provides a second rail profile loss detection method, such as Figure 2 As shown, the method includes:

[0056] Step S202: According to a preset first time interval or a preset first spatial interval, the laser sensor emits a structural laser symmetrically relative to both sides of the rail; according to a preset second time interval or a preset second spatial interval, the laser sensor collects return data.

[0057] In the specific implementation, the carrier platform is provided with a laser sensor; the returned data includes: returned laser data, current time data corresponding to the returned laser data, position data, attitude data of the carrier platform, speed data of the carrier platform, and distance data traveled by the carrier platform within the target time; the attitude data includes roll angle, yaw angle and pitch angle.

[0058] Specifically, except for the returned laser data which is collected based on the laser sensor, the other data mentioned above are collected based on the carrier platform (ie, the cart).

[0059] Specifically, the laser sensor has two functions at the same time, which is to emit laser and collect return data. The laser sensor is also called a line structured light profile scanner or a 2D laser scanner.

[0060] Specifically, the above-mentioned rail profile loss detection method is applied to a rail profile loss detection system, which is arranged on a carrier platform, that is, a cart, or it can also be understood that the loss detection system includes a carrier platform. The loss detection system includes a carrier platform (i.e., a cart), a server, a laser sensor, an interaction device, a photoelectric encoding device (also called a photoelectric encoder), a positioning device, a posture acquisition device, a clock device, a synchronization device, and a storage device.

[0061] Specifically, the execution process of the above step S202 is:

[0062] 1) The interactive device receives instructions sent by the user. Specifically, the instructions may be: emit a laser once every 1 second (i.e., the first time interval) or every time the cart moves 6 mm (i.e., the first spatial interval); or, for example, collect return data once every 1 second (i.e., the second time interval) or every time the cart moves 6 mm (i.e., the second spatial interval).

[0063] 2) After receiving the above instructions, the synchronization device immediately receives the speed data of the cart and the distance data of the cart traveled within the target time collected by the photoelectric encoding device; receives the position data of the cart collected by the positioning device; and receives the time data in the clock device.

[0064] Specifically, the above speed data may be embodied in the form of pulse data.

[0065] 3) The synchronization device, according to the user's instructions and the data collected in 2), enables the laser sensor to emit laser according to the user's needs and collect return data according to the user's needs. That is, the synchronization device can synchronously trigger the above devices to work simultaneously to ensure the consistency of the collected return data.

[0066] Specifically, four laser sensors are generally provided, and two of them are symmetrically arranged on the cart relative to the rails on both sides. Therefore, the return data obtained in this way is incomplete, and one laser sensor can only collect data on one side of the rail.

[0067] Step S204, saving the returned data to a preset synchronization database.

[0068] Specifically, the synchronization device will match the multiple returned laser data in the returned data with their corresponding time information and space information, and package and store them in the synchronization database.

[0069] Step S206, converting the returned laser data into a returned laser point cloud.

[0070] Specifically, this step is a data format conversion step, that is, converting laser data into visible fragmented point clouds. Figure 3 As shown, Figure 3 The horizontal coordinate is x and the vertical coordinate is y.

[0071] Step S208, performing angle correction on the multiple returned laser point clouds according to the preset calibration angle.

[0072] Specifically, the interactive device can receive the calibration angle preset by the user. The server performs angle correction on the returned laser point cloud according to the calibration angle. The angle-corrected returned laser point cloud is as follows: Figure 4 shown.

[0073] Step S210, physically splicing the angle-corrected returned laser point cloud to obtain a first complete rail point cloud.

[0074] Specifically, the data after angle correction is still fragmented, so it needs to be spliced ​​into a complete rail point cloud, that is, the first complete rail point cloud.

[0075] Step S212: filtering and correcting the first complete rail point cloud to obtain a second complete rail point cloud.

