A new method for detecting wear of metro pantograph based on 2D laser measurement

Abstract

The application discloses a new method for detecting metro pantograph abrasion based on 2D laser measurement, and is applied to the field of metro pantograph abrasion detection. In the process of metro operation, the image effect obtained by detection is not ideal due to poor light conditions in the metro tunnel, accurate detection values cannot be obtained, the metro runs at a high speed, and there is also a problem of missed detection. The high sampling rate of the 2D laser can collect full-coverage profile data of the pantograph passing at a high speed. The obtained metro pantograph profile point cloud data are subjected to point cloud preprocessing, feature point recognition and point cloud registration to obtain a complete profile of the carbon slide plate of the metro pantograph. In combination with a standard carbon slide plate profile, the pantograph abrasion value can be calculated. The scheme of the application effectively avoids the influence of the tunnel light environment on the overall measurement accuracy of the pantograph in the image detection method, thereby greatly improving the metro pantograph abrasion detection precision.

Description

Technical Field

[0001] This invention belongs to the field of rail transit, and specifically relates to a pantograph wear detection technology for subways. Background Technology

[0002] With the increasing number of subway lines and the continuous expansion of the subway system, the safety and reliability of subway operations have received increasing attention. The pantograph, as a crucial component of the power supply system during subway operation, is a vital link in ensuring the smooth and safe operation of the subway. The wear and tear of the pantograph's carbon contact plate directly affects the stability and safety of the subway's power supply. During daily subway operation, without real-time monitoring of the pantograph, the accumulated wear of the carbon contact plate can significantly lead to temporary power outages or even major traffic accidents. Therefore, effectively and accurately detecting the wear of the subway pantograph's carbon contact plate is a critical issue.

[0003] Currently, there are two main methods for detecting pantograph wear in subways: contact measurement and non-contact measurement. Contact measurement primarily involves manual inspection from the top of the tunnel. This method can accurately and effectively measure the wear value of the pantograph's carbon contactor, analyze existing defects, and avoids missed detections. However, manual inspection has significant drawbacks. It can only be performed after the subway has been shut down and is in storage. Furthermore, the process requires workers to climb the tunnel repeatedly, compromising their safety. Additionally, the low efficiency fails to meet the requirements of efficient subway management. To address the problems of manual inspection, reduce costs, and improve efficiency, non-contact measurement has gradually gained traction. Image detection, as a primary non-contact method, enables online inspection during subway operation, eliminating the need for in-situ inspections, saving labor costs, and ensuring the safety of subway operations. However, this method also has limitations: during subway operation, poor lighting conditions in the tunnel result in unsatisfactory image quality, making accurate detection values ​​impossible. Additionally, the high speed of the subway increases the possibility of missed detections. A key issue is how to achieve efficient and high-precision detection of wear on the carbon sliding plate of the pantograph in high-speed subway vehicles under tunnel conditions. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention proposes a novel method for detecting pantograph wear in subways based on 2D laser measurement. This method is less affected by the lighting environment in subway tunnels and can adapt to the higher speeds of subway trains.

[0005] The technical solution adopted in this invention is: a new method for detecting pantograph wear in subways based on 2D laser measurement, comprising:

[0006] A1. A point laser sensor facing the oncoming vehicle direction is used to obtain the pantograph approach signal, and the pantograph approach signal is transmitted to the 2D laser sensor behind the point laser sensor in the form of a high level.

[0007] After receiving a high-level signal, the A2 and 2D laser sensors start working and continuously collect data on the pantograph for a period of time, obtaining pantograph point cloud profile data in CSV format.

[0008] A3. Preprocess the pantograph point cloud profile data in CSV format acquired by the 2D laser sensor, specifically including data filtering and data screening, to achieve data denoising and smoothing.

[0009] A4. Normalize the pantograph point cloud profile data obtained after preprocessing in step A3, and extract the carbon skateboard point cloud profile data based on the normalized pantograph point cloud profile data.

[0010] A5. Register the carbon pantograph point cloud profile data obtained in step A4 using the standard carbon pantograph point cloud profile data, thereby calculating the pantograph wear value.

[0011] Step A4 involves processing the left and right profile data of the pantograph point cloud to extract the left and right profile data of the pantograph carbon skid plate; specifically, it includes the following sub-steps:

[0012] A41. Normalize the left or right profile data of the pantograph point cloud obtained after preprocessing in step A3, and map the ordinate of each point cloud profile data to the interval [0,1].

