A machine vision-based rod bending stiffness distribution measurement system

Abstract

The application discloses a kind of based on machine vision's rod piece bending stiffness distribution measurement system, it is related to material mechanics parameter measurement technical field, including: on camera measurement platform, rod piece is loaded, and the image before and after loading is collected;The image collected is cropped and gray image conversion operation is carried out, and according to noise level and deflection interval carries out image availability evaluation;The discrete edge point set is obtained by applying improved subpixel edge detection technology, and the edge point pair is screened out, pixel coordinates are converted into actual coordinates, the edge coordinate difference value before and after loading is calculated, and then deflection distribution is calculated;Using Tikhonov regularization, generalized cross-validation method is used to calculate the second derivative of discrete deflection data, and the bending moment distribution function formula is established based on cantilever beam model;Finally, combined with Euler-Bernoulli beam differential equation, space stiffness inversion is carried out.The application effectively and accurately obtains the stiffness field distribution of rod piece, and provides a new technical path for the research and engineering application of various variable stiffness structures.

Description

Technical Field

[0001] This invention relates to the field of material mechanical parameter measurement technology, and in particular to a machine vision-based system for measuring the bending stiffness distribution of rods, the main parameter of which is the bending stiffness distribution of the rods. Background Technology

[0002] Bending stiffness is a core mechanical parameter of slender structural members, and its accurate measurement is of great significance for the safety assessment and optimization design of engineering structures. Its spatial distribution characteristics directly determine the safety performance and service life of engineering structures. It plays a key role in composite material structure design, biomimetic component optimization, and flexible mechanical performance assessment, such as carbon fiber composite beams in the aerospace field and flexible joint components of biomimetic robots.

[0003] Existing methods for measuring variable stiffness are largely limited in that they can only obtain the global average stiffness of a structure, failing to effectively capture the spatial gradient distribution of stiffness. Whether it's traditional contact methods like tension and bending, or non-contact techniques such as vibration analysis and ultrasound, all are constrained by their fundamental principles, making it difficult to achieve high-resolution in-situ gradient identification. While some emerging optical methods have improved accuracy, they still haven't established an inversion model from the deformation field to the stiffness distribution. Therefore, developing new methods that combine gradient identification with both in-situ and non-contact measurement capabilities remains a key challenge.

[0004] This invention proposes a machine vision-based system for measuring the bending stiffness distribution of rods. By introducing subpixel edge detection and Tikhonov regularization, it utilizes high-resolution imaging equipment to collect minute deformations on the target surface, achieving non-contact, sampling-free stiffness characteristic identification. This enables efficient and accurate acquisition of the stiffness field distribution of rods, providing a novel technical approach for the research and engineering applications of various variable stiffness structures. Summary of the Invention

[0005] The measurement system used in this invention includes a camera system, a lighting system, a computer system, a clamping system, and the rod to be measured, characterized in that:

[0006] The camera system includes a camera, a camera support frame, camera accessories, and a data transmission cable. Its main feature is that it can capture clear images with a resolution of not less than 1024*512.

[0007] The lighting system, including fill lights, stands, and softboxes, is characterized by fill lights and stands providing a shooting environment with a light intensity of no less than 5000 lux, and softboxes providing a high-contrast background.

[0008] The computer system, including a computer and a data connection line, is characterized by reading image information and running a measurement method, transmitting images captured by a camera to the computer, processing edge data, and performing member stiffness inversion.

[0009] The clamping system includes a three-jaw clamp and a weight of known weight. Its main feature is that the two ends of the object to be measured are defined as a fixed end and a free end, respectively. One end of the object is clamped by the three-jaw clamp and is subject to fixed constraints; the other end is a free end, and a metal ball is suspended from the free end by a thin rope to apply a concentrated load.

[0010] The main characteristics of the rod under test are that its geometry conforms to the Euler-Bernoulli beam theory in mechanics of materials, and it satisfies the geometric constraint of a length-to-thickness ratio ≥300 to ensure negligible shear deformation. It is fixed using a cantilever beam method, with one end rigidly fixed by a three-jaw clamp and the other end free, extended by 20% as a loading extension. Its deflection function satisfies the following differential equation:

[0011] Formula 1

[0012] in, Indicates the position coordinates along the length of the beam. Let be the deflection function of the beam. Let x be the bending moment distribution function along the x-direction. Let be the bending stiffness function distributed along the length of the beam.

