Railway vehicle bottom clearance detection method and system

Abstract

The present invention relates to a method and system for detecting the undercarriage clearance of railway vehicles. The method comprises the following steps: (1) monitoring and obtaining a two-dimensional matrix S0[m,n] of the undercarriage clearance of the entire train; (2) obtaining the number of train sections K and the number of data points N in the length and width directions of each section through a train segmentation algorithm; k and M k , the two-dimensional matrix S of the bottom clearance of each car k (3) Through the point cloud construction algorithm, the bottom clearance two-dimensional matrix S of each car is k [m, n] corresponds to the actual position of the bottom of the vehicle, and S k The actual coordinate P corresponding to each data point in [m, n] k，m，n [x, y]; (4) Using a clearance extraction algorithm, extract the lowest T clearance values ​​and their locations under each car. The undercarriage clearance detection method and system provided by the present invention can be applied to all railway vehicles, has wider applicability, can significantly improve the efficiency and quality of marshaling yard operations, and promote railway freight safety.

Description

Technical Field

[0001] The present invention relates to the technical field of rail transit safety detection, and in particular to a method and system for detecting undercarriage clearance of railway vehicles used in shunting operations in railway marshalling yards. Background Art

[0002] In order to ensure that the gap between the bottom of railway vehicles and the rail surface is sufficient during operation and marshaling, it is necessary to measure the bottom clearance of railway vehicles. At present, the bottom clearance detection algorithm for railway freight cars is to use a point laser to scan the bottom of the train to form one-dimensional bottom clearance data and analyze it. This method can identify concave-bottom vehicles (a type of vehicle among railway freight cars) and calculate the lowest clearance at the scanning line, which basically solves the bottom clearance detection problem of concave-bottom vehicles and can improve the efficiency of dismantling such concave-bottom vehicles. However, this method has certain limitations. The defects are: (1) The one-dimensional clearance data cannot fully display the clearance height of the two-dimensional plane of the bottom of the vehicle, that is, the lowest bottom clearance of the vehicle is not in the clearance data, so the subsequent recognition algorithm cannot obtain the actual lowest clearance; (2) This method mainly recognizes concave-bottom vehicles, and has limited recognition effect on other freight cars such as passenger cars, locomotives, and open car boxes; (3) This method only obtains one lowest bottom clearance value, but in fact there may be other local minimum heights, which may also become accident hazards. Therefore, although the current underbody clearance extraction method plays a certain role in identifying the underbody clearance of vehicles with concave bottoms, the acquired data is not comprehensive enough, and there is still a certain gap from fully detecting the underbody clearance of all types of trucks. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to address the above-mentioned shortcomings and provide a method and system for detecting the underbody clearance of railway vehicles that can construct three-dimensional point cloud data of the underbody.

[0004] The present invention is achieved through the following technical solutions:

[0005] A method for detecting underbody clearance of a railway vehicle comprises the following steps:

[0006] (1) Monitor and obtain the two-dimensional matrix S0[m,n] of the vehicle bottom clearance of the entire train, whose size is M*N (m=1,2,...,M,n=1,2,...,N), where M and N are the number of data points in the width direction and length direction of the entire train respectively;

[0007] (2) Combined with the two-dimensional matrix of the vehicle bottom clearance of the entire train, the number of train sections K and the number of data points N in the length and width directions of each section are obtained through the vehicle division algorithm. k and M k , the two-dimensional matrix S of the bottom clearance of each car k ;

[0008] (3) Through the point cloud construction algorithm, the bottom clearance two-dimensional matrix S of each car is k [m, n] corresponds to the actual position of the bottom of the vehicle, and S is obtained k The actual coordinate P corresponding to each data point in [m, n] in the car k，m，n [x, y];

[0009] (4) For each car's bottom clearance two-dimensional matrix S k ,Through the clearance extraction algorithm, the lowest T clearance values ​​and their positions under each car are extracted, where T is a preset value.