[0013] A42. Perform interval sampling on the normalized left or right profile data of the pantograph point cloud, and then divide it into three equal intervals. Starting from the left or right pantograph horn, these intervals are designated as the first, second, and third equal intervals, respectively. The two division points are designated as the first division point and the second division point, respectively, based on their proximity to the left or right pantograph horn.

[0014] A43. Define the length of the point cloud line segment as l, and the step size of the line segment each time it moves as m; traverse the first and second equal intervals;

[0015] A44. For each traversal operation, calculate the variance of the ordinate of the point cloud profile data contained in the currently traversed point cloud segment.

[0016] A45. Find the minimum variance during the traversal of the first two equal intervals, and obtain the position index of the point cloud line segment in the first and second equal intervals when the minimum variance is obtained.

[0017] A46. Construct a straight line that passes through the point cloud profile data coordinates corresponding to the position index in step A45 and the coordinates of the second division point.

[0018] A47. Calculate the distances from the point cloud profile data corresponding to the position index in step A45 to the line described in step A46 between the second division point and the point cloud profile data coordinates. The point with the maximum distance is the dividing point between the pantograph horn and the carbon skateboard.

[0019] A48. Based on the dividing point obtained in step A47, extract the left or right profile data of the pantograph carbon slide plate.

[0020] The beneficial effects of this invention are as follows: The subway pantograph wear detection method based on 2D laser measurement provided by this invention utilizes the high sampling rate of 2D laser to collect full-coverage profile data of pantographs passing at high vehicle speeds. Using the obtained pantograph profile point cloud data, point cloud preprocessing, feature point recognition, and point cloud registration are performed to obtain the complete profile of the subway pantograph carbon skid plate. Combined with the profile of a standard carbon skid plate, the pantograph wear value can be calculated. This invention solves the difficulties in measuring subway pantograph wear using current image detection methods. Furthermore, based on the principle of laser triangulation, this method only requires the laser to contact the object during measurement to obtain the distance data of the detection target and construct the actual profile of the detection target. This effectively avoids the influence of tunnel lighting environment on the overall measurement accuracy of the pantograph in image detection methods, thus greatly improving the accuracy of subway pantograph wear detection. Utilizing the advantages of high sampling frame rate, high data accuracy, and strong anti-interference capability of 2D laser sensors, efficient and high-precision measurement of subway pantograph wear is achieved, laying the foundation for further improving the accuracy of subway pantograph wear detection. Attached Figure Description

[0021] Figure 1 This is a flowchart of the method of the present invention;

[0022] Figure 2 This is a schematic diagram of the 2D laser sensor detecting the pantograph carbon slider in this invention;

[0023] Figure 3 This is a schematic diagram of the feature point search process in this invention. Detailed Implementation

[0024] To facilitate understanding of the technical content of this invention by those skilled in the art, the following description, in conjunction with the accompanying drawings, further illustrates the invention.

[0025] like Figure 1 As shown, this invention provides a method for detecting pantograph wear in subways based on 2D laser measurement, comprising the following steps:

[0026] S1, such as Figure 2As shown, two 2D laser sensors are installed on the top of the tunnel and can be finely adjusted in three degrees of freedom. The point laser sensor is installed on the top of the tunnel in front of the 2D laser sensors, according to the direction of the subway train and the centerline of the track. The vertical distance between the installation position and the installation point of the 2D laser sensor is 10m.

[0027] In this step, two 2D laser sensors should be installed symmetrically to the centerline of the track, with the distance between the sensors and the centerline being 270mm; the point laser sensor is fixedly installed on the top of the tunnel according to the centerline of the track.

[0028] S2. When the subway pantograph passes the point laser sensor, the point laser sensor scans the pantograph's carbon sliding plate and transmits the collected pantograph proximity signal to the 2D laser sensor in the form of a high level. Each time the pantograph passes by, the point laser sensor transmits a high level signal to the 2D laser sensor once.

[0029] In this embodiment, the point laser sensor needs to be set to a fixed sampling frame rate of 3kHz to ensure that the point laser sensor can be stably triggered when the subway speed is 30-40km / h; after the point laser is triggered once, it will not be triggered a second time within a 2-second time range.