[0076] Specifically, select a suitable range of through-filtering to filter outliers except for the rails, and obtain the point cloud within the rail range, such as Figure 5 As shown, Figure 5 It is also called full-section rail profile, that is, the second complete rail point cloud.

[0077] Step S214: performing carrier coordinate system transformation on the second complete rail point cloud to obtain a third complete rail point cloud.

[0078] Specifically, the return data is collected by the laser sensor, but the subsequent calculations are all calculated with the carrier platform as the origin (or reference), so a coordinate system conversion step is required.

[0079] The conversion formula is as follows:

[0080]

[0081] Among them, x0 and y0 are the coordinates of the points originally measured by the laser sensor (i.e., the coordinates of the second complete rail point cloud), θ0 is the vertical deviation angle of the laser sensor installation, and d x ,d y are the translation distances (or deviation distances) of the laser sensor and the carrier platform in the X and Y directions respectively, m ,y m is the point coordinate of the laser sensor in the carrier coordinate system (i.e., the coordinate of the third complete rail point cloud).

[0082] Step S216, performing the vibration compensation correction on the third complete rail point cloud according to the roll angle, yaw angle and pitch angle in the synchronization database to obtain the actual profile of the rail.

[0083] Specifically, in general, the movement of the vehicle body on the rail is a serpentine motion, which will cause the scanning angle of the line laser sensor to be unstable. In addition, the vehicle body will shake due to certain external factors when it moves. Therefore, in order to achieve high-precision dynamic measurement of the rail profile, vehicle motion compensation must be performed. This solution uses an inertial sensor (i.e., attitude acquisition device) installed on the rail car to collect the attitude data and acceleration data of the three axes of roll angle, heading angle and pitch angle of the vehicle body (i.e., carrier platform) in real time, and corrects the system attitude. This technology is also called the technology of using inertial navigation to correct the system attitude.

[0084] Specifically, the roll angle, heading angle and pitch angle are called attitude data, which are collected in real time by the attitude acquisition device and packaged with other data and stored in the synchronization database, and then the server can query it from the synchronization database.

[0085] In specific implementation, the vibration compensation correction of the third complete rail point cloud is calculated according to the roll angle, yaw angle and pitch angle by the following formula:

[0086]

[0087] After mathematically simplifying (2), we obtain:

[0088]

[0089] Among them, x s ,y s is the point coordinate of the actual contour, that is, the coordinate of the point after vibration compensation correction, x m ,y m is the point coordinate of the laser sensor device in the third complete rail point cloud; θ R is the roll angle of the carrier platform, θ H is the yaw angle of the carrier platform, θ P is the pitch angle of the carrier platform.

[0090] In specific implementation, steps S214 to S216 are referred to as steps of performing vibration compensation correction on the second complete rail point cloud to obtain the actual contour of the rail.

[0091] Step S218, matching the actual contour with the pre-acquired standard contour by using an iterative closest point technique to obtain a matching result.

[0092] Specifically, the point cloud accuracy of the rail profile section depends on the point cloud registration accuracy of the two lasers on the left and right sides of the single side. Therefore, this embodiment designs an iterative closest point algorithm based on the improved normal vector of the rail curve to better match the data collected by the sensor with the standard data of the Ministry of Railways. The iterative process is calculated by the server.

[0093] The specific steps of step S218 are as follows:

[0094] 1) Determine the source point cloud based on the actual contour.

[0095] Specifically, since this solution uses return data collected by four symmetrical laser sensors and restores the actual contour of the rail based on the four return data, this solution can quickly restore the real contour of the rail by aligning the return data of the left and right lasers, that is, this solution is a three-dimensional acquisition method. That is, the actual contour of this solution is more accurate, and the natural source point cloud is also more accurate.

[0096] 2) Determine the target point cloud based on the pre-acquired standard contour.

[0097] 3) Determine corresponding point pairs based on the source point cloud and the target point cloud.

[0098] 4) Construct rotation matrix and translation matrix based on corresponding point pairs.