[0010] Furthermore, in the vehicle bottom clearance detection method, in step (2), the vehicle classification algorithm is:

[0011] ①Convolve the two-dimensional matrix S0[m,n] of the vehicle bottom clearance of the entire train with the filter F1[m,n] to obtain the filtered waveform S F [m, n], that is:

[0012]

[0013] ②From S F The calculation starts from the first row / column of data. If the average value of the row / column data is greater than the preset value H and the standard deviation is less than the preset value ∈ v , then delete all rows / columns from the first row / column to the row / column in S0[m,n], and finally obtain a two-dimensional matrix S1[m,n] with a size of M1*N1;

[0014] ③The vector Multiplying S1 on the left gives β, which is:

[0015] β=α*S1

[0016] Select all the values ​​in β that are less than the preset value ∈ β The subscripts of form the sequence I[k];

[0017] ④Traverse the sequence I[k] and delete all I[k′]-I[k]≤∈ i The elements I[k′] of the sequence I′[k] (k=1, 2, ..., K); where ∈ i is a preset threshold, and k′>k;

[0018] ⑤At this time, the sequence I′[k] (k=1, 2, ..., K) is the subscript for dividing each section of data, that is, the two-dimensional matrix S of the bottom clearance of each car k [m, n] is the part from column I′[k] to column I′[k+1] in the two-dimensional matrix S1, and the number of data points in the length direction of each car is N. k=I′[k+1]-I[k], the number of data points in the width direction of each car M k =M.

[0019] Furthermore, in the vehicle bottom clearance detection method, in step (3), the point cloud construction algorithm is:

[0020] ① Determine the coordinate origin: construct a coordinate system with the upper left corner of each car as the origin;

[0021] ②Determine the y-direction coordinate of the point cloud: S k The vertical coordinate of the i-th row of the matrix

[0022] ③Determine the x-direction coordinate of the point cloud: S k The vertical coordinate of the jth column of the matrix

[0023] ④ Then S k The actual coordinates corresponding to each data point in [m, n] are the coordinates of the point in the mth row and nth column of the kth car from the origin of this car. The height of this point from the rail surface is S k [m, n];

[0024] Among them, L k 、W k is the length and width of the car section.

[0025] Furthermore, in the vehicle bottom clearance detection method, in step (4), the clearance extraction algorithm is:

[0026] ①If T is greater than S k Number of data points M*N k , then M*N k Give T;

[0027] ②S k The elements in are laid out in order into array s k [j], where s k The first (m k -1)*n k +1 to m k *n k The element is S k mth k Row element, j = 1, 2, ..., M*N k ;

[0028] ③Array s k [j] Arrange in descending order to obtain the descending array s′ k [j] and the corresponding subscript conversion array r[j];

[0029] ④Select s′k The first T elements in [j] are used as the T clearance values ​​d of the kth car. k，t ;

[0030] ⑤Select the first T elements in r[j] as the T lowest values ​​of the kth car in the array s′ k The subscript r k，t ;

[0031] ⑥Calculate the S of each lowest point in this section of the car k Coordinates (m k，t , n k，t ):

[0032]

[0033] n k，t =r k，t -m k，t ;

[0034] ⑦ By (m k，t , n k，t ) Calculate the coordinates of each lowest point in the original point cloud matrix (M k，t , N k，t ), that is, d k，t Corresponding position coordinates:

[0035] M k，t =m k，t ,

[0036]

[0037] Furthermore, the vehicle bottom clearance detection method, the step (4) further comprises: k The actual coordinate P corresponding to each data point in [m, n] k，m，n [x, y], get d k，t Corresponding to the position coordinates in this section of the car.

[0038] A railway vehicle undercarriage clearance detection system, the system comprising:

[0039] The monitoring module is used to monitor and obtain the two-dimensional matrix S0[m,n] of the vehicle bottom clearance of the entire train in real time. Its size is M*N (m=1,2,...,M,n=1,2,...,N), where M and N are the number of data points in the width direction and length direction of the entire train respectively;

[0040] The car division algorithm module is used to combine the two-dimensional matrix of the vehicle bottom clearance of the entire train, and obtain the number of train sections K and the number of data points N in the length and width directions of each section through the car division algorithm. k and M k , the two-dimensional matrix S of the bottom clearance of each cark ;

[0041] Point cloud construction algorithm module is used to construct the bottom clearance two-dimensional matrix S of each car through the point cloud construction algorithm. k [m, n] corresponds to the actual position of the bottom of the vehicle, and S is obtained k The actual coordinate P corresponding to each data point in [m, n] in the car k，m，n [x, y]; and

[0042] Clearance extraction algorithm module, used for the two-dimensional matrix S of the clearance under each car k ,Through the clearance extraction algorithm, the lowest T clearance values ​​and their positions under each car are extracted, where T is a preset value.