[0030] The S3 and 2D laser sensors start working after receiving a high-level signal. The maximum acquisition range of a single sensor on the horizontal axis is 720mm and the maximum acquisition range on the vertical axis is 1352mm. They continuously acquire data for 3 seconds to obtain the point cloud profile of the pantograph carbon skid in CSV format. The 2D laser sensor works once every time a high-level signal is received, and the controller transmits the acquired pantograph carbon skid CSV point cloud data to the local industrial control computer via a switch.

[0031] In this step, the two 2D laser sensors need to be set to a fixed sampling frame rate of 1kHz to ensure accurate acquisition of pantograph profile data even at high subway speeds. Based on the installation distance between the two 2D laser sensors and the positioning laser sensor, and considering the subway speed of 30-40 km / h, the batch processing count should be set to 3000 to ensure the 2D lasers fully cover the entire pantograph's passage. Simultaneously, the two 2D laser sensors need to be set to emit light alternately to avoid interference between them; each time the subway passes, both the positioning laser sensor and the 2D laser sensors will only trigger twice, obtaining two sets of pantograph CSV point cloud profile data.

[0032] Two 2D laser sensors each collect half of the pantograph profile data, which are recorded as the left pantograph profile data and the right pantograph profile data.

[0033] S4. Preprocess the pantograph carbon skid plate CSV point cloud data acquired by the 2D laser sensor, set special threshold conditions to filter the data, and perform data filtering to complete the data denoising and smoothing, eliminating invalid noise and burrs in the pantograph carbon skid plate CSV point cloud data.

[0034] Furthermore, step S4 also includes the following steps:

[0035] S41. Remove invalid point clouds from the pantograph point cloud profile data obtained in step S3.

[0036] In this step, the pantograph carbon slider CSV point cloud data (x) obtained in step S3 is used. r ,y r The valid data is mainly distributed in the range [-300, 0], while the invalid data is mainly located in the range [-999, -400]. Based on the point cloud spacing of 0.225 mm, the actual pantograph profile length, and the significant numerical difference between invalid and valid point cloud data, a point cloud filtering threshold Y is set. v = -400, point cloud valid value determination completed:

[0037] y r ≥Y v (1)

[0038] Calculate the number of valid points T in each frame of point cloud data; based on the length of the pantograph carbon slider of 1050mm and the point cloud spacing of 0.225mm, set the threshold T for the number of valid points T in each frame of point cloud data from a single 2D laser sensor. d =2800, to achieve point cloud profile validity determination:

[0039] T≥T d (2)

[0040] Point cloud profiles that meet conditions (1) and (2) are considered valid profiles, while those that do not are considered invalid profiles. This allows for the selection of relatively complete point cloud profile data (x) of the pantograph carbon skateboard. d ,y d Eliminate invalid point cloud data in the collected data;

[0041] S42. The effective pantograph point cloud profile data obtained in step S41 is filtered and smoothed.

[0042] The effective pantograph point cloud profile data obtained in step S41 is filtered and smoothed using the Savitzky-Golay convolutional smoothing algorithm. Specifically, the steps are as follows: Taking 2w+1 data points centered at x=0, a filtering window width s=2w+1 is set, meaning the sampling point set within the window is x=(-w,-w+1,....,0,....,w-1,w). An n-1 order polynomial is constructed to fit the data point set within the window, and fitting parameters a0, a1, a2...a... are set. n-2 a n-1 And the formula for fitting data points within the window:

[0043] y = a0 + a1x + a2x 2 +...+a n-2 x n-2 +a n-1 x n-1 (3)

[0044] We obtain n fitting coefficients a0, a1, a2, a... n-2 a n-1 The equation is given by formula (4). If the equation has a solution, then s>n. The point cloud fitting parameters A within the filtering window are obtained by least squares fitting. Thus, we can obtain:

[0045]

[0046] Convert to matrix form:

[0047] Y (2w+1)×1 =X (2w+1 ) ×n ·A n×1 +B (2w+1)×1 (5)

[0048] Through the transpose of X, X T Find the least squares solution for A.

[0049]

[0050] The model prediction value of Y can then be obtained.

[0051]

[0052] Here, E represents the identity matrix.

