Train open-top wagon measurement and inspection system and method

Abstract

A train open-top wagon measurement and inspection system and method. The system comprises: a LiDAR group, which is used for generating a laser to scan a train and collecting profile information of the train, wherein the profile information is point cloud data; a wagon identification apparatus, which is used for measuring the wheelbase of the train, determining the wagon type of each wagon of the train on the basis of the wheelbase and the point cloud data, and generating a coupler signal between wagons of the train; and a detection module, which is used for segmenting, on the basis of the coupler signal, the profile information of the train into wagon profile information taking a wagon as a unit, calculating the volume of loaded objects in the wagon on the basis of the wagon profile information when the wagon is an open-top wagon, and identifying foreign matters in the loaded objects.

Description

Train open box measurement and inspection system and method

[0001] This application claims priority to Chinese patent application No. 202311865604.8 filed on December 29, 2023, the contents of which are incorporated herein by reference. Technical Field

[0002] The present disclosure relates to the field of train inspection, and in particular to a train open box measurement and inspection system and method. Background Art

[0003] The train open car measurement and inspection system is mainly used in the field of train coal transport car safety inspection technology. During the operation of the coal transport train, the load information in the open car is obtained through sensors to determine the loading status of the car.

[0004] The operating principle of a related open-carriage train measurement and inspection system is as follows: first, an area scan camera or line scan camera is installed on the top of the gantry to capture images of the train carriages. Then, sensors are installed on the sides of the gantry or on the rails to measure speed or segment the carriages. Finally, each segmented carriage image is identified. However, area scan cameras or line scan cameras have high lighting requirements and can only determine the presence of loads within the carriages through images; they cannot measure the volume of the loads. Summary of the Invention

[0005] The present disclosure provides a train open box measurement and inspection system and method.

[0006] According to the first aspect of the present disclosure, a train open box measurement and inspection system is provided, wherein the system includes: a laser radar group, used to generate laser to scan the train and collect contour information of the train, wherein the contour information is point cloud data; a car identification device, used to measure the wheelbase of the train, determine the car type of each car in the train based on the wheelbase and the point cloud data, and generate hook signals between the cars of the train; and a processor, which is communicatively connected to the car identification device and the laser radar group, and is configured to: divide the contour information of the train into car contour information in units of cars based on the hook signals; and calculate the volume of the load in the car based on the car contour information when the car type is an open box, and identify foreign objects in the load.

[0007] In some embodiments, the car identification device includes: a magnetic steel sensor group, including multiple magnetic steel sensors, and the multiple magnetic steel sensors are arranged on one side of the train track with a predetermined installation spacing, for respectively collecting the arrival time of the train wheels; the processor is also configured to: calculate the speed of the train based on the time difference of the arrival times collected by the multiple magnetic steel sensors in the magnetic steel sensor group and the installation spacing of the multiple magnetic steel sensors; calculate the wheelbase of the train based on the time difference between adjacent wheels sensed by the same magnetic steel sensor in the magnetic steel sensor group and the speed; and determine the car type and the hook signal based on the wheelbase and the point cloud data.

[0008] In some embodiments, the magnetic steel sensor group includes: an upward magnetic steel sensor group, arranged in the upward direction of the train track, for sensing the arrival time of a train coming from the upward direction; and / or, a downward magnetic steel sensor group, arranged in the downward direction of the train track, for sensing the arrival time of a train coming from the downward direction.

[0009] In some embodiments, the laser radar group includes: a first laser radar, which is arranged above the train track and at a height higher than the height of the train, and is used to scan the top of the train to obtain the top contour information of the train; and a second laser radar, which is arranged on one side of the train track and at a height higher than the bottom surface of the train track and lower than the height of the first laser radar, and is used to scan the side of the train to obtain the side contour information of the train.

[0010] In some embodiments, the scanning surfaces of the first lidar and the second lidar are both perpendicular to the traveling direction of the train.

[0011] In some embodiments, the laser radar group also includes: a gantry, mounted on the train track, including a beam and two brackets, the beam spanning the train track, and the two brackets are respectively arranged on both sides of the train track to support the beam; the first single-line laser radar is arranged in the middle of the beam; and the second single-line laser radar is arranged on at least one of the two brackets.

[0012] In some embodiments, the processor is further configured to: calculate the translation step length of the contour information of two adjacent frames based on the speed of the train and the scanning frequency of the lidar group; map the multiple frames of contour information continuously scanned and collected by the lidar group to the same point cloud coordinate system according to the translation step length to obtain the overall contour information of the train; and divide the overall contour information of the train into car contour information in units of cars based on the hook signal.

[0013] In some embodiments, the vehicle box contour information includes vehicle box top contour information and vehicle box side contour information; and the processor is further configured to: filter the box point cloud in the vehicle box top contour information according to the height change trend of the point cloud in the vehicle box top contour information to obtain the top contour information of the load; obtain the bottom height of the vehicle box according to the vehicle box side contour information, and calculate the height of each point in the vehicle box top contour information relative to the bottom of the vehicle box; rasterize the vehicle box top contour information, and calculate the average height of the points in the vehicle box top contour information in each grid relative to the bottom of the vehicle box; calculate the unit volume of the load in each grid according to the area of ​​the grid and the average height of the points in the grid, and calculate the sum of the unit volumes of the load in all the grids to obtain the total volume of the load in the vehicle box.

[0014] In some embodiments, the processor is further configured to: calculate the elevation difference between the maximum height and the minimum height of each point in the grid; determine whether the elevation difference exceeds a first preset threshold; and when the elevation difference exceeds the first preset threshold, determine that there is foreign matter in the load.

[0015] In some embodiments, the processor is further configured to: calculate the average height of the points in all grids adjacent to the grid to obtain the average height of the adjacent grid point cloud; compare the average height of the points in the grid with the average height of the adjacent grid point cloud; and when the difference between the average height of the points in the grid and the average height of the adjacent grid point cloud is greater than a second preset threshold, determine that there is foreign matter in the load.