[0099] 5) Based on the rotation matrix and translation matrix, the source point cloud is transformed to the coordinate system of the target point cloud.

[0100] 6) Determine the error function based on the target point cloud and the transformed source point cloud;

[0101] 7) Determine whether the error function meets the preset function conditions.

[0102] 8) If the error function satisfies the preset function condition, the final profile of the rail is generated;

[0103] Specifically, if the error function does not satisfy the preset function condition, the iterative calculation continues to be performed. The preset function condition is pre-set by the user through the interactive device and can be changed arbitrarily through the interactive device.

[0104] 9) Match the final contour with the pre-acquired standard contour to obtain a matching result.

[0105] Preferably, a source point cloud that meets a preset curvature rule is selected as the first source point cloud, or a source point cloud whose stability meets a preset stability rule is selected as the first source point cloud; corresponding point pairs are determined based on the first source point cloud and the target point cloud.

[0106] Specifically, the preset curvature rule and the preset stability rule are information contained in the user's instruction received by the interactive device. The preset curvature rule and the preset stability rule can be changed at any time according to user needs.

[0107] Specifically, the iterative closest point algorithm uses the least squares estimation to calculate the transformation matrix. The principle is simple and has good accuracy. However, due to the iterative calculation, the algorithm calculation speed is slow, and the iterative closest point algorithm is prone to fall into a local optimal solution. This embodiment proposes a method for extracting the rail waist points with larger characteristics as key points (i.e., the first source point cloud) based on the change of the normal vector angle for a single-frame two-dimensional point cloud of a line structure, and calculates the curvature of the key points to find the initial matching point pairs, so as to improve the efficiency of the iterative closest point algorithm.

[0108] Specifically, the rail point cloud is segmented into the rail head point cloud and the rail waist and rail bottom point cloud (i.e., the source point cloud whose stability meets the preset stability rule) through threshold segmentation, and the rail waist and rail bottom point cloud is used as the starting point set P to match the standard point set Q of the designed rail. In the kth iteration, the points with large changes are extracted as key points according to the change of the normal vector angle, and a point set S corresponding to the target point set T in the starting point set P is found from Q by calculating the curvature of the key points. The transformation matrix (i.e., the general term for the rotation matrix and the translation matrix) from T (target, standard) to S (actual input points) is calculated, and the point set is updated until the average distance between the two point sets is less than the set threshold L, and the rotation and translation matrix at this time is recorded. According to the calculated translation matrix and rotation matrix, the same rotation and translation transformation is applied to the point cloud of the rail head part, and the transformed point cloud is compared with the point cloud of the rail head with the standard profile.

[0109] Specifically, in layman's terms, the rails are divided into the rail head and the rail waist. Generally, the rail head is more severely worn (considered unstable), while the rail waist is almost not worn (considered relatively stable). Therefore, the source point cloud of the actual contour rail waist (i.e., the first source point cloud) and the target point cloud of the standard contour rail waist are first iterated by trial and error. This makes it easier to iterate successfully, that is, it is easy to match the actual contour rail waist with the standard contour rail waist, reducing the number of trial and error. Then, according to the iterative results of the rail waist, the two rail heads are directly matched (or fused) and compared, which reduces the iterative trial and error process of the rail head part and greatly reduces the workload of iteration.

[0110] Step S220, based on the matching result, the difference between the actual coordinates of the final profile of the preset target position and the standard coordinates of the standard profile is compared to obtain the loss value of the rail profile.

[0111] In a specific implementation, the loss value includes at least one of the following: a side loss value, a vertical loss value, and a total loss value.