[0043] Furthermore, in the vehicle bottom clearance detection system, the vehicle separation algorithm module performs the following vehicle separation algorithm operations:

[0044] ①Convolve the two-dimensional matrix S0[m,n] of the vehicle bottom clearance of the entire train with the filter F1[m,n] to obtain the filtered waveform S F [m, n], that is:

[0045]

[0046] ②From S F The calculation starts from the first row / column of data. If the average value of the row / column data is greater than the preset value H and the standard deviation is less than the preset value ∈ v , then delete all rows / columns from the first row / column to the row / column in S0[m,n], and finally obtain a two-dimensional matrix S1[m,n] with a size of M1*N1;

[0047] ③The vector Multiplying S1 on the left gives β, which is:

[0048] β=α*S1

[0049] Select all the values ​​in β that are less than the preset value ∈ β The subscripts of form the sequence I[k];

[0050] ④Traverse the sequence I[k] and delete all I[k′]-I[k]≤∈ i The elements I[k′] of the sequence I′[k] (k=1, 2, ..., K); where ∈ i is a preset threshold, and k′>k;

[0051] ⑤At this time, the sequence I′[k] (k=1, 2, ..., K) is the subscript for dividing each section of data, where K≤K0, that is, the two-dimensional matrix S of the bottom clearance of each section of the car k[m, n] is the part from column I′[k] to column I′[k+1] in the two-dimensional matrix S1, and the number of data points in the length direction of each car is N. k =I′[k+1]-I[k], the number of data points in the width direction of each car M k =M.

[0052] Furthermore, in the vehicle bottom clearance detection system, the point cloud construction algorithm module performs the following point cloud construction algorithm operations:

[0053] ① Determine the coordinate origin: construct a coordinate system with the upper left corner of each car as the origin;

[0054] ②Determine the y-direction coordinate of the point cloud: S k The vertical coordinate of the i-th row of the matrix

[0055] ③Determine the x-direction coordinate of the point cloud: S k The vertical coordinate of the jth column of the matrix

[0056] ④ Then S k The actual coordinates corresponding to each data point in [m, n] are the coordinates of the point in the mth row and nth column of the kth car from the origin of this car. The height of this point from the rail surface is S k [m, n];

[0057] Among them, L k 、W k is the length and width of the car section.

[0058] Furthermore, in the vehicle bottom clearance detection system, the clearance extraction algorithm module performs the following clearance extraction algorithm operations:

[0059] ①If T is greater than S k Number of data points M*N k , then M*N k Give T;

[0060] ②S k The elements in are laid out in order into array s k [j], where s k The first (m k -1)*n k +1 to m k *n k The element is S k mth k Row element, j = 1, 2, ..., M*N k ;

[0061] ③Array s k [j] Arrange in descending order to obtain the descending array s′k [j] and the corresponding subscript conversion array r[j];

[0062] ④Select s′ k The first T elements in [j] are used as the T clearance values ​​d of the kth car. k，t ;

[0063] ⑤Select the first T elements in r[j] as the T lowest values ​​of the kth car in the array s′ k The subscript r k，t ;

[0064] ⑥Calculate the S of each lowest point in this section of the car k Coordinates (m k，t , n k，t ):

[0065]

[0066] n k，t =r k，t -m k，t ;

[0067] ⑦ By (m k，t , n k，t ) Calculate the coordinates of each lowest point in the original point cloud matrix (M k，t , N k，t ), that is, d k，t Corresponding position coordinates:

[0068] M k，t =m k，t ,

[0069]

[0070] Furthermore, the vehicle bottom clearance detection system, the clearance extraction algorithm also includes: k The actual coordinate P corresponding to each data point in [m, n] k，m，n [x, y], get d k，t Corresponding to the position coordinates in this section of the car.