[0016] In some embodiments, the processor is further configured to convert the vehicle box contour information into an image; and the system further comprises a memory configured to store the image.

[0017] According to a second aspect of the present disclosure, a method for measuring and inspecting an open box of a train is provided, wherein the method comprises: generating a laser to scan the train and collecting contour information of the train, wherein the contour information is point cloud data; measuring the wheelbase of the train, determining the car type of each car in the train based on the wheelbase and the point cloud data, and generating hook signals between the cars of the train; dividing the contour information of the train into car contour information in units of cars based on the hook signals; and when the car type is an open box, calculating the volume of the load in the car based on the car contour information, and identifying foreign objects in the load. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to better understand the present disclosure, the present disclosure will be described in detail according to the following drawings:

[0019] FIG1 is a schematic structural diagram of a train open box measurement and inspection system according to some exemplary embodiments of the present disclosure;

[0020] FIG2 is a schematic diagram of train wheelbase identification according to some exemplary embodiments of the present disclosure;

[0021] FIG3 is a schematic diagram of hook signal calculation according to some exemplary embodiments of the present disclosure;

[0022] FIG4 is a scanning principle diagram of a laser radar group according to some exemplary embodiments of the present disclosure;

[0023] FIG5A is a schematic diagram of train top contour information according to some exemplary embodiments of the present disclosure;

[0024] FIG5B is a schematic diagram of train side profile information according to some exemplary embodiments of the present disclosure;

[0025] FIG6A is a workflow of a processor processing train profile information according to some exemplary embodiments of the present disclosure;

[0026] FIG6B is a schematic diagram showing the composition of vehicle box contour information according to some exemplary embodiments of the present disclosure;

[0027] FIG7A schematically illustrates the height difference between a train car and a load according to some exemplary embodiments of the present disclosure;

[0028] FIG7B schematically shows a schematic diagram of vehicle box top contour information according to some exemplary embodiments of the present disclosure;

[0029] FIG7C schematically shows a schematic diagram of the top contour information of a vehicle box after filtering out the box point cloud according to some exemplary embodiments of the present disclosure;

[0030] FIG8 schematically shows a schematic diagram of one method of determining whether a load contains foreign matter according to an embodiment of the present disclosure;

[0031] FIG9 schematically shows a flow chart of a train open box measurement and inspection method according to the present disclosure. DETAILED DESCRIPTION

[0032] Specific embodiments of the present disclosure will be described in detail below. It should be noted that the embodiments described herein are intended to be illustrative only and are not intended to limit the present disclosure. In the following description, a large number of specific details are set forth to provide a thorough understanding of the present disclosure. However, it will be apparent to those skilled in the art that these specific details are not necessarily required to practice the present disclosure. In other examples, known structures, materials, or methods are not specifically described to avoid obscuring the present disclosure.

[0033] Throughout this specification, references to "one embodiment," "an embodiment," "an example," or "an example" mean that a particular feature, structure, or characteristic described in connection with the embodiment or example is included in at least one embodiment of the present disclosure. Thus, appearances of the phrases "in one embodiment," "in an embodiment," "an example," or "an example" in various places throughout this specification are not necessarily all referring to the same embodiment or example. Furthermore, the particular features, structures, or characteristics may be combined in any suitable combinations and / or subcombinations in one or more embodiments or examples. Furthermore, one of ordinary skill in the art will understand that the term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0034] The terms used herein are only for describing specific embodiments and are not intended to limit the present disclosure. The terms "comprise," "include," etc. used herein indicate the presence of features, steps, operations, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, or components.

[0035] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those skilled in the art unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification and should not be interpreted in an idealized or overly rigid manner.

[0036] Related technologies usually use area array cameras or line array cameras to collect train car information, which has high requirements for ambient lighting. In addition, the images can only determine whether there are loads in the car, but cannot measure the volume of the loads.

[0037] The disclosed embodiments provide a train open box measurement and inspection system, comprising: a car identification device, a laser radar group, and a processor. The laser radar group is used to generate a laser scan of the train and collect the train's contour information, which is point cloud data. The car identification device is used to measure the train's wheelbase, determine the car type of each car in the train based on the wheelbase and point cloud data, and generate hook signals between the train cars. The processor is communicatively connected to the car identification device and the laser radar group, and is configured to: segment the train's contour information into car box contour information based on the hook signals; and, when the car type is an open box, calculate the volume of the load in the car based on the car box contour information and identify foreign objects in the load.

[0038] FIG1 is a schematic structural diagram of a train open box measurement and inspection system according to some exemplary embodiments of the present disclosure.

[0039] As shown in Figure 1, in some embodiments, the car identification device may include a magnetic sensor group, which includes a plurality of magnetic sensors (S0~S1 and / or X0~X1). The plurality of magnetic sensors are arranged on one side of the train track T with a predetermined installation spacing, and are used to respectively collect the arrival time of the wheels of the train.

[0040] The magnetic sensor assembly is installed inside the train tracks. The magnetic sensor operates by generating a static magnetic field around the magnetic probe. When a ferromagnetic object, such as a rifle or vehicle, enters this static magnetic field, a new magnetic field is induced. The target's motion creates interference that changes the magnetic field, causing the magnetometer needle to deflect and oscillate, generating an electrical signal that enables detection of people and vehicles. As the train's wheels sequentially pass each magnetic sensor in the assembly, each sensor outputs an electrical signal. The moment this signal is generated represents the arrival time of the wheel.