[0112] Specifically, the above matching results are as follows: Figure 6 As shown. Figure 6By comparing the two edge lines (one edge line is a schematic diagram of the standard profile issued by the Ministry of Railways, and the other edge line is a schematic diagram of the final profile fitted), the loss value of the rail profile can be obtained. Figure 6 The horizontal coordinate is x and the vertical coordinate is y. The standard profile is the data input by the user received by the interactive device, or obtained by the server from the network. According to the regulations of the Ministry of Railways, the target position refers to the vertical loss measured at 1 / 3 of the width of the top surface of the rail (from the standard working edge), and the side loss is 16mm below the tread of the rail (according to the standard section). The total wear is the sum of the vertical loss and the 1 / 2 side loss.

[0113] This embodiment proposes a rail profile wear detection technology solution using an on-track artificial push trolley equipped with multi-source sensors. The solution can quickly and dynamically detect the actual profile of the rail and the rail wear (also known as loss). Compared with dynamic detection equipment, this case has faster acquisition speed and higher measurement accuracy. This case has a high degree of integration, supports rapid synchronous positioning between sensors, and compensates for the vibration effects generated during the movement of the carrier platform through the attitude acquisition device. This method is simpler than traditional fitting linearity and can more effectively suppress the error caused by vibration. The iterative nearest point matching algorithm adopted in this solution can quickly and accurately match and fuse with the standard rail profile, and then calculate the rail wear value. Compared with static detection equipment, this case belongs to intelligent detection equipment, and its detection efficiency and accuracy are far superior to static measurement. The device described in this case only requires one operator and can support running on the track at a speed of about 6km / h, and supports the collection of two profile data of the left and right tracks every 6mm. The time for manual measurement to collect single profile data is about 1 to 2 minutes, and the frequent bending operation is labor-intensive.

[0114] Embodiment 3

[0115] This embodiment provides a third rail profile loss detection method, which specifically includes:

[0116] 1. On the trolley (also called the carrier platform), four structural laser sensors (which can emit structural lasers and collect lasers returned from the rails) are set symmetrically with respect to the rails on both sides, and a synchronization device (also called a synchronization board) is set. The synchronization device is connected to a clock device (collecting time data), a positioning device (collecting positioning data), a posture sensing device (collecting the roll angle, yaw angle and pitch angle of the trolley in real time), a photoelectric encoding device (which can collect the speed and travel distance of the trolley), and a server is also set on the trolley.

[0117] 2. The server uses a synchronization device to make the four laser sensors symmetrical to the rails on both sides emit four beams of structured laser at the same time, and synchronizes the laser sensors to collect the returned laser data. The server also uses a clock device, a positioning device, and a photoelectric encoding device to package and store the time, speed, positioning, posture, and returned laser data to obtain a synchronization file for easy access at any time. The spatial interval and time interval for emitting and collecting the returned laser can be pre-set, for example, the car emits or collects a laser every 6 mm, or emits or collects a laser every 1 second.

[0118] 3. The collected return laser data is fragmented and cannot fully reflect the overall shape of the rail.

[0119] 4. According to the preset calibration angle, the fragmented return laser data are stitched together, that is, the deflection angle of the return laser data is corrected and stitched together to obtain complete return laser data.

[0120] 5. Select a suitable range for filtering the complete returned laser data, filter out the outliers except the rails, obtain the point cloud within the rail range, and obtain the filtered data, that is, the full-section rail profile.

[0121] 6. However, the full-section rail profile only solves the fragmentation problem. The data is still based on data collected by sensors rather than by a cart. Therefore, it is necessary to correct the error of the coordinate system and obtain the profile after correcting the coordinate system error.

[0122] 7. The car will shake violently during driving, so the contour after correcting the coordinate system error should be compensated for the shaking according to the roll angle, yaw angle and pitch angle to obtain the contour after shaking correction.

[0123] 8. Using the iterative closest point technique, the above jitter-corrected profile is compared and matched with the pre-acquired standard profile to obtain the rail profile loss.

[0124] 1) Based on the jitter-corrected contour, the source point cloud is obtained.

[0125] 2) Based on the pre-acquired standard contour, the target point cloud is obtained.