[0071] The advantages and effects of the present invention are:

[0072] 1. The vehicle bottom clearance detection method and system provided by the present invention can be applied to all railway vehicles including but not limited to locomotives, open wagons, covered wagons, tank cars, flat cars, grain cars, insulated cars, ore cars, cement tank cars, JSQ6 type cars, D type cars, special vehicles, etc. It has wider applicability, can significantly improve the operating efficiency and quality of marshalling yards, and promote railway freight safety.

[0073] 2. The vehicle underbody clearance detection method provided by the present invention utilizes the height information of the two-dimensional plane under the entire train to directly separate the trains and perform underbody clearance detection. At the same time, it constructs three-dimensional point cloud data, simplifies the process (no vehicle model and other information are required), and provides users (marshaling yards) with more comprehensive and reliable underbody clearance data. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] Figure 1 A schematic diagram of point cloud construction in the vehicle bottom clearance detection method provided by the present invention is shown. DETAILED DESCRIPTION

[0075] In order to make the purpose, technical solutions and advantages of the implementation of the present invention clearer, the technical solutions in the embodiments of the present invention are described in more detail below in conjunction with the drawings in the embodiments of the present invention. The described embodiments are part of the embodiments of the present invention, rather than all the embodiments. The embodiments described below with reference to the drawings are exemplary and are intended to be used to explain the present invention, and should not be understood as limiting the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making creative work are within the scope of protection of the present invention. The embodiments of the present invention are described in detail below in conjunction with the drawings:

[0076] Figure 1 A schematic diagram of point cloud construction in the vehicle bottom clearance detection method provided by the present invention is shown. The method includes the following steps:

[0077] (1) Monitor and obtain the two-dimensional matrix S0[m,n] of the vehicle bottom clearance of the entire train, whose size is M*N (m=1,2,...,M,n=1,2,...,N). Among them, M and N are the number of data points in the width direction and length direction of the entire train, respectively. Specifically, the number of data points is determined according to the amount of data collected by the actual sensor. The number of data points in the length direction of the entire train N = the time the entire train passes the sensor * the sensor data collection frequency (for example, it takes 180 seconds for the train to pass, and the sensor collection frequency is 1000 data points per second, then N=180000). The number of data points M in the width direction of the entire train is a preset parameter of the sensor, which can be set to 100 to 2000 according to actual needs.

[0078] (2) According to the train division algorithm, the number of train sections K and the number of data points N in the length and width directions of each section are obtained. k and M k , the two-dimensional matrix S of the bottom clearance of each car k .

[0079] Specifically, the car division algorithm is:

[0080] ①Convolve the two-dimensional matrix S0[m,n] of the vehicle bottom clearance of the entire train with the filter F1[m,n] to obtain the filtered waveform S F [m, n], that is:

[0081]

[0082] ②From S F The calculation starts from the first row / column of data. If the average value of the row / column data is greater than the preset value H and the standard deviation is less than the preset value ∈ v , then delete all rows / columns from the first row / column to the row / column in S0[m,n], and finally get a two-dimensional matrix S1[m,n] with a size of M1*N1. v It is used to remove redundant data points outside the boundaries of the vehicle width and length directions, and only retain the valid vehicle bottom height data. The preset value H can be taken as 600mm~800mm according to the on-site sensor calibration. The preset value ∈ v It can be set to 0.5-1mm according to the on-site sensor calibration.

[0083] ③The vector Multiplying S1 on the left gives β, which is:

[0084] β=α*S1,

[0085] Select all the values ​​in β that are less than the preset value ∈ β The subscripts form the sequence I[k].

[0086] Among them, the vector is a row vector with N1 elements, where each element is Default value∈ β It is a preset threshold based on the collection situation and can be 500mm to 700mm.

[0087] ④Traverse the sequence I[k] and delete all I[k′]-I[k]≤∈ i The elements I[k′] are obtained as the sequence I[k′] (k=1, 2, ..., K); where ∈ i is the preset threshold, which represents the minimum number of data points for each section of the vehicle, and can be set to 600. And k′>k;

[0088] ⑤At this time, the sequence I′[k] (k=1, 2, ..., K) is the subscript for dividing each section of data, that is, the two-dimensional matrix S of the bottom clearance of each car k [m, m] is the part from column I′[k] to column I′[k+1] in the two-dimensional matrix S1, and the number of data points in the length direction of each car is N. k =I′[k+1]-I[k], the number of data points in the width direction of each car M k=M, the number of train sections K is the number of data in I′[k].