[0041] Optionally, the number of magnetic steel sensors included in the magnetic steel sensor group is at least 2, so as to obtain the time difference of the arrival times collected by multiple magnetic steel sensors. In order to further improve the calculation accuracy of the time difference, the number of magnetic steel sensors can be any number greater than 2, which is not limited here. Determining the passage of train wheels based on the sensing data of at least two magnetic steel sensors can avoid detection errors caused by some special circumstances. For example, when other objects enter the sensing range of a certain magnetic steel sensor for a short time or a certain magnetic steel sensor itself has an error or failure, the sensing data of multiple magnetic steel sensors usually differ from the regularity of the train wheels passing through each magnetic steel sensor in turn, thereby effectively eliminating the situation where the train wheels are not passing. Among three or more magnetic steel sensors, some magnetic steel sensors can be set as redundant magnetic steel sensors, which can be switched with the faulty magnetic steel sensor in the event of a failure to ensure the continuity of detection.

[0042] Optionally, the magnetic steel sensor group can be further divided into an upward magnetic steel sensor group (S0-S1) and a downward magnetic steel sensor group (X0-X1). As shown in Figure 1, the upward magnetic steel sensor group is located in the upward direction of the train track T to sense the arrival time of trains coming from the upward direction; the downward magnetic steel sensor group is located in the downward direction of the train track T to sense the arrival time of trains coming from the downward direction.

[0043] Optionally, a group of magnetic steel sensors may be used to collect both the arrival time of trains in the upward direction and the arrival time of trains in the downward direction, thereby reducing the use of magnetic steel sensors.

[0044] When the magnetic sensors detect a train's arrival time, the processor calculates the train's speed based on the time difference between the arrival times detected by multiple magnetic sensors in the magnetic sensor group and the spacing between the sensors. Furthermore, the processor calculates the train's wheelbase based on the time difference and speed detected by adjacent wheels sensed by the same magnetic sensor in the group. Based on the wheelbase and point cloud data, the processor determines the car type and the hook signals between adjacent cars.

[0045] Assuming that the magnetic steel sensor group collects the train arrival time difference Δt1, and the installation spacing of the magnetic steel sensors is x, the processor can calculate the train speed according to the formula v=x / Δt1.

[0046] Assuming that the time difference between the adjacent wheels sensed by the same magnetic steel sensor is Δt2, the wheelbase of adjacent wheels of the train is:

[0047] l=Δt2*v.

[0048] Alternatively, assuming that v1 is the vehicle speed calculated based on the time difference between the arrival of the previous wheel collected by the magnetic sensor group and the installation spacing of the magnetic sensors, and v2 is the vehicle speed calculated based on the time difference between the arrival of the next wheel collected by the magnetic sensor group and the installation spacing of the magnetic sensors, and the time difference between the adjacent wheels sensed by the same magnetic sensor is Δt2, then the wheelbase of adjacent wheels of the train is:

[0049] The processor segments the wheelbase according to a preset segmentation process, identifies the vehicle body type of the segmented wheelbase according to a preset vehicle body identification process, and calculates the hook signal of the vehicle body.

[0050] "Segmentation" refers to separating the collected wheelbase data of a train into actual carriage sections. Currently, most train carriages in China have 4, 5, 6, or 8 axles. Their wheelbases meet the following three rules:

[0051] Rule 1: The wheelbase of the carriage is symmetrical about the center point of the carriage. As shown in Figure 2, the L 1,2 =L 5,6 、L 2,3 =L 4,5 , L in car 1 7,8 =L 9,10 (where L i,j Refers to the distance between the i-th wheel and the j-th wheel, the same below);

[0052] Rule 2: The distance from the first wheel to the last wheel of a carriage is greater than 7 meters. As shown in Figure 2, the L 1,6 >7 meters, L in carriage 1 7,10 >7 meters;

[0053] Rule 3: The wheelbase between the two bogies is greater than the wheelbase at the hook, and the wheelbase at the hook is greater than the wheelbase of the bogies. For example, the L in the locomotive in Figure 2 3,4 >L 6,7 >L 1,2 , L in car 1 8,9 >L 10,11 >L 7,8 .

[0054] The method of segmenting the train wheelbase is as follows.

[0055] Initialization: Set T and N to 1, and i to 0. T indicates that the train wheelbases before the Tth wheelbase have been divided into single carriages, N indicates that the train is currently dividing carriages using the Nth wheelbase, and i indicates the number of wheelbases that have not yet been divided.

[0056] Regularly reading a wheelbase: When two wheels of a train pass a set of magnetic sensors, the data acquisition card connected to these sensors immediately calculates the wheelbase information and sends it to the car identification computer. The car identification computer reads this wheelbase information and increments the number of wheelbases that can be divided, i, by 1.

[0057] Satisfy the 4-axle rule: Apply the 3 wheelbase rules to the 4-axle. That is, if it is a 4-axle vehicle, it must satisfy the following 4 rules:

[0058] Rule 1: Is the Nth wheelbase value approximately equal to the N+2th wheelbase value, that is, is the absolute value of the difference between the Nth wheelbase value and the N+2th wheelbase value less than 100 mm?

[0059] Rule 2: Is the sum of the wheelbases of N, N+1, and N+2 greater than 7000 mm?

[0060] Rule 3: Is the N+1th wheelbase value greater than the N+3th wheelbase value?

[0061] Rule 4: Whether the N+3th wheelbase value is greater than the Nth wheelbase value.

[0062] When i≤2, that is, the number of wheelbases that have not yet been divided is less than 3, the four-axis regularity analysis is not performed because there is insufficient data available for analysis.

[0063] When i = 3, that is, the number of wheelbases that have not yet been divided is equal to 3, the three wheelbases (Nth to N+3th) can be analyzed for rules 1 and 2. If the rules do not meet, it is considered that the current division does not conform to the 4-axis rule; if the rules meet, it is considered that the current division may conform to the 4-axis rule, and the next axis is considered, that is, i = 4.