[0126] 3) Based on the source point cloud and the target point cloud, corresponding point pairs are obtained.

[0127] 4) Based on the corresponding point pairs, construct the rotation matrix and translation matrix.

[0128] 5) Based on the rotation matrix and the translation matrix, the source point cloud is transformed into the coordinate system of the target point cloud to obtain the transformed source point cloud.

[0129] 6) Calculate the error function between the transformed source point cloud and the target point cloud. If the error function value is greater than a preset threshold, iterate the above operation until the given error requirement is met, that is, until it is less than or equal to the preset threshold, stop the iteration, and obtain the final contour of the target rail.

[0130] The above 1)-6) is an iterative process, which uses least squares estimation to calculate the transformation matrix. The principle is simple and has good accuracy. However, due to the iterative calculation, the algorithm calculation speed is slow and it is easy to fall into the local optimal solution.

[0131] In order to solve the problem of slow calculation speed or falling into the local optimal solution, the following optimization scheme is adopted:

[0132] ① Find the initial matching point by calculating the curvature of the key point.

[0133] ② By extracting points with larger or more stable features (such as rail waist points) for matching, after the actual rail waist is matched with the standard rail waist (that is, the final contour of the rail waist has been obtained), the same rotation and translation transformation is applied to the point cloud of the rail head, thus reducing the trial and error process of the rail head matrix transformation. Here, the final contour of the rail head + the final contour of the rail waist = the final contour of the target rail.

[0134] 7) Generally speaking, it is believed that the rail waist is stable, not easy to wear, and relatively complete. Therefore, this application mainly performs wear detection on the rail head part, that is, the rail head part is compared with the standard data to obtain the wear value.

[0135] According to the regulations of the railway department, vertical wear is measured at 1 / 3 of the width of the rail top surface (from the standard working edge), and side wear is measured at 16mm below the rail tread (according to the standard section). The total wear is the sum of vertical wear and 1 / 2 side wear. Therefore, side wear, vertical wear and total wear can be calculated.

[0136] Based on the above functional requirements, this technology focuses on the signal-data-information translation process. Figure 7 The system design shown in the figure includes three layers: physical layer, data layer and logical layer.

[0137] The physical layer is a description of the integration of multi-source heterogeneous sensors. The physical layer integrates photoelectric encoding devices, GNSS (Global Navigation Satellite System, i.e. positioning devices), IMU (i.e. attitude acquisition devices), and line structured light scanners (i.e. laser sensors). Through multi-source laser sensors, the changes in the real space that the system is concerned about are converted into photoelectric signals; the system lower computer uses a synchronous control circuit to synchronously trigger and acquire the multi-source sensors of the physical layer, and obtain various data in the data layer such as mileage (i.e. distance data), system position (i.e. position data), system attitude, and cross-sectional laser data (i.e. return laser data).

[0138] The data layer is the interaction layer between multi-source sensors and the host computer. It is responsible for translating the signals of the physical layer into data and realizing the mutual transmission of data between the host computer and the physical layer. Signal translation follows a specific protocol, which is agreed upon by the internal scheme and is not within the scope of this patent protection; the logic layer is responsible for data analysis and processing. Through a series of operations, the profile and wear of the rail are calculated.

[0139] This case first designed a rail profile wear detection trolley (or device) that uses an on-track artificial push trolley equipped with multi-source sensors. The system is mainly composed of four linear structure laser sensors, a set of IMU (Inertial Measurement Unit, inertial navigation unit), a GPS global positioning system (including antenna and receiver), a synchronous control board and a server.