[0053] S5. Using the pantograph point cloud profile data obtained after preprocessing in step S4. The profile data is normalized and mapped to a specific interval [0,1] for further processing. Data is sampled and the profile data is divided into three equal parts. The first two parts are selected to narrow down the search interval for feature points in the point cloud profile. Simultaneously, a point cloud line segment and its step size are defined. Starting from the first part of the profile interval, the selected two parts are iteratively searched to find feature points. The variance of the point cloud contained in the line segment after each movement is calculated. Combined with the actual profile characteristics of the pantograph carbon slide plate, the index point A1 of the profile position of the line segment when its variance is minimized is obtained. The second division point of the point cloud profile is selected as A2. Figure 3 As shown, a straight line is determined by two points, and then the distance formula between the two points is used to iterate and calculate the remaining points between the two points in turn to find the point with the largest distance. This yields the optimal feature point X of the pantograph CSV point cloud profile, thus completing the extraction of the left and right profile data of the pantograph carbon slide plate.

[0054] The original profile data were normalized using the min-max normalization method. After processing, the profile data y in the interval [0,1] is obtained. s :

[0055]

[0056] For the normalized point cloud profile data (x, y) s Perform interval sampling to obtain a new dataset (x). c ,y c Then, the interval is divided into three equal parts to reduce the amount of data to be processed. Those skilled in the art should know that the pantograph's bow-shaped structure specifically includes the left and right pantograph horn parts and the middle carbon sliding plate part. Starting from the pantograph horn, the interval closest to the pantograph horn is called the first equal interval, and the intervals successively farther away from the pantograph horn are called the second equal interval, the third equal interval, and so on. The dividing point between the pantograph horn and the carbon sliding plate is located within the first and second equal intervals. The dividing point between the first and second equal intervals is called the first dividing point, and the other dividing point is called the second dividing point.

[0057] First dividing point M j (x j ,y j ) and the second dividing point M k (x k ,y k The first two parts of the outline interval contain k point clouds. The length of the point cloud line segment is defined as l, the step size of each line segment movement is m, and the traversal terminates at point M. j (x j ,y j ), to obtain the total number of rounding iterations i for each equally divided interval:

[0058]

[0059] In this embodiment, the length of the point cloud line segment is l = 20 point cloud data points, and the step size of the line segment movement each time is m = 2 point cloud data points.

[0060] For each traversal operation, the variance of the point cloud contained in the traversed line segment is calculated, and the variance is calculated based on the point cloud value y of the line segment. i And the mean y of the point cloud line segments, to obtain the variance of the point cloud line segments:

[0061]

[0062] After completing i iterations, calculate the variance after each iteration, and combine this variance with the actual profile characteristics of the pantograph to find the minimum variance value s. 2 min When the minimum variance value is obtained, the position index of the point cloud line segment in the first two parts of the profile interval is recorded as A1(x). a1 ,y a1 Simultaneously, record the second division point of the point cloud outline into three equal parts as M. k (x k ,y k ), by A1(x a1 ,y a1 ) and M k (x k ,y k The two points form a straight line y a :

[0063]

[0064] Using the line y a For A1(x) a1 ,y a1 ) and M k (x k ,y k Iterate through all points between the two points and calculate the distance from each point to the line y. a The distance between the point and the line is obtained by finding the distance d. i :

[0065]

[0066] Statistics A1(x) a1 ,y a1 ) and M k (x k ,y k All points between two points on the line y a The distance is used to find the point A with the maximum distance. m (xam ,y am This point is the boundary between the pantograph horn and the carbon sliding plate, which is the optimal feature point of the pantograph that needs to be searched. The horn part is removed, and the point cloud of the carbon sliding plate part is retained to obtain the left and right outlines of the subsequent registration point cloud.

[0067] S6. Using the standard carbon skateboard point cloud profile data and the carbon skateboard point cloud profile data obtained in step S5, select the unworn area at the bend of the carbon skateboard and define their respective registration intervals. Then, define the registration interval of the standard carbon skateboard profile as the target point cloud P, and the registration interval of the carbon skateboard profile in S5 as the source point cloud Q. Initialize the distance between the two point clouds and define the rotation matrix R and the translation matrix T. Normalize the two point clouds to obtain point sets p and q. Calculate the Euclidean distance between each pair of points in point sets p and q, and determine the corresponding point pair p of point sets p and q based on the minimum Euclidean distance. j With q j Based on the least squares method, the objective function D(R,T) is defined as the sum of the Euclidean distances between all corresponding point pairs of two point sets. The goal is to find the minimum objective function D. min The corresponding optimal rotation matrix R and translation matrix T are then used to perform rotation and translation transformations on the source point cloud Q and the carbon skateboard point cloud profile data obtained in step S5, to obtain a new source point cloud Q′ and the carbon skateboard point cloud profile after one iteration.