[0064] When i ≥ 4, that is, the number of wheelbases that have not yet been divided exceeds 3, the four wheelbases (Nth to N+3th) are checked to see if they meet the aforementioned four rules. If they do, they are considered to meet the four-axis rule; if not, they are considered not to meet the four-axis rule.

[0065] Satisfy the 5-axle rule: Similar to the 4-axle rule, the 3 wheelbase rules are applied to the 5-axle. That is, if it is a 5-axle vehicle, it must meet the following 5 rules:

[0066] Rule 1: Is the Nth wheelbase value approximately equal to the N+3th wheelbase value, that is, is the absolute value of the difference between the Nth wheelbase value and the N+3th wheelbase value less than 100 mm?

[0067] Rule 2: Is the N+1th wheelbase value approximately equal to the N+2th wheelbase value, that is, is the absolute value of the difference between the N+1th wheelbase value and the N+2th wheelbase value less than 100 mm?

[0068] Rule 3: Is the sum of the wheelbases of N, N+1, N+2, and N+3 greater than 7000 mm?

[0069] Rule 4: Is the N+1th wheelbase value greater than the N+4th wheelbase value?

[0070] Rule 5: Whether the N+4th wheelbase value is greater than the Nth wheelbase value.

[0071] When i ≤ 3, the 5-axis rule analysis is not performed due to insufficient data available for analysis. When i = 4, analysis of rules 1, 2, and 3 can be performed. When i ≥ 5, all five rules are analyzed. If they meet, the 5-axis rule is considered to be met; if not, the 5-axis rule is not met.

[0072] Satisfy the 6-axle rule: Similar to the 4-axle rule, the 3 wheelbase rules are applied to the 6-axle. That is, if it is a 6-axle vehicle, it must meet the following 5 rules:

[0073] Rule 1: Is the Nth wheelbase value approximately equal to the N+4th wheelbase value, that is, is the absolute value of the difference between the Nth wheelbase value and the N+4th wheelbase value less than 100 mm?

[0074] Rule 2: Is the N+1th wheelbase value approximately equal to the N+3th wheelbase value, that is, is the absolute value of the difference between the N+1th wheelbase value and the N+3th wheelbase value less than 100 mm?

[0075] Rule 3: Is the sum of the five wheelbases from Nth to N+4th greater than 7000 mm?

[0076] Rule 4: Is the N+2th wheelbase value greater than the N+5th wheelbase value?

[0077] Rule 5: Whether the N+5th wheelbase value is greater than the Nth wheelbase value.

[0078] When i ≤ 3, the 6-axis rule analysis is not performed due to insufficient data available for analysis. When i = 4, rule 2 analysis can be performed. When i = 5, rules 1, 2, 3, and 4 analysis can be performed. When i ≥ 6, all five rules are analyzed. If the conditions are met, the 6-axis rule is considered to be met; if not, the 6-axis rule is not met.

[0079] Meet the 8-axle rule: Similar to the 4-axle rule, the 3 wheelbase rules are applied to the 8-axle vehicle. That is, if it is an 8-axle vehicle, it must meet the following 6 rules:

[0080] Rule 1: Is the Nth wheelbase value approximately equal to the N+6th wheelbase value, that is, is the absolute value of the difference between the Nth wheelbase value and the N+6th wheelbase value less than 100 mm?

[0081] Rule 2: Is the N+1th wheelbase value approximately equal to the N+5th wheelbase value, that is, is the absolute value of the difference between the N+1th wheelbase value and the N+5th wheelbase value less than 100 mm?

[0082] Rule 3: Is the N+2th wheelbase value approximately equal to the N+4th wheelbase value, that is, is the absolute value of the difference between the N+2th wheelbase value and the N+4th wheelbase value less than 100 mm?

[0083] Rule 4: Is the sum of the seven wheelbases from Nth to N+6th greater than 7000 mm?

[0084] Rule 5: Is the N+3th wheelbase value greater than the N+7th wheelbase value?

[0085] Rule 6: Whether the N+7th wheelbase value is greater than the Nth wheelbase value.

[0086] When i ≤ 4, the 8-axis pattern analysis is not performed due to insufficient data available for analysis. When i = 5, analysis for pattern 3 can be performed. When i = 6, analysis for patterns 2 and 3 can be performed. When i = 7, analysis for patterns 1, 2, 3, and 4 can be performed. When i ≥ 8, analysis for all six patterns is performed. If all the patterns meet, the 8-axis pattern is considered to be met; if not, the 8-axis pattern is not met.

[0087] All rules are not satisfied: that is, all 4-, 5-, 6-, and 8-axis rules are not satisfied.

[0088] N = N + 1: If all the rules are not met, N = N + 1, meaning the next division starts from the N + 1th axis. i = i - 1, meaning the number of undivided axis distances minus 1. This means the Nth axis cannot be used for division and is temporarily set aside. At this point, T will not equal N. The next process begins, and the 4th, 5th, 6th, and 8th axis regularity analysis is repeated.

[0089] Divide the wheelbases from T to N into a single car section: If the pattern analysis shows that the requirements are met, then the current wheelbase values ​​can be determined as the wheelbase values ​​and the number of axles in a car section, and the axles from N to N + the number of axles of the car section - 1 can be divided into a single car section. For example, if the 4-axle pattern is met, then the N, N+1, N+2, and N+3 are the four wheelbase values ​​of a 4-axle car section, and the number of axles in the car section is 4.

[0090] Divide the Tth to Nth axles into one car section: If T=N, that is, there are no wheelbase values ​​that cannot be divided, then do not perform this step. If T>N, that is, there are wheelbase values ​​that cannot be divided, then divide the wheelbase values ​​that have not been divided into one car section, that is, divide the wheelbase values ​​from the Tth to the N-1th into one car section.