[0089] (3) According to the point cloud construction algorithm, the bottom clearance two-dimensional matrix S of each car is k [m, m] corresponds to the actual position of the bottom of the vehicle, and S k The actual coordinate P corresponding to each data point in [m, m] k，m，n [x, y].

[0090] Specifically, the point cloud construction algorithm is:

[0091] L k and W k The length and width of the car section can be obtained by collecting the vehicle type information corresponding to the vehicle number when the train passes. Each vehicle type has a fixed length and width.

[0092] ① Determine the coordinate origin: Construct a coordinate system with the upper left corner of each car as the origin.

[0093] ②Determine the y-direction coordinate of the point cloud: S k The vertical coordinate of the i-th row of the matrix

[0094] ③Determine the x-direction coordinate of the point cloud: S k The vertical coordinate of the jth column of the matrix

[0095] ④ Then S k The actual coordinates corresponding to each data point in [m, n] are the coordinates of the point in the mth row and nth column of the kth car from the origin of this car. The height of the point from the rail surface is S k [m,n].

[0096] (4) According to the clearance extraction algorithm, the lowest T clearance values ​​and their positions under each car are extracted, where T is a preset value.

[0097] Specifically, the clearance extraction algorithm is:

[0098] For each car's bottom clearance two-dimensional matrix S k .

[0099] ①If T is greater than S k Number of data points M*N k , then M*N k Give T;

[0100] ②S k The elements in are laid out in order into array s k [j], where s k The first (m k -1)*n k+1 to m k *n k The element is S k mth k Row element, j = 1, 2, ..., M*N k ;

[0101] ③Array s k [j] Arrange in descending order to obtain the descending array s′ k [j] and the corresponding subscript conversion array r[j];

[0102] ④Select s′ k The first T elements in [j] are the T lowest values ​​d of the kth car. k，t , that is, the headroom value;

[0103] ⑤Select the first T elements in r[j] as the T lowest values ​​of the kth car in the array s′ k The subscript r k，t ;

[0104] ⑥Calculate the S of each lowest point in this section of the car k Coordinates (m k，t , n k，t ):

[0105]

[0106] n k，t =r k，t -m k，t ;

[0107] ⑦ By (m k，t , n k，t ) Calculate the coordinates of each lowest point in the original point cloud matrix (M k，t , N k，t ), that is, d k，t Corresponding position coordinates:

[0108] M k，t =m k，t ,

[0109]

[0110] The original point cloud matrix refers to the data matrix initially collected for the entire train. Specifically, the lowest altitude coordinates of the data after train separation need to correspond to the initially collected data for the entire train. For example, if the coordinates of the lowest point of car 5 are (230, 140), the coordinates in the original data are (230, 4140).

[0111] ⑧ By S k The actual coordinate P corresponding to each data point in [m, n] k，m，n [x, y], we can also get dk，t Corresponding to the position coordinates in this section of the car.

[0112] End detection.

[0113] The monitoring system corresponding to the above-mentioned railway vehicle undercarriage clearance detection method includes a monitoring module, a vehicle division algorithm module, a point cloud construction algorithm module and a clearance extraction algorithm module. The monitoring module is used to monitor and obtain the undercarriage clearance two-dimensional matrix S0[m,n] of the entire train in real time, whose size is M*N (m=1,2,...,M,n=1,2,...,N), where M and N are the number of data points in the width direction and length direction of the entire train respectively. The vehicle division algorithm module is used to combine the undercarriage clearance two-dimensional matrix of the entire train and obtain the number of train sections K and the number of data points N in the length direction and width direction of each section through the vehicle division algorithm. k and M k , the two-dimensional matrix S of the bottom clearance of each car k The point cloud construction algorithm module is used to construct the bottom clearance two-dimensional matrix S of each vehicle through the point cloud construction algorithm. k [m, n] corresponds to the actual position of the bottom of the vehicle, and S is obtained k The actual coordinate P corresponding to each data point in [m, n] in the car k，m，n [x, y]. Clearance extraction algorithm module is used to calculate the two-dimensional matrix S of the clearance under each car. k ,Through the clearance extraction algorithm, the lowest T clearance values ​​and their positions under each car are extracted, where T is a preset value.

