Measuring Method and Measuring System for Brake Pads of Rail Vehicles

Abstract

The present invention discloses a measurement method and system for a brake pad of a rail vehicle. The measurement method includes: an image acquisition step of acquiring an image of the brake pad of the rail vehicle; a segmentation parameter obtaining step of obtaining segmentation parameters based on the STOA-DE algorithm according to the brake pad image; a binary image obtaining step of segmenting the brake pad image to obtain a binary image according to the optimal segmentation parameter selected after judging the segmentation parameters through a PCNN network; and a brake pad thickness obtaining step of obtaining an edge contour line according to the binary image and reversely mapping the edge contour line to 3D point cloud data to obtain the brake pad thickness. The present invention uses technologies such as machine vision to collect 3D images of the brake pads of rail vehicles for measurement, assisting the inspection work of quality inspection personnel, reducing the manual work intensity, and improving the maintenance quality and efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent inspection of rail trains, and more particularly, to a method and a system for measuring brake pads of rail vehicles. Background Art

[0002] The braking device of a rail vehicle is an important component of the vehicle, and the wear condition of the brake pads is directly related to the safety of the vehicle operation. Therefore, after the rail vehicle has run for a certain period of time, it is necessary to accurately measure the thickness of the brake pads and replace the brake pads with a thickness less than the standard value to ensure compliance with the safe operation of the rail vehicle.

[0003] Currently, the inspection of the brake pad thickness mainly relies on manual inspection. Manual inspection has the following risks: (1) Operators are prone to fatigue, and the inspection quality overly depends on the responsibility of the maintenance personnel, with risks of missed inspection and misjudgment; (2) The efficiency of manual inspection is low, the inspection time is long, which delays the other maintenance processes of the rail vehicle and there is a backlog phenomenon; (3) Lack of informatization support, it is inconvenient to store and retrieve inspection records and common problem information, which is not conducive to statistics and traceability, cannot be remotely retrieved, and is not convenient for problem troubleshooting and supervision and management.

[0004] Therefore, there is an urgent need to develop a method and a system for measuring brake pads of rail vehicles that overcome the above defects. Summary of the Invention

[0005] In view of the above problems, the present invention provides a method for measuring brake pads of rail vehicles, which includes:

[0006] Image acquisition step: acquiring an image of the brake pads of the rail vehicle;

[0007] Segmentation parameter acquisition step: obtaining segmentation parameters based on the STOA-DE algorithm according to the brake pad image;

[0008] Binary image acquisition step: after judging the segmentation parameters through the PCNN network and selecting the optimal segmentation parameters, segmenting the brake pad image according to the optimal segmentation parameters to obtain a binary image;

[0009] Brake pad thickness acquisition step: obtaining an edge contour line according to the binary image, and reversely mapping the edge contour line to 3D point cloud data to obtain the brake pad thickness.

[0010] The above measurement method, wherein the binary image acquisition step includes:

[0011] Fitness function value acquisition step: inputting the segmentation parameters into the PCNN network for iteration to obtain a fitness function value;

[0012] Fitness function value judgment step: Compare the fitness function value with the standard fitness function value;

[0013] Optimal segmentation parameter obtaining step: When the fitness function value meets the segmentation effect, determine the segmentation parameter corresponding to the fitness function value as the optimal segmentation parameter, otherwise return to the segmentation parameter obtaining step;

[0014] Image segmentation step: Segment the brake pad image through the PCNN network according to the optimal segmentation parameter to obtain the binary image.

[0015] The above measurement method, wherein the segmentation parameter obtaining step includes:

[0016] Parameter setting step: Assign values to the initial parameters of the STOA-DE algorithm and set the search range of the segmentation parameters;

[0017] Global search step: Obtain the position update trajectory and random parameters towards the optimum according to the search range through the Sooty tern optimization algorithm;

[0018] Local search step: Judge the random parameters, select a local search strategy according to the judgment result, and obtain the segmentation parameter of the optimal position through the local search strategy according to the position update trajectory.

[0019] The above measurement method, wherein the brake pad thickness obtaining step includes:

[0020] Edge feature obtaining step: After performing Gaussian filtering on the binary image, extract edge features using gradient changes;

[0021] Edge feature processing step: Fit a straight line through Hough transform according to the edge features to obtain the edge contour line and the maximum circumscribed rectangle region;

[0022] Brake pad thickness calculation step: Determine the coordinates of the central pixel point according to the maximum circumscribed rectangle region, extend from the central pixel point to the edge contour line respectively, traverse each pixel point step by step, search for the coordinates of two pixel points corresponding to the edge contour line, and then reverse map to the 3D point cloud data to calculate the distance in the width direction of the brake pad.