[0091] N = N + number of axles: Because the previous wheelbase values ​​have been divided, the next division starts from the N + number of axles. i = i - number of axles, that is, the number of wheelbases that have not been divided is reduced by the number of axles; T = N, which means that the first N + number of axles - 1 wheelbases have been divided, and there are no undivided wheelbase values.

[0092] An example of a train wheelbase segmentation method is given below in conjunction with FIG2 .

[0093] For example, when a train, as shown in Figure 2, passes through a set of magnetic sensors in the system of the present disclosure, 13 wheelbase information will be generated sequentially for each of the 14 wheels. Suppose the wheelbase information sequence measured by the data acquisition card corresponding to the magnetic sensors is 1802, 1803, 8378, 1796, 1792, 4233, 1762, 7538, 1753, 2895, 1756, 7530, and 1769, in millimeters. Starting with the first wheelbase, "1802," when the third axle is accumulated, the four-axis pattern is checked. Clearly, the first three wheelbase values ​​do not meet the four-axis pattern for the first three axes. This continues in this order. When the six axles have been accumulated, the six-axis pattern is checked and found to meet the requirements: 1802 ≈ 1792, 1803 ≈ 1796, 1802 + 1803 + 8378 + 1796 + 1792 > 7000, and 8378 > 4233 > 1802. Therefore, 1802, 1803, 8378, 1796, 1792, and 4233 can be grouped into one car. Then, starting with the seventh wheelbase, "1762," it's clear that axles 7 through 10 meet the four-axis pattern and can therefore also be grouped into one car. The remaining axles are then grouped into the final car.

[0094] When a train passes, various reasons, such as train vibration, may cause the signal of one axle to be lost. For example, when a three-car train, as shown in Figure 2, passes through a set of magnetic sensors, the signal of the fifth wheel is lost. Then, a total of 14 wheels will generate 12 wheelbase information in sequence. If the wheelbase information sequence measured by the data acquisition card corresponding to the magnetic sensors is 1802, 1803, 8378, 3588, 4233, 1762, 7538, 1753, 2895, 1756, 7530, 1769 (a total of 12 wheelbases, the original 4th and 5th wheelbase values ​​are combined into one wheelbase value, in millimeters), the process starts with the first wheelbase, "1802." When checking the regularity of the 4th, 5th, 6th, and 8th axles, it is found that none of them meet the requirements. So we put the first wheelbase value aside and re-analyze the rules starting from the second wheelbase value "1803'". We found that it did not meet the requirements again. Then we analyzed from the third wheelbase value... and so on. When we started from the sixth wheelbase "1762", we found that the four axles from the 6th to the 9th met the 4-axle rule, so they could also be divided into one car. The wheelbases from the 1st to the 5th that were not divided before were divided into one car, and the rest were divided into the last car.

[0095] After segmenting the wheelbases, the processor compares the wheelbase of each car with the wheelbase information of corresponding known open-top cars pre-stored in the processor to determine whether the car is an open-top car. For example, by analyzing the wheelbase data of wheels on currently operating domestic trains, the following rules were found: when the first wheelbase of a car is less than 1500 mm, the car is a freight car; when both the first and third wheelbases of a car are less than 2000 mm, the car is a freight car; if the first wheelbase is greater than or equal to 2000 mm and the third wheelbase is greater than 2000 mm, the car is a locomotive; if the first wheelbase is greater than or equal to 2000 mm and the second wheelbase is less than 8000 mm, the car is a locomotive; if the first wheelbase is greater than or equal to 2000 mm and the second wheelbase is less than 8000 mm, the car is a locomotive; and if the first wheelbase is greater than or equal to 2000 mm and the second wheelbase is greater than or equal to 8000 mm, the car is a passenger car. After determining a carriage as a truck, the system can identify open boxes, double boxes, containers, bulk cargo, flatbed cars and other scenes based on the point cloud data obtained by the lidar.

[0096] Furthermore, the processor can also calculate the car hook signal. The hook is a connecting component between two adjacent cars, and the hook signal indicates the switching time between train cars.

[0097] FIG3 is a schematic diagram of hook signal calculation according to some exemplary embodiments of the present disclosure.

[0098] As shown in Figure 3, the dotted line represents the scanning plane of the second laser radar L2, located on the side of the train track. The arrow indicates the train's direction of travel, and Q represents the center of the train's hook. When the current car acquires the first set of wheelbases, L, the distance D between the last axle of the previous car and the first axle of the current car is known. Because the installation distance G between the magnetic sensor and the second laser radar L2 is fixed, the current distance between the hook and the second laser radar L2 can be calculated as G - (D / 2) - L. Within this distance, the train can be assumed to be moving at a constant speed. Combined with the current speed, the time t required for the hook to reach the scanning plane of the second laser radar L2 can be calculated, and the hook signal is emitted after this time. The distance G between the magnetic sensor and the second laser radar L2 ensures that when the current car acquires the first set of wheelbases, the hook between the current car and the previous car has not yet passed through the scanning plane of the second laser radar L2.

[0099] As shown in Figure 1, in some embodiments, the laser radar group includes a first laser radar L1 and a second laser radar L2. The first laser radar L1 is located above the train track T, at a height higher than the height of the train, and is used to scan the top of the train to obtain the top outline information of the train; the second laser radar L2 is located on the side of the train track T, at a height higher than the bottom surface of the train track T and lower than the height of the first laser radar L1, and is used to scan the side of the train to obtain the side outline information of the train.

[0100] Specifically, the first laser radar L1 and the second laser radar L2 can be single-line lasers or multi-line laser radars. Single-line laser sensors operate on a similar principle to multi-line laser radars, generating laser lines to measure the surface of an object. By translating a certain step size within the same point cloud coordinate system, point cloud data containing rich contour information of the train surface is generated.