[0068] Based on the sum of squared Euclidean distances D′(R,T) between corresponding point pairs in the target point cloud and the source point cloud obtained from the iterative results of the field test data, the iteration termination condition L is still set according to the sum of squared Euclidean distances between corresponding point pairs. m =0.3, and based on the convergence of the iterations using on-site test data, the number of iterations does not exceed 10. Therefore, the maximum number of iterations required for the rotation and translation of the source point cloud Q is set to 0.3. max =20 is another iteration condition; repeat the iteration calculation, and the iteration terminates after one of the conditions is met, and the final registration result between the two point clouds is obtained, so that the wear value of the pantograph can be calculated.

[0069] The registration interval for the standard carbon skateboard profile is defined as the target point cloud P(x). p ,y p In step S5, the registration interval for the actual measured profile of the carbon skateboard is the source point cloud Q(x). q ,y q ), P(x p ,y p ), Q(x) q ,y q The number of point clouds in each interval is n. Define the rotation angle θ and the translation variables Δx and Δy, and initialize the rotation matrix R and the translation matrix T:

[0070]

[0071]

[0072] For P(x) p ,y p ) and Q(x q ,y q Normalize the two point sets to obtain point sets p and q:

[0073]

[0074] p = PP mean (16)

[0075]

[0076] q = QQ mean (18)

[0077] Calculate the Euclidean distances between all pairs of points in two point sets p and q, and determine the points in point set p by using the minimum Euclidean distance. The corresponding point in point set q This yields a set of corresponding point pairs. and

[0078]

[0079] Repeat step (19) to determine all corresponding point pairs in point sets p and q.

[0080] Define the sum of squared Euclidean distances between corresponding point pairs in two point sets p and q as the objective function D(R,T):

[0081]

[0082] Calculate the minimum objective function D min The corresponding optimal rotation matrix R and translation matrix T, after the above normalization, can simplify the objective function:

[0083]

[0084] Convert equation (21) into matrix form:

[0085]

[0086] After deriving and rearranging equation (22), we can obtain:

[0087]

[0088] p k q kGiven constant values, solve for the minimum objective function D. min The problem can be transformed into a maximum value problem, let:

[0089]

[0090] With θ as the variable, we can find the maximum value of G(R,T) by taking the derivative of G(R,T):

[0091]

[0092] Further derivation:

[0093]

[0094] Solving for θ using inverse trigonometric functions:

[0095]

[0096] Using the obtained θ, the rotation matrix R can be obtained, and the translation matrix T can also be obtained:

[0097]

[0098] The rotation and translation matrices R and T obtained from the above registration can be used to perform a point cloud transformation on the source point cloud Q, thereby obtaining a new point cloud Q′:

[0099] Q′=R·Q+T (29)

[0100] Calculate the sum of squared Euclidean distances between corresponding point pairs in the new point cloud Q′ and the target point cloud P:

[0101]

[0102] Based on the sum of squared Euclidean distances D′(R,T) between corresponding point pairs in the target point cloud and the source point cloud obtained from the iterative results of the field test data, the iteration termination condition L is still set according to the sum of squared Euclidean distances between corresponding point pairs. m =0.3, and simultaneously set the maximum number of iterations required for the source point cloud Q to perform rotations and translations. max =20 is another iteration termination condition. Repeat the above registration steps, and perform an iteration judgment every time. If the condition is satisfied: the sum of squared Euclidean distances between corresponding point pairs in the two point clouds D′(R,T)≤L min Or the number of iterations I ≥ iteration max If the iteration terminates, the rotation and translation matrices of each iteration are used to perform rotation and translation on the carbon skateboard point cloud profile data obtained in step S5 to obtain the final point cloud profile.

[0103] The above operation only processes the data collected by one of the two sensors. The same operation is performed on the profile data collected by the other sensor to obtain the final point cloud profile after the two sensors have collected and processed. By comparing the final left and right point cloud profiles with the standard carbon skateboard point cloud profile data, the actual measured carbon skateboard profile wear data can be obtained.