[0114] Specifically, the car allocation algorithm module performs the following car allocation algorithm operations:

[0115] ①Convolve the two-dimensional matrix S0[m,n] of the vehicle bottom clearance of the entire train with the filter F1[m,n] to obtain the filtered waveform S F [m, m], that is:

[0116]

[0117] ②From S F The calculation starts from the first row / column of data. If the average value of the row / column data is greater than the preset value H and the standard deviation is less than the preset value ∈ v , then delete all rows / columns from the first row / column to the row / column in S0[m,m], and finally get the two-dimensional matrix S i [m, n], whose size is M1*N1;

[0118] ③The vector Multiplying S1 on the left gives β, which is:

[0119] β=α*S1

[0120] Select all the values ​​in β that are less than the preset value ∈ β The subscripts of form the sequence I[k];

[0121] ④Traverse the sequence I[k] and delete all I[k′]-I[k]≤∈ i The elements I[k′] of the sequence I′[k] (k=1, 2, ..., K); where ∈ i is a preset threshold, and k′>k;

[0122] ⑤At this time, the sequence I′[k] (k=1, 2, ..., K) is the subscript for dividing each section of data, where K≤K0, that is, the two-dimensional matrix S of the bottom clearance of each section of the car k [m, n] is the part from column I′[k] to column I′[k+1] in the two-dimensional matrix S1, and the number of data points in the length direction of each car is N. k =I′[k+1]-I[k], the number of data points in the width direction of each car M k =M.

[0123] Specifically, the point cloud construction algorithm module performs the following point cloud construction algorithm operations:

[0124] ① Determine the coordinate origin: construct a coordinate system with the upper left corner of each car as the origin;

[0125] ②Determine the y-direction coordinate of the point cloud: S k The vertical coordinate of the i-th row of the matrix

[0126] ③Determine the x-direction coordinate of the point cloud: S k The vertical coordinate of the jth column of the matrix

[0127] ④ Then S k The actual coordinates corresponding to each data point in [m, n] are the coordinates of the point in the mth row and nth column of the kth car from the origin of this car. The height of this point from the rail surface is S k [m, n];

[0128] Among them, L k 、W k is the length and width of the car section.

[0129] Specifically, the headroom extraction algorithm module performs the following headroom extraction algorithm operations:

[0130] ①If T is greater than S k Number of data points M*N k , then M*N k Give T;

[0131] ②S kThe elements in are laid out in order into array s k [j], where s k The first (m k -1)*n k +1 to m k *n k The element is S k mth k Row element, j = 1, 2, ..., M*N k ;

[0132] ③Array s k [j] Arrange in descending order to obtain the descending array s′ k [j] and the corresponding subscript conversion array r[j];

[0133] ④Select s′ k The first T elements in [j] are used as the T clearance values ​​d of the kth car. k，t ;

[0134] ⑤Select the first T elements in r[j] as the T lowest values ​​of the kth car in the array s′ k The subscript r k，t ;

[0135] ⑥Calculate the S of each lowest point in this section of the car k Coordinates (m k，t , n k，t ):

[0136]

[0137] n k，t =r k，t -m k，t ;

[0138] ⑦ By (m k，t , n k，t ) Calculate the coordinates of each lowest point in the original point cloud matrix (M k，t , N k，t ), that is, d k，t Corresponding position coordinates:

[0139] M k，t =m k，t ,

[0140]

[0141] By S k The actual coordinate P corresponding to each data point in [m, n] k，m，n [x, y], get d k，t Corresponding to the position coordinates in this section of the car.

[0142] The above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the scope of implementation of the present invention. Any equivalent changes and modifications made within the scope of protection of the present invention should be considered to fall within the scope of protection of the present invention.