[0101] In some embodiments, the laser radar group further includes a gantry D. The gantry D is mounted on the train track T and includes a crossbeam and two brackets. The crossbeam spans the train track T, and the two brackets are respectively provided on both sides of the train track T to support the crossbeam. The first single-line laser radar L1 is provided in the middle of the crossbeam; and the second single-line laser radar L2 is provided on at least one of the two brackets.

[0102] FIG4 is a scanning principle diagram of a lidar group according to some exemplary embodiments of the present disclosure.

[0103] As shown in Figure 4, in some embodiments, when the magnetic steel sensor group detects the arrival of a train, the first laser radar L1 and the second laser radar L2 are started to scan the train. When the acquisition of one carriage is completed, the point cloud data of the carriage is stored, and the point cloud acquisition of the next carriage is started at the same time.

[0104] In some embodiments, the scanning surfaces of the first laser radar L1 and the second laser radar L2 are both perpendicular to the direction of travel of the train, and the scanning surfaces of the first laser radar L1 and the second laser radar L2 are located in the same plane, ensuring that the laser radar group can obtain complete contour information of the top and sides of the train during the passage of the train.

[0105] During the passage of the train, the first laser radar L1 and the second laser radar L2 continuously scan the train at a fixed frequency f to collect the train's contour information. In some embodiments, the processor is further configured to calculate the translation step length of two adjacent frames of contour information based on the train's speed and the scanning frequency of the laser radar group. The formula for calculating the translation step length of two adjacent frames of contour information is Δs = v / f, where v represents the speed of the train and Δs represents the translation step length. During the movement of the train, each frame of data collected by the first laser radar L1 and the second laser radar L2 is sequentially translated by an integer multiple of Δs. Thus, the processor can sequentially translate the multiple frames of contour information continuously scanned and collected by the laser radar group according to the translation step length Δs, map the collected contour information to the same point cloud coordinate system, and obtain the overall contour information of the train.

[0106] As shown in Figures 5A and 5B, Figure 5A shows the top contour information of the train according to some exemplary embodiments of the present disclosure, and Figure 5B shows the side contour information of the train according to some exemplary embodiments of the present disclosure. The hook signal obtained by the car identification device indicates the switching timing between the train cars. The processor can divide the overall contour information of the train into car contour information based on the hook signal. The overall contour information of the train includes the top contour information of the train as shown in Figure 5A and the side contour information of the train as shown in Figure 5B. After segmentation based on the hook information, the top contour information and the side contour information of each car are obtained. The car contour information is point cloud data, which can carry more and more fine-grained information than images.

[0107] FIG6A is a workflow of a processor processing train profile information according to some exemplary embodiments of the present disclosure.

[0108] As shown in Figure 6A, in some embodiments, the processor connects to a car identification device and a laser radar system to obtain the car type and hook signal from the car identification device and train profile information from the laser radar system. Based on the hook signal, the processor segments the overall train profile information into multiple car profiles, then processes each car profile separately to obtain information about the load in each car.

[0109] FIG6B is a schematic diagram showing the composition of vehicle box contour information according to some exemplary embodiments of the present disclosure.

[0110] As shown in Figure 6B, in some embodiments, the processor stores the car profile information for each car separately, specifically including the train ID, car number, car model, car top profile information, and car side profile information. The train ID and car number can be generated by the processor to identify the corresponding train and car, facilitating the retrieval of the corresponding car profile information when needed. The car model can be generated based on the car type identified by the car identification device to facilitate differentiation between car types. The car top profile information and car side profile information are both point cloud data, which are subsequently processed by the processor.

[0111] The processor can sequentially store the compartment profile information of each compartment in a compartment data queue. When the compartment data queue is not empty, the processor sequentially retrieves the compartment profile information of each compartment from the head of the compartment data queue for processing. The method provided in this embodiment of the disclosure is used to calculate the loading volume of an open-top container. If and only if the compartment type is an open-top container, the processor further processes the compartment profile information to determine the loading status of the compartment.

[0112] FIG. 7A schematically illustrates the height difference between a train car and a load according to some exemplary embodiments of the present disclosure.

[0113] As shown in Figure 7A, the dotted box represents the train car, and the dotted curve in the middle represents the top edge of the load. There is an obvious height difference between the edge of the train car and the load. After obtaining the top contour information of a single car through the first laser radar L1 located in the top area of ​​the train, the processor determines the position of the box edge in four directions based on the height change trend of the point cloud in the car top contour information, and filters the box point cloud in the car top contour information to obtain the top contour information of the load.

[0114] Figure 7B schematically illustrates a diagram of the top contour information of a vehicle box according to some exemplary embodiments of the present disclosure. As shown in Figure 7B , the point cloud of the box area in the figure appears as a regular rectangle. Figure 7C schematically illustrates the top contour information of the vehicle box after filtering out the box point cloud according to some exemplary embodiments of the present disclosure. As shown in Figure 7C , after filtering, only the point cloud of the center area is retained.

[0115] Furthermore, the processor is also configured to obtain the bottom height of the car box based on the side profile information of the car box, and then calculate the height of each point in the car box top profile information relative to the bottom of the car box in combination with the car box top profile information after filtering out the box body point cloud. Since point clouds are discrete data, the volume is relatively difficult to obtain, and therefore, defining the boundaries of objects and processing the continuity of contours are relatively difficult. In some embodiments of the present disclosure, the processor quantizes the three-dimensional point cloud data using a rasterization method, converts it into two-dimensional elevation data, and then calculates the volume of the load represented by the point cloud data in each grid in combination with the grid area.

[0116] In some embodiments, the processor rasterizes the top contour information of the carriage in the horizontal plane of the point cloud coordinate system. After the grid is demarcated, the average height of the points in each grid relative to the bottom of the carriage is calculated. The average height of the point cloud in the grid is used as the elevation information stored in the grid, and the point cloud data is quantified. Then, based on the area s of the grid and the average height h of the points in the grid, the processor calculates the unit volume of the load in each grid, and calculates the sum of the unit volumes of the load in all grids to obtain the total volume of the load in the carriage. Based on the total volume of the load, the processor can also calculate data such as the space utilization rate of the carriage, the space utilization rate of the entire train, and the average space utilization rate of the train carriages.