[0023] The above measurement method, wherein the fitness function value obtaining step calculates the fitness function value using image entropy.

[0024] The present invention also provides a measurement system for a brake pad of a rail vehicle, wherein it includes:

[0025] An image acquisition unit for acquiring a brake pad image of a rail vehicle;

[0026] A segmentation parameter acquisition unit that acquires segmentation parameters based on the brake pad image using the STOA-DE algorithm;

[0027] A binary image acquisition unit that, after judging the segmentation parameters through a PCNN network and selecting the optimal segmentation parameters, segments the brake pad image according to the optimal segmentation parameters to obtain a binary image;

[0028] A brake pad thickness acquisition unit that obtains an edge contour line based on the binary image and reversely maps the edge contour line to 3D point cloud data to obtain the brake pad thickness.

[0029] The above measurement system, wherein the binary image acquisition unit includes:

[0030] A fitness function value acquisition module that inputs the segmentation parameters into the PCNN network for iteration to obtain a fitness function value;

[0031] A fitness function value judgment module that compares the fitness function value with a standard fitness function value;

[0032] An optimal segmentation parameter acquisition module that, when the fitness function value meets the segmentation effect, determines the segmentation parameter corresponding to the fitness function value as the optimal segmentation parameter, otherwise returns to the segmentation parameter acquisition step;

[0033] An image segmentation module that segments the brake pad image according to the optimal segmentation parameters through the PCNN network to obtain the binary image.

[0034] The above measurement system, wherein the segmentation parameter acquisition unit includes:

[0035] A parameter setting module that assigns initial parameters to the STOA-DE algorithm and sets the search range of the segmentation parameters;

[0036] A global search module that obtains an orientation-optimal position update trajectory and random parameters through the Sooty Tern optimization algorithm according to the search range;

[0037] A local search module that judges the random parameters, selects a local search strategy according to the judgment result, and obtains the segmentation parameters of the optimal position through the local search strategy according to the position update trajectory.

[0038] The above measurement system, wherein the brake pad thickness acquisition unit includes:

[0039] An edge feature acquisition module that, after performing Gaussian filtering on the binary image, extracts edge features using gradient changes;

[0040] An edge feature processing module obtains the edge contour line and the maximum circumscribed rectangle area by fitting a straight line through Hough transform according to the edge feature;

[0041] A brake pad thickness calculation module determines the coordinates of the central pixel point according to the maximum circumscribed rectangle area, extends from the central pixel point to the edge contour line respectively, traverses each pixel point step by step, and after searching for the coordinates of two pixel points corresponding to the edge contour line, maps them back to the 3D point cloud data to calculate the distance in the width direction of the brake pad.

[0042] The above measurement system, wherein the fitness function value obtaining module calculates the fitness function value by using image entropy.

[0043] The efficacy of the present invention compared with the prior art lies in that the present invention uses technologies such as machine vision to collect 3D images of the brake pads of rail vehicles for measurement, assisting the inspection work of quality inspection personnel, reducing the manual work intensity, improving the overhaul quality and efficiency; at the same time, through the information management platform, the brake pad detection results are recorded in the form of images and data, forming a database that can be statistically analyzed and traced, facilitating problem troubleshooting and supervision management.

[0044] Other features and advantages of the present invention will be described in the following specification, and part of them will be obvious from the specification, or understood by implementing the present invention. The objectives and other advantages of the present invention can be realized and obtained by the structures pointed out in the specification, claims and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0046] Figure 1 It is a flowchart of the measurement method of the present invention;

[0047] Figure 2 is Figure 1 a sub-step flowchart of step S2 in

[0048] Figure 3 is Figure 1 a sub-step flowchart of step S3 in

[0049] Figure 4 is Figure 1 a sub-step flowchart of step S4 in

[0050] Figure 5It is the application flowchart of the measurement method of the present invention;

[0051] Figure 6 It is the structural schematic diagram of the measurement system of the present invention;

[0052] Figure 7 It is the application flowchart of the STOA-DE algorithm of the present invention;

[0053] Figure 8 It is the effect diagram of the brake pad measurement of the present invention. Detailed implementation manners

[0054] To make the objectives, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Apparently, the described embodiments are some but not all of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0055] The schematic embodiments of the present invention and the descriptions thereof are used to explain the present invention, but not to limit the present invention. In addition, the same or similar reference numerals of elements / components used in the drawings and embodiments are used to represent the same or similar parts.