Claims

1. A method for detecting the clearance under a railway vehicle, characterized in that: The method comprises the following steps: (1) Monitor and obtain the two-dimensional matrix S0[m,n] of the vehicle bottom clearance of the entire train, whose size is M*N, m=1,2,...,M,n=1,2,...,N, where M and N are the number of data points in the width direction and length direction of the entire train respectively; (2) Combined with the two-dimensional matrix of the vehicle bottom clearance of the entire train, the number of train sections K and the number of data points N in the length and width directions of each section are obtained through the vehicle division algorithm. k and M k , the two-dimensional matrix S of the bottom clearance of each car k [m, n]; (3) Through the point cloud construction algorithm, the bottom clearance two-dimensional matrix S of each car is k [m,n] corresponds to the actual position of the bottom of the vehicle, and S is obtained k The actual coordinate P corresponding to each data point in [m,n] in the car section k,m,n [x,y]; (4) For each car's bottom clearance two-dimensional matrix S k [m,n], through the clearance extraction algorithm, extract the lowest T clearance values ​​and their positions under each car, where T is a preset value; In step (2), the vehicle division algorithm is: ①Convolve the two-dimensional matrix S0[m,n] of the vehicle bottom clearance of the entire train with the filter F1[m,n] to obtain the filtered waveform S F [m, n], that is: ②From S F The calculation starts from the first row / column of data. If the average value of the row / column data is greater than the preset value H and the standard deviation is less than the preset value ∈ v , then delete all rows / columns from the first row / column to the row / column in S0[m,n], and finally obtain a two-dimensional matrix S1[m,n] with a size of M1*N1; ③The vector Multiplying S1 on the left gives β, which is: β＝α*S1 Select all the values ​​in β that are less than the preset value ∈ β The subscripts of form the sequence I[k]; ④Traverse the sequence I[k] and delete all I[k′]-I[k]≤∈ i The element I[k′] is obtained as the sequence I′[k], k=1,2,...,K; where ∈ i is a preset threshold, and k′>k; ⑤At this time, the sequence I′[k], k=1,2,..., K is the subscript for dividing each section of data, that is, the two-dimensional matrix S of the bottom clearance of each section of the car k [m,n] is the part from column I′[k] to column I′[k+1] in the two-dimensional matrix S1, and the number of data points in the length direction of each car is N. k =I'[k+1]-I[k], the number of data points in the width direction of each car M k =M.

2. The vehicle bottom clearance detection method according to claim 1, characterized in that: In step (3), the point cloud construction algorithm is: ① Determine the coordinate origin: construct a coordinate system with the upper left corner of each car as the origin; ②Determine the y-direction coordinate of the point cloud: S k The vertical coordinate of the i-th row of the matrix ③Determine the x-direction coordinate of the point cloud: S k The vertical coordinate of the jth column of the matrix ④ Then S k The actual coordinates corresponding to each data point in [m, n] are the coordinates of the point in the mth row and nth column of the kth car from the origin of this car. The height of this point from the rail surface is S k [m, n]; Among them, L k 、W k is the length and width of the car section.

3. The vehicle bottom clearance detection method according to claim 1, characterized in that: In step (4), the headroom extraction algorithm is: ①If T is greater than S k Number of data points M*N k , then M*N k Give T; ②S k The elements in are laid out in order into array s k [j], where s k The first (m k -1)*n k +1 to m k *n k The element is S k mth k Row element, j = 1, 2, ..., M*N k ; ③Array s k [j] Arrange in descending order to obtain the descending array s′ k [j] and the corresponding subscript conversion array r[j]; ④Select s′ k The first T elements in [j] are used as the T clearance values ​​d of the kth car. k，t ; ⑤Select the first T elements in r[j] as the T lowest values ​​of the kth car in the array s′ k The subscript r k，t ; ⑥Calculate the S of each lowest point in this section of the car k Coordinates (m k，t , n k，t ): n k,t =r k,t -m k，t ; ⑦ By (m k，t , n k，t ) Calculate the coordinates of each lowest point in the original point cloud matrix (M k，t , N k，t ), that is, d k，t Corresponding position coordinates: M k，t =m k，t , 4. The vehicle bottom clearance detection method according to claim 3, characterized in that: The step (4) further includes: k The actual coordinate P corresponding to each data point in [m, n] k，m，n [x, y], get d k，t Corresponding to the position coordinates in this section of the car.