[0117] Furthermore, in some embodiments, the processor is further configured to detect foreign objects contained within the load. For cargo in a pile, such as coal, the generated point cloud has a relatively smooth outline, with no significant elevation changes within local areas. Therefore, the elevation differences of the rasterized point cloud can be used to determine whether there are other foreign objects placed within the pile.

[0118] Specifically, in some embodiments, after the processor rasterizes the contour information of the top of the box, it can calculate the elevation difference between the maximum height and the minimum height of the points representing the contour of the top of the box in each grid, and determine whether the elevation difference exceeds a first preset threshold. When the elevation difference exceeds the first preset threshold range, it is determined that there is an abnormal height change in the load, and then it is determined that the load here carries foreign matter.

[0119] FIG8 schematically shows another schematic diagram of determining whether a load contains foreign matter according to an embodiment of the present disclosure.

[0120] As shown in Figure 8, in other embodiments, after the processor rasterizes the top contour information of the box, it calculates the average height of the points in all grids adjacent to the grid (m, n) to obtain the average height h' of the adjacent grid point cloud; compares the average height h of the points in the grid (m, n) with the average height h' of the adjacent grid point cloud; when the average height difference between the average height h of the points in the grid and the average height h' of the adjacent grid point cloud is greater than a second preset threshold, it is determined that there are foreign objects in the load.

[0121] Both of the above embodiments obtain the point cloud mutation information of the carriage load by rasterizing the elevation difference of the point cloud and perform foreign object detection, thereby enriching the functions of the train open box measurement and inspection system provided by the embodiment of the present disclosure.

[0122] In some embodiments, the processor is further configured to convert the carriage outline information into an image, where the point cloud within the image can represent its elevation information using different colors. By retrieving the image, the user can intuitively understand the distribution of the load within the carriage and extract useful information. As shown in Figure 1, the train open-carriage measurement and inspection system provided by the embodiments of the present disclosure may also include a memory configured to store the image.

[0123] FIG9 schematically shows a flow chart of a train open box measurement and inspection method according to the present disclosure.

[0124] As shown in FIG9 , a train open box measurement and inspection method provided by the present disclosure includes operations S910 to S940 .

[0125] In operation S910 , a laser is generated to scan a train and collect contour information of the train, where the contour information is point cloud data.

[0126] In some embodiments, when the sensor measures the arrival time of the train, the first laser radar L1 and the second laser radar L2 are started to scan the train, and the contour information of the top of the train is collected by the first laser radar L1, and the contour information of the side of the train is collected by the second laser radar L2.

[0127] In operation S920, the wheelbase of the train is measured, the car type of each car in the train is determined based on the wheelbase and the point cloud data, and a hook signal between the cars of the train is generated.

[0128] In some embodiments, the wheelbase can be calculated by measuring the arrival time of the train wheels using magnetic sensors. When the magnetic sensors detect the train's arrival time, the processor calculates the train's speed based on the time difference between the arrival times detected by multiple magnetic sensors and the installation spacing of the multiple magnetic sensors. The train's wheelbase is also calculated based on the time difference and speed between two adjacent wheels sensed by the same magnetic sensor in the magnetic sensor group. Based on the wheelbase and point cloud data, the processor can determine the car type and the hook signal between adjacent cars.

[0129] In some embodiments, a second laser radar L2 located on one side of the train track T can also detect gaps between cars. Based on the point cloud data collected by the second laser radar L2, the number of point clouds is significantly reduced when the second laser radar L2 detects a gap between cars. When the second laser radar L2 detects a gap between cars, the processor generates a hook signal between the currently adjacent cars.

[0130] In operation S930, the train profile information is segmented into car profile information in units of cars according to the hook signal.

[0131] In some embodiments, the hook signal indicates the switching timing between two adjacent carriages on the train. Based on the hook information, the overall profile information of the train is divided into multiple carriage profile information, which facilitates the subsequent statistics of the loading status in units of carriages.

[0132] In operation S940, when the box type is an open box, the volume of the load in the box is calculated based on the box contour information, and foreign objects in the load are identified.

[0133] In some embodiments, when the car box type is not an open box, its loading condition cannot be determined based on the contour information of the car box; when the car box type is an open box, the bottom height of the car box is obtained based on the side contour information of the car box, and then combined with the top contour information of the car box after filtering out the box point cloud, the height of each point in the top contour information of the car box relative to the bottom of the car box is calculated; the three-dimensional point cloud data is quantified by rasterization and converted into two-dimensional elevation data, and then combined with the grid area, the volume of the load represented by the point cloud data in each grid is calculated.

[0134] In some embodiments, after the top contour information of the box is rasterized, the elevation difference between the maximum height and the minimum height of the points representing the top contour of the box in each grid can be calculated to determine whether the elevation difference exceeds a first preset threshold. When the elevation difference exceeds the first preset threshold range, it is determined that there is an abnormal height change in the load, and then it is determined that the load here carries foreign matter.

[0135] In some embodiments, the average height of the points in all grids adjacent to the grid (m, n) can also be calculated to obtain the average height h' of the adjacent grid point cloud; the average height h of the points in the grid (m, n) is compared with the average height h' of the adjacent grid point cloud; when the average height difference between the average height h of the points in the grid and the average height h' of the adjacent grid point cloud is greater than a second preset threshold, it is determined that there are foreign objects in the load.