[0140] This solution uses a manual push cart as a carrier to work on the track. The GPS system synchronizes the time with the lower computer, thereby synchronizing the time of each sensor, the IMU obtains the posture of each sensor, and the pulse count of the photoelectric encoding device is synchronized for positioning. The synchronization control board (i.e., the synchronization device) sends parsing instructions to the lower computer through the network, notifying the lower computer to receive the profile data from the line structured light profile scanner, GPS data, IMU data, and pulse data from the photoelectric encoding device, and finally transmits the data to the upper computer for solution. Through a series of operations such as error correction, data fusion, and model matching, the profile and wear value of the rail are finally obtained.

[0141] The rail profile loss detection method provided in the embodiment of the present invention has the same technical features as the rail profile loss detection method provided in the above embodiment, so it can also solve the same technical problems and achieve the same technical effects.

[0142] Embodiment 4

[0143] The embodiment of the present invention provides a rail profile loss detection device, such as Figure 8 As shown, the loss detection device comprises:

[0144] The structural laser processing module 81 is used to emit structural lasers symmetrically relative to both sides of the rail and collect return data;

[0145] A vibration compensation correction module 82 is used to perform splicing and vibration compensation correction on the returned data to obtain the actual contour;

[0146] An iteration module 83, used for matching the actual contour with the pre-acquired standard contour by an iterative closest point technique to obtain a matching result;

[0147] The loss calculation module 84 is used to obtain the loss value of the rail profile according to the matching result.

[0148] The present invention provides a loss detection device for rail profile, which is applied to a carrier platform. The method comprises: emitting structural lasers symmetrically relative to both sides of the rail and collecting return data; splicing and vibration compensation correction of the return data to obtain the actual profile of the rail; matching the actual profile with a pre-acquired standard profile by iterative closest point technology to obtain a matching result; and determining the loss value of the rail profile according to the matching result. The present invention calculates the loss value of the rail profile more accurately by three technical solutions: emitting symmetrical structural lasers, performing vibration compensation correction on the return lasers, and matching the profile by iterative closest point technology.

[0149] The rail profile loss detection device provided in the embodiment of the present invention has the same technical features as the rail profile loss detection method provided in the above embodiment, so it can also solve the same technical problems and achieve the same technical effects.

[0150] Embodiment 5

[0151] The embodiment of the present invention provides a rail profile loss detection system, wherein the loss detection system 90 is used to execute the above-mentioned loss detection method; Fig. 9 As shown, the loss detection system includes a carrier platform 91, a server 92, a laser sensor 93, an interactive device 94, a photoelectric encoding device 95, a positioning device 96, a posture acquisition device 97, a clock device 98, a synchronization device 99 and a storage device 910; the server, the laser sensor, the interactive device, the photoelectric encoding device, the positioning device, the posture acquisition device, the clock device, the synchronization device and the storage device are all arranged on the carrier platform.

[0152] The interactive device is used for receiving instructions.

[0153] Specifically, the interactive device is used to receive instructions sent by the user, such as the time and space for emitting or collecting laser signals, or receive a preset standard profile or preset rule sent by the user.

[0154] The synchronization device is used to control the laser sensor, the photoelectric encoding device, the positioning device, the posture acquisition device and the clock device to work synchronously according to instructions.

[0155] The laser sensor is used to emit structural laser symmetrically on both sides of the rail and collect return data.

[0156] The photoelectric encoding device is used to collect the speed data of the carrier platform and the distance data traveled by the carrier platform within the target time.

[0157] The positioning device is used to collect the position data of the carrier platform.

[0158] The posture acquisition device is used to collect the posture data of the carrier platform.

[0159] Clock device, used to collect time data.

[0160] The synchronization device is also used to store the return data, speed data, distance data, position data, posture data and time data into the synchronization database.

[0161] The storage device is used to store the synchronization database.

[0162] The server is used to splice the returned data and perform vibration compensation correction to obtain the actual contour, and match the actual contour with the pre-acquired standard contour through the iterative closest point technology to obtain the matching result; based on the matching result, the loss value of the rail contour is obtained.

[0163] Specifically, the system physical diagram of this scheme is as follows Fig.10 shown.