[0056] Regarding the "first", "second", "S1", "S2",... used herein, they do not particularly refer to the meaning of order or sequence, nor are they used to limit the present invention. They are only used to distinguish the elements or operations described with the same technical terms.

[0057] Regarding the "including", "comprising", "having", "containing", etc. used herein, they are all open-ended terms, that is, they mean including but not limited to.

[0058] Regarding the "multiple" herein, it includes "two" and "more than two"; regarding the "multiple groups" herein, it includes "two groups" and "more than two groups".

[0059] Refer to Figure 1 and Figure 5 , Figure 1 It is the flowchart of the measurement method of the present invention; Figure 5 It is the application flowchart of the measurement method of the present invention. As Figure 1 and Figure 5 shown, a measurement method for the brake pads of rail vehicles of the present invention includes:

[0060] Image acquisition step S1: Acquire the image of the brake pads of the rail vehicle;

[0061] Segmentation parameter acquisition step S2: Obtain segmentation parameters based on the brake pad image using the STOA-DE algorithm;

[0062] Binary image acquisition step S3: After judging the segmentation parameters through the PCNN network and selecting the optimal segmentation parameters, segment the brake pad image according to the optimal segmentation parameters to obtain a binary image;

[0063] Brake pad thickness acquisition step S4: Obtain the edge contour line according to the binary image, and reversely map the edge contour line to the 3D point cloud data to obtain the brake pad thickness.

[0064] Please refer to Figure 2 , Figure 2 for Figure 1 the sub-step flowchart of step S2 in Figure 2 As shown, the segmentation parameter acquisition step S2 includes:

[0065] Parameter setting step S21: Assign values to the initial parameters of the STOA-DE algorithm and set the search range of the segmentation parameters;

[0066] Global search step S22: Obtain the position update trajectory and random parameters with the best orientation through the Sooty Tern Optimization Algorithm according to the search range;

[0067] Local search step S23: Judge the random parameters, select a local search strategy according to the judgment result, and obtain the segmentation parameters of the optimal position through the local search strategy according to the position update trajectory.

[0068] Among them, the Sooty Tern Optimization Algorithm (STOA) is a new optimization algorithm proposed by G. Dhiman and A. Kaur in 2019 for industrial engineering problems. Its inspiration comes from the foraging behavior of seabirds in nature. The sooty tern is an omnivorous bird that feeds on earthworms, insects, fish, etc. This algorithm has strong global search ability and high accuracy. However, it still has some problems such as the imbalance between exploration and exploitation and the low population diversity in the later stage of iteration, resulting in the premature convergence of the algorithm. At the same time, this also promotes the research work on improving the optimization algorithm, enabling the improved algorithm to be applied to more optimization problems.

[0069] Specifically, the Sooty Tern Optimization Algorithm includes migration behavior (global exploration) and attack behavior (local search). Among them, the migration behavior, that is, the exploration part, is mainly divided into three parts: conflict avoidance, aggregation, and update.

[0070] (1) Conflict avoidance:

[0071] C st = S A×P st |(Z) (7);

[0072] Among them, P st represents the current position of the Sooty Tern, and C st represents the position that should be in without colliding with other Sooty Terns, and S A represents a variable factor for collision avoidance, used to calculate the position after collision avoidance, and its constraint condition is as shown in formula (8).

[0073] S A = C f - (Z × (C f / Max iterations )) (8);

[0074] Z = 0, 1, 2,..., Max iterations (9);

[0075] Among them, C f is a control variable used to adjust S A , Z represents the current iteration number, so S A linearly decreases from C f to 0. In the present invention, the value of C f is set to 2. Therefore, S A will gradually decrease from 2 to 0.

[0076] (2) Aggregation:

[0077] Aggregation means approaching the best position among adjacent Sooty Terns on the premise of avoiding collisions, that is, approaching the position of the optimal solution, and its mathematical expression is as follows:

[0078] M st = C B × (P bst (Z) - P st (Z)) (10);

[0079] Among them, M st represents the process of P st at different positions moving towards the position of the optimal solution P bst , and C B is a random variable that makes the exploration more comprehensive and changes according to the following formula:

[0080] C B = 0.5 × R and (11);

[0081] Among them, R and is a random number between 0 and 1.