[0104] This invention achieves non-contact, online detection of pantograph wear in subway systems. A positioning laser sensor and a 2D laser sensor are installed at the top of the tunnel. The positioning laser sensor detects the approach of the pantograph in advance, triggering the 2D laser to collect pantograph profile data. Using a threshold setting method, effective profile data is extracted from both the numerical and quantitative aspects of the point cloud. Combined with the Savitzky-Golay convolution smoothing algorithm, the profile data is smoothed and filtered, thus deriving a pantograph profile data preprocessing algorithm. The actual pantograph profile characteristics are analyzed, and feature points are found through interval iterative search, interval variance calculation, and point-line distance relationships, resulting in a pantograph feature point search algorithm. A registration interval is defined between the measured carbon skateboard profile and the standard carbon skateboard profile. An objective function for the two point sets is established based on the least squares method. By solving for the rotation and translation matrices of the minimum objective function, iterative rotation and translation between the two point sets are achieved, thus deriving a pantograph point cloud profile registration algorithm. This method simplifies the pantograph wear detection process in subway systems, improves the accuracy of pantograph wear detection, and overcomes the shortcomings of detection results being greatly affected by the tunnel environment. It also reduces labor and time costs. It has the advantages of high accuracy, high efficiency, good stability, and low cost.

[0105] Those skilled in the art will recognize that the embodiments described herein are intended to help the reader understand the principles of the invention, and should be understood that the scope of protection of the invention is not limited to such specific statements and embodiments. Various modifications and variations can be made to the invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the invention should be included within the scope of the claims of the invention.

Claims

1. A novel method for detecting pantograph wear in subway systems based on 2D laser measurement, characterized in that, include: A1. A point laser sensor facing the oncoming vehicle direction is used to obtain the pantograph approach signal, and the pantograph approach signal is transmitted to the 2D laser sensor behind the point laser sensor in the form of a high level. After receiving a high-level signal, the A2 and 2D laser sensors start working and continuously collect data on the pantograph for a period of time, obtaining pantograph point cloud profile data in CSV format. A3. Preprocess the pantograph point cloud profile data in CSV format acquired by the 2D laser sensor, specifically including data filtering and data screening, to achieve data denoising and smoothing. A4. Normalize the pantograph point cloud profile data obtained after preprocessing in step A3. Based on the normalized pantograph point cloud profile data, extract the carbon skid plate point cloud profile data. Step A4 includes processing the left and right profile data of the pantograph point cloud to complete the extraction of the left and right profile data of the pantograph carbon skid plate. Specifically, it includes the following sub-steps: A41. Normalize the left or right profile data of the pantograph point cloud obtained after preprocessing in step A3, and map the ordinate of each point cloud profile data to... Within the range; A42. Perform interval sampling on the normalized left or right profile data of the pantograph point cloud, and then divide it into three equal intervals. Starting from the left or right pantograph horn, these intervals are designated as the first, second, and third equal intervals, respectively. The two division points are designated as the first division point and the second division point, respectively, based on their proximity to the left or right pantograph horn. A43. Define the length of the point cloud line segment as... The step size of the line segment each time is Traverse the first and second equally divided intervals; A44. For each traversal operation, calculate the variance of the ordinate of the point cloud profile data contained in the currently traversed point cloud segment. A45. Find the minimum variance during the traversal of the first two equal intervals, and obtain the position index of the point cloud line segment in the first and second equal intervals when the minimum variance is obtained. A46. Construct a straight line that passes through the point cloud profile data coordinates corresponding to the position index in step A45 and the coordinates of the second division point. A47. Calculate the distances from the point cloud profile data corresponding to the position index in step A45 to the line described in step A46 between the second division point and the point cloud profile data coordinates. The point with the maximum distance is the dividing point between the pantograph horn and the carbon skateboard. A48. Based on the dividing point obtained in step A47, extract the left or right profile data of the pantograph carbon sliding plate. A5. Register the carbon pantograph point cloud profile data obtained in step A4 using the standard carbon pantograph point cloud profile data, thereby calculating the pantograph wear value.

2. A novel method for detecting pantograph wear in subways based on 2D laser measurement, as described in claim 1, is characterized in that... Two 2D laser sensors were used to collect left profile data and right profile data of the pantograph point cloud, respectively.

3. A novel method for detecting pantograph wear in subways based on 2D laser measurement, as described in claim 2, is characterized in that... Two 2D laser sensors are set to a fixed sampling frame rate of 1kHz.

4. A novel method for detecting pantograph wear in subways based on 2D laser measurement, as described in claim 3, is characterized in that... The number of batch processing points is set based on the installation distance between the two 2D laser sensors and the positioning laser sensor, combined with the speed of the subway.

5. A novel method for detecting pantograph wear in subways based on 2D laser measurement, as described in claim 4, is characterized in that... The two 2D laser sensors are set to emit light alternately.

Images