[0164] In actual use, the worker 102 can push the carrier platform 91 (or the cart) and move along the rail 101. Four laser sensors 93 are symmetrically arranged on the carrier platform, which can respectively emit lasers to the rails on both sides and collect return data respectively.

[0165] The rail profile loss detection system provided in the embodiment of the present invention has the same technical features as the rail profile loss detection method provided in the above embodiment, so it can also solve the same technical problems and achieve the same technical effects.

[0166] 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 replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for detecting rail profile loss, characterized in that: Applied to a carrier platform, the method comprises: Emit structured lasers symmetrically on both sides of the rail and collect return data; The returned data are spliced ​​and corrected for vibration compensation to obtain the actual profile of the rail; The rail point cloud is divided into a rail head point cloud and a rail waist and rail bottom point cloud, and the actual contour is matched with the pre-acquired standard contour by an iterative closest point technology to obtain a matching result; Determining the loss value of the rail profile according to the matching result; Wherein, the carrier platform is provided with a laser sensor; The steps of emitting structured lasers symmetrically relative to both sides of the rail and collecting return data include: According to a preset first time interval or a preset first spatial interval, the laser sensor emits a structural laser symmetrically relative to two sides of the rail; According to a preset second time interval or a preset second spatial interval, the laser sensor collects the return data; wherein the return data includes: return laser data, current time data corresponding to the return laser data, position data, attitude data of the carrier platform, speed data of the carrier platform, and distance data traveled by the carrier platform within a target time, and the attitude data includes a roll angle, a yaw angle, and a pitch angle; Saving the returned data to a preset synchronization database; The step of splicing and vibration compensation correction of the returned data to obtain the actual profile of the rail includes: Converting the returned laser data into a returned laser point cloud; According to a preset calibration angle, the angle of the returned laser point clouds is corrected; Physically splicing the angle-corrected returned laser point cloud to obtain a first complete rail point cloud; Performing filtering and correction on the first complete rail point cloud to obtain a second complete rail point cloud; Performing vibration compensation correction on the second complete rail point cloud to obtain the actual profile of the rail; The step of performing vibration compensation correction on the second complete rail point cloud to obtain the actual profile of the rail comprises: Performing carrier coordinate system transformation on the second complete rail point cloud to obtain a third complete rail point cloud; The vibration compensation correction is performed on the third complete rail point cloud according to the roll angle, the yaw angle and the pitch angle in the synchronization database to obtain the actual profile of the rail.

2. The method according to claim 1, characterized in that The vibration compensation correction is performed on the third complete rail point cloud according to the roll angle, the yaw angle and the pitch angle by calculating the following formula: Among them, x s ,y s is the point coordinate of the actual contour, x m ,y m θ is the point coordinate of the laser sensor device on the third complete rail point cloud; R is the roll angle, θ H is the yaw angle, θ P is the pitch angle.

3. The method according to claim 1, characterized in that The step of matching the actual contour with the pre-acquired standard contour by iterative closest point technology to obtain a matching result comprises: Determine a source point cloud according to the actual contour; Determine a target point cloud according to the pre-acquired standard contour; Determining corresponding point pairs based on the source point cloud and the target point cloud; Constructing a rotation matrix and a translation matrix based on the corresponding point pairs; Based on the rotation matrix and the translation matrix, transform the source point cloud to the coordinate system of the target point cloud; Determine an error function based on the target point cloud and the transformed source point cloud; Determining whether the error function satisfies a preset function condition, and if the error function satisfies the preset function condition, generating a final profile of the rail; The final contour is matched with the pre-acquired standard contour to obtain the matching result.

4. The method according to claim 3, characterized in that: The step of determining corresponding point pairs based on the source point cloud and the target point cloud comprises: Selecting the source point cloud that meets the preset curvature rule as the first source point cloud, or selecting the source point cloud that meets the preset stability rule as the first source point cloud; The corresponding point pairs are determined based on the first source point cloud and the target point cloud.