[0082] (3) Update:

[0083] Update refers to updating the trajectory towards the optimal solution, and its trajectory D st The mathematical expression is:

[0084] D st = C st + M st (12).

[0085] Among them, the attack behavior (local search):

[0086] During the migration process, the sooty tern can increase its flight altitude by flapping its wings, adjust its own speed and attack angle. When attacking prey, their hovering behavior in the air can be defined by the following mathematical model:

[0087] x′ = R adius ×sin(i) (13)

[0088] y′ = R adius ×cos(i) (14)

[0089] z′ = R adius ×i (15)

[0090] R adius = u×e kv (16)

[0091] Among them, R adius represents the radius of each helix, and i represents a variable between [0, 2π]. u and v are constants defining its helix shape, both set to 1 in this article, and e is the base of the natural logarithm. The STOA-DE algorithm of the present invention continuously updates the position according to the following formula:

[0092]

[0093] Among them, q is a random parameter. In this embodiment, when q is greater than or equal to 0.5, the position is updated by formula (17), and when q is less than 0.5, the position is updated by formula (18). Thus, the search efficiency and search accuracy are improved, so as to achieve a better segmentation effect and segmentation accuracy for the brake pads.

[0094] Among them, F is a proportionality coefficient, taken as 0.4 in the present invention, r 1 , r 2 , r 3 , r 4 are different integers within the range of [1, N], and N represents the population size.

[0095] The following is an explanation of the PCNN network. The mathematical expression of the simplified PCNN network model of the present invention is:

[0096] Fij I(n) = S ij (1)

[0097] L ij I(n) = ∑W ijkl Y kl I(n - 1) (2)

[0098] U ij I(n) = F ij (1 + βL ij I(n)) (3)

[0099]

[0100]

[0101] Where: F ij I(n) represents the input of the PCNN; S ij is the external input, such as all pixel points of an image; L ij I(n) is the link input; U ij I(n) is the internal activity item, θ ij I(n) represents the dynamic threshold, Y ij I(n) is the output of the neural network; β is the link strength coefficient, W ijkl is the link matrix; α θ is the threshold decay coefficient, V θ is the threshold amplification coefficient. Usually, the calculation of W ijkl is divided into two steps. First, calculate the Euclidean distance between adjacent neurons, and then take the reciprocal of the obtained value. Generally, it can be set as:

[0102]

[0103] From formula (1) and formula (2), it can be obtained that the input of the simplified feedback channel removes the excitation of neighboring neurons and only retains the external input excitation. The input value of the link channel only considers the excitation of neighboring neurons. The simplified PCNN model has four parameters: W ijkl , V θ , α θ and β. Among these parameters, three mainly have a greater impact on the segmentation result: the link strength coefficient β, the threshold decay coefficient α θ , the threshold amplification coefficient V θ . Specifically, the role of the threshold amplification coefficient V θ is to make the dynamic threshold θ ij of the neuron that has just fired jump to a relatively large value, so that the neuron cannot fire again within a short time. The role of the decay coefficient α θ is to make the dynamic threshold θ of the neuron that has just completed firingij gradually decreases, and when it is less than the internal activity item U ij (n), the neuron fires again. The link strength coefficient β reflects the ability of the central neuron to influence the neighboring neurons. The larger the β value, the greater the possibility that the neighboring neurons are influenced by the central neuron. From the segmentation result, the image edge will be more complete. Similarly, the smaller the β value, the smaller the chance that the neighboring neurons are captured, and the details of the obtained segmentation result will be relatively richer.

[0104] Please refer to Figure 7 , Figure 7 which is the application flow chart of the STOA-DE algorithm of the present invention. The following combines Figure 7 to specifically illustrate the application process of the STOA-DE algorithm of the present invention. First, select the initial parameters: assign values to the initial parameters of the STOA-DE algorithm, and set the search ranges of the segmentation parameters β, V E , α E to be 0.001 to 400, where the initial parameters are C f , u, v; secondly, after calculating the individual fitness value and the population average fitness value; obtain the position update trajectory D st towards the optimal according to the search range through the global exploration of the sooty tern optimization algorithm (Formula 12); then according to the position update trajectory D st perform local optimization through the mutation operator of the differential evolution algorithm (Formula 18) to obtain the segmentation parameters β, V E , α E of the optimal position.