[0136] In the train open box measurement and inspection system and method according to the embodiment of the present disclosure, the type of train car is determined by measuring the wheelbase and the collected point cloud data, the hook signal between the cars is generated, and the point cloud data reflecting the train contour is binned to achieve separate processing of the point cloud data of each car. Based on the point cloud data of each car, the volume of the load in each car can be identified, and foreign objects in the load can be identified.

[0137] The flowcharts and block diagrams in the accompanying drawings illustrate the possible implementation architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each box in the flowchart or block diagram can represent a module, program segment, or a part of code, and the above-mentioned module, program segment, or a part of code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flowchart, and the combination of boxes in the block diagram or flowchart, can be implemented with a dedicated hardware-based system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0138] The embodiments of the present disclosure are described above. However, these embodiments are for illustrative purposes only and are not intended to limit the scope of the present disclosure. Although each embodiment has been described separately above, this does not mean that the measures in each embodiment cannot be used in combination to advantage. The scope of the present disclosure is defined by the appended claims and their equivalents. Without departing from the scope of the present disclosure, those skilled in the art may make various substitutions and modifications, which should all fall within the scope of the present disclosure.

Claims

1. A train open-box measurement and inspection system, wherein, The system includes: A lidar group for generating laser to scan the train and collecting the contour information of the train, where the contour information is point cloud data; A carriage identification device for measuring the wheelbase of the train, determining the carriage type of each carriage in the train according to the wheelbase and the point cloud data, and generating a coupler signal between the carriages of the train; and A processor communicatively connected to the carriage identification device and the lidar group, the processor being configured to: segment the contour information of the train into carriage contour information in units of carriages according to the coupler signal; and when the carriage type is an open carriage, calculate the volume of the load in the carriage according to the carriage contour information and identify foreign objects in the load.

2. The system according to claim 1, wherein The carriage identification device includes: A magnetic sensor group including a plurality of magnetic sensors, the plurality of magnetic sensors being arranged on one side of the train track at a predetermined installation spacing for respectively collecting the arrival times of the wheels of the train; The processor is further configured to: calculate the vehicle speed of the train according to the time difference of the arrival times collected by the plurality of magnetic sensors in the magnetic sensor group and the installation spacing of the plurality of magnetic sensors; calculate the wheelbase of the train according to the time difference between the same magnetic sensor in the magnetic sensor group sensing adjacent wheels and the vehicle speed; and determine the carriage type and the coupler signal according to the wheelbase and the point cloud data.

3. The system according to claim 2, wherein The magnetic sensor group includes: An up-track magnetic sensor group arranged in the up-track direction of the train track for sensing the arrival time of a train coming from the up-track direction; and / or, A down-track magnetic sensor group arranged in the down-track direction of the train track for sensing the arrival time of a train coming from the down-track direction.

4. The system according to claim 1, wherein, The lidar group includes: A first lidar arranged above the train track at a height higher than the height of the train for scanning the top of the train to obtain the top contour information of the train; and A second lidar arranged on one side of the train track at a height higher than the bottom surface where the train track is located and lower than the height of the first lidar for scanning the side of the train to obtain the side contour information of the train.

5. The system according to claim 4, wherein, The scanning planes of the first lidar and the second lidar are both perpendicular to the traveling direction of the train.

6. The system according to claim 4, wherein, The lidar group further includes: A gantry erected on the train track, including a cross beam and two supports, the cross beam spanning the train track, and the two supports being respectively arranged on both sides of the train track to support the cross beam; The first single-line lidar is arranged in the middle of the cross beam; and The second single-line lidar is arranged on at least one of the two supports.

7. The system according to claim 1, wherein The processor is further configured to: Calculate the translation step of two adjacent frames of the contour information based on the vehicle speed of the train and the scanning frequency of the lidar group; Map multiple frames of contour information continuously scanned and collected by the lidar group to the same point cloud coordinate system according to the translation step to obtain the overall contour information of the train; and Based on the coupler signals, the overall contour information of the train is segmented into car body contour information in units of car bodies.

8. The system according to claim 1, wherein The car body contour information includes car body top contour information and car body side contour information; and The processor is further configured to: filter the box point cloud in the car body top contour information according to the height change trend of the point cloud in the car body top contour information to obtain the top contour information of the load; obtain the bottom height of the car body according to the car body side contour information, and calculate the height of each point in the car body top contour information relative to the bottom of the car body; rasterize the car body top contour information, and calculate the average height of the points in each grid of the car body top contour information relative to the bottom of the car body; calculate the unit volume of the load in each grid according to the area of the grid and the average height of the points in the grid, and calculate the sum of the unit volumes of the load in all the grids to obtain the total volume of the load in the car body.

9. The system according to claim 8, wherein The processor is further configured to: calculate the elevation difference between the maximum height and the minimum height of the points in each grid; judge whether the elevation difference exceeds a first preset threshold; and when the elevation difference exceeds the first preset threshold, determine that there is a foreign object in the load.

10. The system according to claim 8, wherein, The processor is further configured to: calculate the average height of the points in all the grids adjacent to the grid to obtain the average height of the adjacent grid point cloud; compare the average height of the points in the grid with the average height of the adjacent grid point cloud; and when the difference between the average height of the points in the grid and the average height of the adjacent grid point cloud is greater than a second preset threshold, determine that there is a foreign object in the load.

11. The system according to claim 1, wherein The processor is further configured to convert the car body contour information into an image; and The system further includes a memory configured to store the image.

12. A method for measuring and inspecting an open box of a train, wherein, The method includes: generating a laser to scan the train to collect the contour information of the train, where the contour information is point cloud data; measuring the wheelbase of the train, determining the car body types of the car bodies in the train according to the wheelbase and the point cloud data, and generating coupler signals between the car bodies of the train; segmenting the contour information of the train into car body contour information in units of car bodies according to the coupler signals; and when the car body type is an open car body, calculating the volume of the load in the car body according to the car body contour information and identifying foreign objects in the load.

Images