5. The method according to claim 3, characterized in that: The step of determining the loss value of the rail profile according to the matching result comprises: According to the matching result, the difference between the actual coordinates of the final profile of the preset target position and the standard coordinates of the standard profile is compared to obtain the loss value of the rail profile, wherein the loss value includes at least one of the following: side loss value, vertical loss value and total loss value.

6. A rail profile loss detection device, characterized in that: Applied to a carrier platform, the loss detection device comprises: The structural laser processing module is used to emit structural lasers symmetrically relative to both sides of the rail and collect return data; A vibration compensation correction module is used to perform splicing and vibration compensation correction on the returned data to obtain an actual contour; An iterative module is used to segment the rail point cloud into a rail head point cloud and a rail waist and rail bottom point cloud, and match the actual contour with a pre-acquired standard contour through an iterative closest point technology to obtain a matching result; A loss calculation module, used for obtaining the loss value of the rail profile according to the matching result; Wherein, the carrier platform is provided with a laser sensor; the steps of emitting structural lasers symmetrically relative to both sides of the rails and collecting return data include: according to a preset first time interval or a preset first spatial interval, the laser sensor emits structural lasers symmetrically relative to both sides of the rails; according to a preset second time interval or a preset second spatial interval, the laser sensor collects the return data; wherein, the return data includes: return laser data, current time data corresponding to the return laser data, position data, posture data of the carrier platform, speed data of the carrier platform, and distance data traveled by the carrier platform within the target time, and the posture data includes roll angle, yaw angle and pitch angle; the return data is saved to a preset synchronization database; The step of performing splicing and vibration compensation correction on the return data to obtain the actual profile of the rail includes: converting the return laser data into a return laser point cloud; performing angle correction on a plurality of the return laser point clouds according to a preset calibration angle; physically splicing the angle-corrected return laser point clouds to obtain a first complete rail point cloud; performing filtering correction on the first complete rail point cloud to obtain a second complete rail point cloud; performing vibration compensation correction on the second complete rail point cloud to obtain the actual profile of the rail; The step of performing vibration compensation correction on the second complete rail point cloud to obtain the actual profile of the rail includes: performing carrier coordinate system conversion on the second complete rail point cloud to obtain a third complete rail point cloud; performing vibration compensation correction on the third complete rail point cloud according to the roll angle, the yaw angle and the pitch angle in the synchronization database to obtain the actual profile of the rail.

7. A rail profile loss detection system, characterized in that: The loss detection system is used to perform the loss detection method according to any one of claims 1 to 5; The loss detection system includes a carrier platform, a server, a laser sensor, an interaction device, a photoelectric encoding device, a positioning device, a posture acquisition device, a clock device, a synchronization device and a storage device; The server, the laser sensor, the interactive device, the photoelectric encoding device, the positioning device, the posture acquisition device, the clock device, the synchronization device and the storage device are all arranged on the carrier platform; The interaction device is used to receive instructions; The synchronization device is used to control the laser sensor, the photoelectric encoding device, the positioning device, the posture acquisition device and the clock device to work synchronously according to the instruction; The laser sensor is used to emit structural lasers symmetrically relative to both sides of the rail and collect return data; The photoelectric encoding device is used to collect speed data of the carrier platform and distance data traveled by the carrier platform within a target time; The positioning device is used to collect the position data of the carrier platform; The posture acquisition device is used to collect the posture data of the carrier platform; The clock device is used to collect time data; The synchronization device is further used to store the return data, the speed data, the distance data, the position data, the posture data and the time data into a synchronization database; The storage device is used to store the synchronization database; The server is used to splice and perform vibration compensation correction on the returned data to obtain an actual profile, match the actual profile with a pre-acquired standard profile through an iterative closest point technology to obtain a matching result; and obtain a loss value of the rail profile based on the matching result.

Images