[0105] Please refer to Figure 3 , Figure 3 Figure 3 is Figure 1 the sub-step flow chart of step S3 in Figure 3 . As

[0106] shown, the binarized image obtaining step S3 includes:

[0107] Fitness function value obtaining step S31: input the segmentation parameters into the PCNN network for iteration to obtain the fitness function value;

[0107] Fitness function value judgment step S32: compare the fitness function value with the standard fitness function value;

[0108] Optimal segmentation parameter obtaining step S33: when the fitness function value meets the segmentation effect, determine the segmentation parameter corresponding to the fitness function value as the optimal segmentation parameter, otherwise return to the segmentation parameter obtaining step;

[0109] Image segmentation step S34: The PCNN network segments the brake pad image according to the optimal segmentation parameters to obtain the binary image.

[0110] Specifically, first, the segmentation parameters β, V E , α E obtained by the STOA-DE algorithm are input into the PCNN network; after PCNN iteration, the fitness function value is obtained; secondly, the fitness function value is compared with the standard fitness function value. If the fitness function value meets the segmentation requirement, the segmentation parameters β, V E , α E corresponding to the fitness function value are set as the optimal segmentation parameters. If the fitness function value does not meet the segmentation requirement, return to step S21 to obtain a new set of segmentation parameters β, V E , α E ; then, the new set of segmentation parameters β, V E , α E are input into the PCNN to obtain the fitness function value, which is compared with the standard fitness function value again to determine the optimal segmentation parameters; then, when the number of algorithm iterations reaches the maximum number of iterations, the global optimal fitness function value and the optimal parameter value are output. Step 5: Input the optimal segmentation parameters into the PCNN to segment the brake pad image.

[0111] Among them, when the optimization algorithm optimizes the parameters, the fitness function must be set in advance as a standard to find the optimal parameters that can reach or approach the function value. In the present invention, entropy is selected as the fitness function. The basic idea is to search for the parameter value that maximizes the sum of the entropies of the target and background regions based on the gray value distributions of the target and the background. Calculate the entropy value of the segmentation result. The larger this value is, the more original image information is retained in the segmentation result and the better the segmentation effect. The calculation formula of entropy is as follows:

[0112] H = -P 0 log 2 P 0 -P 1 log 2 P 1

[0113] When using PCNN to segment an image, since the output is a pulse sequence, the result is a binary image containing only 0 and 1; P 0 and P 1 respectively represent the ratios of the pixel points with gray values of 0 and 1 in the output image to the entire image; H represents the image entropy.

[0114] For PCNN, the parameters that need to be optimized are the link strength coefficient β, the threshold amplification coefficient V E , and the threshold decay coefficient α E, because these three parameters have the greatest impact on the segmentation result.

[0115] Please refer to Figure 4 , Figure 4 for Figure 1 the sub-step flowchart of step S4 in Figure 4 As shown, the brake pad thickness obtaining step S4 includes:

[0116] Edge feature obtaining step S41: After performing Gaussian filtering on the binary image, extract edge features using gradient changes;

[0117] Edge feature processing step S42: Obtain the edge contour line and the maximum circumscribed rectangle region by fitting a straight line through Hough transform according to the edge features;

[0118] Brake pad thickness calculation step S43: Determine the coordinates of the central pixel point according to the maximum circumscribed rectangle region, extend from the central pixel point to the edge contour line respectively, traverse each pixel point step by step, search for the coordinates of two pixel points corresponding to the edge contour line, and then reverse map to the 3D point cloud data to calculate the distance in the width direction of the brake pad.

[0119] Specifically, in step S4, the segmented binary image is subjected to Gaussian filtering, the edge contour line is extracted using gradient changes, and a straight line is fitted by Hough transform. At the same time, the maximum circumscribed rectangle region is obtained, the central pixel coordinates are obtained, starting from this point, extend to the contour edge respectively, traverse each point step by step, search for the coordinates of two pixels on the corresponding boundary, reverse map to the 3D point cloud data, and calculate the distance in the width direction of the brake pad, which is the brake pad thickness. Figure 8 Shows the effect diagrams of each step of brake pad measurement. Among them, (a) is the original image of the brake pad; (b) is the segmentation result of the brake pad; (c) is the brake pad contour; (d) is the brake pad contour mapped to the 3D image; (e) is the final measurement result.