5. A railway vehicle undercarriage clearance detection system, characterized in that: The system includes: The monitoring module is used to monitor and obtain the two-dimensional matrix S0[m,n] of the vehicle bottom clearance of the entire train in real time. Its size is M*N, m=1,2,...,M, n=1,2,...,N, where M and N are the number of data points in the width direction and length direction of the entire train respectively; The car division algorithm module is used to combine the two-dimensional matrix of the vehicle bottom clearance of the entire train, and obtain the number of train sections K and the number of data points N in the length and width directions of each section through the car division algorithm. k and M k , the two-dimensional matrix S of the bottom clearance of each car k [m, n]; Point cloud construction algorithm module is used to construct the bottom clearance two-dimensional matrix S of each car through the point cloud construction algorithm. k [m, n] corresponds to the actual position of the bottom of the vehicle, and S is obtained k The actual coordinate P corresponding to each data point in [m, n] in the car k，m，n [x, y]; and Clearance extraction algorithm module, used for the two-dimensional matrix S of the clearance under each car k [m, n], through the clearance extraction algorithm, extract the lowest T clearance values ​​and their positions under each car, where T is a preset value; The car allocation algorithm module performs the following car allocation algorithm operations: ①Convolve the two-dimensional matrix S0[m,n] of the vehicle bottom clearance of the entire train with the filter F1[m,n] to obtain the filtered waveform S F [m, n], that is: ②From S F The calculation starts from the first row / column of data. If the average value of the row / column data is greater than the preset value H and the standard deviation is less than the preset value ∈ v , then delete all rows / columns from the first row / column to the row / column in S0[m,n], and finally obtain a two-dimensional matrix S1[m,n] with a size of M1*N1; ③The vector Multiplying S1 on the left gives β, which is: β＝α*S1 Select all the values ​​in β that are less than the preset value ∈ β The subscripts of form the sequence I[k]; ④Traverse the sequence I[k] and delete all I[k′]-I[k]≤∈ i The element I[k′] is obtained as the sequence I′[k], k=1,2,...,K; where ∈ i is the preset threshold, and k′>k; ⑤At this time, the sequence I′[k], k=1,2,...,K is the subscript for dividing each section of data, where K≤K0, that is, the two-dimensional matrix S of the bottom clearance of each section of the car k [m,n] is the part from column I′[k] to column I′[k+1] in the two-dimensional matrix S1, and the number of data points in the length direction of each car is N. k =I'[k+1]-I[k], the number of data points in the width direction of each car M k =M.

6. The vehicle bottom clearance detection system according to claim 5, characterized in that: The point cloud construction algorithm module performs the following point cloud construction algorithm operations: ① Determine the coordinate origin: construct a coordinate system with the upper left corner of each car as the origin; ②Determine the y-direction coordinate of the point cloud: S k The vertical coordinate of the i-th row of the matrix ③Determine the x-direction coordinate of the point cloud: S k The vertical coordinate of the jth column of the matrix ④ Then S k The actual coordinates corresponding to each data point in [m, n] are the coordinates of the point in the mth row and nth column of the kth car from the origin of this car. The height of this point from the rail surface is S k [m, n]; Among them, L k 、W k is the length and width of the car section.

7. The vehicle bottom clearance detection system according to claim 5, characterized in that: The headroom extraction algorithm module performs the following headroom extraction algorithm operations: ①If T is greater than S k Number of data points M*N k , then M*N k Give T; ②S k The elements in are laid out in order into array s k [j], where s k The first (m k -1)*n k +1 to m k *n k The element is S k mth k Row element, j = 1, 2, ..., M*N k ; ③Array s k [k] Arrange in descending order to get the descending array s′ k [k] and the corresponding subscript conversion array r[j]; ④Select s′ k The first T elements in [j] are used as the T clearance values ​​d of the kth car. k，t ; ⑤Select the first T elements in r[j] as the T lowest values ​​of the kth car in the array s′ k The subscript r k，t ; ⑥Calculate the S of each lowest point in this section of the car k Coordinates (m k，t , n k，t ): n k,t =r k,t -m k,t ; ⑦ By (m k，t , n k，t ) Calculate the coordinates of each lowest point in the original point cloud matrix (M k，t , N k，t ), that is, d k，t Corresponding position coordinates: M k,t =m k，t , 8. The vehicle bottom clearance detection system according to claim 7, characterized in that: The headroom extraction algorithm also includes: k The actual coordinate P corresponding to each data point in [m, n] k，m，n [x, y], get d k，t Corresponding to the position coordinates in this section of the car.

Images