[0120] Please refer to Figure 6 , Figure 6 is the structural schematic diagram of the measurement system of the present invention. As Figure 6 shown, a measurement system for rail vehicle brake pads of the present invention includes:

[0121] An image acquisition unit 11, which acquires the brake pad image of the rail vehicle;

[0122] A segmentation parameter obtaining unit 12, which obtains segmentation parameters based on the STOA-DE algorithm according to the brake pad image;

[0123] The binary image acquisition unit 13 judges the segmentation parameters through the PCNN network, selects the optimal segmentation parameters, and then segments the brake pad image according to the optimal segmentation parameters to obtain a binary image;

[0124] The brake pad thickness acquisition unit 14 obtains the edge contour line according to the binary image, and reversely maps the edge contour line to the 3D point cloud data to obtain the brake pad thickness.

[0125] Among them, the image acquisition unit 11 includes a movable mechanism and a 3D camera mounted on the movable mechanism. After driving the 3D camera to the acquisition position through the movable mechanism, image acquisition is performed through the 3D camera. The 3D camera can receive signals through network communication and transmit three-dimensional images. The present invention uses a 3D camera, combined with an improved image segmentation algorithm, to perform automated image acquisition and processing, which can improve the measurement accuracy to 0.02 mm, reduce the manual workload, and improve the automation level and quality of detection.

[0126] Further, the segmentation parameter acquisition unit 12 includes:

[0127] The parameter setting module 121 assigns initial parameters to the STOA-DE algorithm and sets the search range of the segmentation parameters;

[0128] The global search module 122 obtains the position update trajectory towards the optimum according to the search range through the sooty tern optimization algorithm;

[0129] The local search module 123 obtains the segmentation parameters at the optimal position through the differential evolution algorithm according to the position update trajectory.

[0130] Still further, the binary image acquisition unit 13 includes:

[0131] The fitness function value acquisition module 131 inputs the segmentation parameters into the PCNN network for iteration to obtain the fitness function value;

[0132] The fitness function value judgment module 132 compares the fitness function value with the standard fitness function value;

[0133] The optimal segmentation parameter acquisition module 133 determines the segmentation parameter corresponding to the fitness function value as the optimal segmentation parameter when the fitness function value satisfies the segmentation effect, otherwise returns to the segmentation parameter acquisition step;

[0134] The image segmentation module 134 segments the brake pad image according to the optimal segmentation parameters through the PCNN network to obtain the binary image.

[0135] Among them, the fitness function value obtaining module 131 calculates the fitness function value by using image entropy.

[0136] Furthermore, the brake pad thickness obtaining unit 14 includes:

[0137] An edge feature obtaining module 141, after performing Gaussian filtering on the binary image, extracts edge features by using gradient changes;

[0138] An edge feature processing module 142, obtains the edge contour line and the maximum circumscribed rectangle area by performing straight line fitting through Hough transform according to the edge features;

[0139] A brake pad thickness calculation module 143 determines the coordinates of the central pixel point according to the maximum circumscribed rectangle area, extends from the central pixel point to the edge contour line respectively, traverses each pixel point step by step, searches for the coordinates of two pixel points corresponding to the edge contour line, and then inversely maps them to the 3D point cloud data to calculate the distance in the width direction of the brake pad.

[0140] In summary, the automated image acquisition technology of the present invention can replace manual measurement, save labor costs, and improve work efficiency; at the same time, automated measurement can save the images of train brake pads in a computer as vouchers for maintenance. In addition, the automated measurement technology of brake pads can accurately measure the brake pads of EMUs, with an accuracy of up to 0.1 mm. Traditional manual measurement requires workers to hold a ruler to measure, and the accuracy is only 5 mm. Higher accuracy can reduce unnecessary replacements and improve economic benefits.

[0141] Although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A measuring method for brake pads of rail vehicles, characterized in that, it includes: Image acquisition step: Acquire the image of the brake pads of the rail vehicle; Segmentation parameter acquisition step: Obtain the segmentation parameters based on the STOA-DE algorithm according to the brake pad image; Binary image acquisition step: After judging the segmentation parameters through the PCNN network and selecting the optimal segmentation parameters, segment the brake pad image according to the optimal segmentation parameters to obtain a binary image; Brake pad thickness acquisition step: Obtain the edge contour line according to the binary image, and reverse map the edge contour line to the 3D point cloud data to obtain the brake pad thickness; Among them, the binary image acquisition step includes: Fitness function value acquisition step: Input the segmentation parameters into the PCNN network for iteration to obtain the fitness function value; Fitness function value judgment step: Compare the fitness function value with the standard fitness function value; Optimal segmentation parameter acquisition step: When the fitness function value meets the segmentation effect, determine the segmentation parameter corresponding to the fitness function value as the optimal segmentation parameter, otherwise return to the segmentation parameter acquisition step; Image segmentation step: Segment the brake pad image through the PCNN network according to the optimal segmentation parameters to obtain the binary image; Among them, the segmentation parameter acquisition step includes: Parameter setting step: Assign values to the initial parameters of the STOA-DE algorithm and set the search range of the segmentation parameters; Global search step: Obtain the position update trajectory and random parameters towards the optimum through the Sooty tern optimization algorithm according to the search range; Local search step: Judge the random parameters, select the local search strategy according to the judgment result, and obtain the segmentation parameter of the optimal position through the local search strategy according to the position update trajectory.

2. The measuring method according to claim 1, characterized in that, the brake pad thickness acquisition step includes: Edge feature acquisition step: After performing Gaussian filtering on the binary image, extract edge features using gradient changes; Edge feature processing step: Fit a straight line through the Hough transform according to the edge features to obtain the edge contour line and the maximum circumscribed rectangle area; Brake pad thickness calculation step: Determine the coordinates of the central pixel point according to the maximum circumscribed rectangle area, extend from the central pixel point to the edge contour line respectively, traverse each pixel point step by step, search for the coordinates of two pixel points corresponding to the edge contour line, and then reverse map to the 3D point cloud data to calculate the distance in the width direction of the brake pad.

3. The measuring method according to claim 1, characterized in that, the fitness function value acquisition step calculates the fitness function value using image entropy.

4. A measuring system for brake pads of rail vehicles, characterized in that, it includes: An image acquisition unit that acquires the image of the brake pads of the rail vehicle; A segmentation parameter acquisition unit that obtains the segmentation parameters based on the STOA-DE algorithm according to the brake pad image; A binary image acquisition unit, after judging the segmentation parameters through a PCNN network and selecting the optimal segmentation parameters, segments the brake pad image according to the optimal segmentation parameters to obtain a binary image; A brake pad thickness acquisition unit, obtains an edge contour line according to the binary image, and reversely maps the edge contour line to 3D point cloud data to obtain the brake pad thickness; Wherein, the binary image acquisition unit includes: A fitness function value acquisition module, inputs the segmentation parameters into the PCNN network for iteration to obtain a fitness function value; A fitness function value judgment module, compares the fitness function value with a standard fitness function value; An optimal segmentation parameter acquisition module, when the fitness function value meets the segmentation effect, determines the segmentation parameter corresponding to the fitness function value as the optimal segmentation parameter, otherwise returns to the segmentation parameter acquisition step; An image segmentation module, segments the brake pad image according to the optimal segmentation parameters through the PCNN network to obtain the binary image; Wherein, the segmentation parameter acquisition unit includes: A parameter setting module, assigns initial parameters of the STOA-DE algorithm and sets the search range of the segmentation parameters; A global search module, obtains an optimal position update trajectory and random parameters through the Sooty Tern optimization algorithm according to the search range; A local search module, judges the random parameters, selects a local search strategy according to the judgment result, and obtains the segmentation parameter of the optimal position through the local search strategy according to the position update trajectory.

5. The measurement system according to claim 4, characterized in that the brake pad thickness acquisition unit includes: An edge feature acquisition module, after performing Gaussian filtering on the binary image, extracts edge features by using gradient changes; An edge feature processing module, fits a straight line through Hough transform according to the edge features to obtain the edge contour line and the maximum circumscribed rectangle area; A brake pad thickness calculation module, determines the coordinates of the central pixel point according to the maximum circumscribed rectangle area, extends from the central pixel point to the edge contour line respectively, traverses each pixel point step by step, searches for the coordinates of two pixel points corresponding to the edge contour line, and reversely maps them to the 3D point cloud data to calculate the distance in the width direction of the brake pad.

6. The measurement system according to claim 4, characterized in that the fitness function value acquisition module calculates the fitness function value by using image entropy.

Images