[0013] The specific technical solution adopted by this invention to solve its technical problem is as follows:

[0014] A machine vision-based system for measuring the bending stiffness distribution of a rod, characterized by comprising the following steps:

[0015] Step S1: Calibrate the perpendicular-coplanar relationship between the rod and the camera spindle;

[0016] Step S2: Image acquisition and preprocessing of the rod;

[0017] Step S3: Extract the edge of the rod;

[0018] Step S4: Stiffness inversion of the members;

[0019] Step S5: Stiffness result optimization.

[0020] A further technical solution is that step S1 includes the following steps:

[0021] Define the central axis of the rod along its length as the horizontal x-axis; define the central optical axis of the camera as the main axis of the camera as the horizontal z-axis; adjust the lens angle through the support frame so that the spatial angle between the z-axis and the x-axis is 90°, and the angle deviation between the z-axis and the x-axis is ≤0.5°; adjust the vertical distance between the x-axis and the z-axis and the horizontal platform surface to be consistent, and the vertical distance deviation with the horizontal platform surface is ≤0.1mm.

[0022] Furthermore, step S2 includes the following sub-steps:

[0023] Step S2-1: Install the rod to be measured, perform spatial relationship calibration, and record the initial image before loading;

[0024] Step S2-2: Apply a concentrated load to the free end of the rod to cause deformation, and record the image after loading;

[0025] Step S2-3: Crop the image sequence and convert it to grayscale images;

[0026] Step S2-4: Image usability assessment;

[0027] Furthermore, in step S2-1, the acquisition range covers the entire length of the rod;

[0028] Furthermore, in step S2-2, a small metal ball is used as a weight and suspended from the free end of the rod with a thin thread to apply a concentrated load; each time a shot is taken, the recording is performed only after the rod has stabilized under the load.

[0029] Furthermore, in steps S2-3, the effective area centered on the rod is retained, while background interference areas such as clamps and loads are removed;

[0030] Furthermore, in steps S2-4, the standard deviation of image background grayscale fluctuation is introduced as a noise level estimation index to assess the noise level; the ratio of maximum deflection to rod length is defined. As a measure of deflection range.

[0031] Step S3 includes the following sub-steps:

[0032] Step S3-1: Apply sub-pixel edge detection technology to obtain a discrete set of edge points;

[0033] Step S3-2: Establish screening criteria based on the geometric characteristics of the rods and screen edge point pairs;

[0034] Step S3-3: Divide the selected edge points into two groups, upper and lower, and select one side to convert the pixel coordinates into actual coordinates;

[0035] Step S3-4: Calculate the difference between the actual edge coordinates after loading and the initial state, and calculate the deflection distribution;

[0036] Furthermore, in step S3-1, the sub-pixel edge detection algorithm based on local area effect proposed by Trujillo-Pino et al. is used to extract the coordinates of the edge points;

[0037] Furthermore, in step S3-2, given that the rod is a slender structure, its two boundary points are usually paired and the spacing is within a fixed range, and the normal vectors are approximately collinear in opposite directions, thus establishing a geometric constraint criterion.

[0038] Step S4 includes the following sub-steps:

[0039] Step S4-1: Take the second derivative of the discrete deflection data to invert the stiffness;

[0040] Step S4-2: Calculate the bending moment distribution based on the cantilever beam model;

[0041] Step S4-3: Perform spatial stiffness inversion using Euler-Bernoulli beam theory;

[0042] Further, in step S4-1, the deflection coordinates are fitted using the least squares method with Tikhonov regularization term using Taylor basis functions;

[0043] Further, in step S4-1, the regularization factor is determined using generalized cross-validation (GCV);

[0044] Furthermore, in step S4-1, the second derivative of the fitted basis function is taken and regarded as the second derivative of the deflection;

[0045] Furthermore, in step S4-2, a formula for the bending moment distribution function is established based on the cantilever beam mechanical model.

[0046] Step S5: Discard the data of the loaded extension section to avoid numerical divergence problems caused by the bending moment at the beam end and the second derivative approaching 0.

[0047] A further technical solution involves noting the following points when using this method:

[0048] When using a camera to record, it is important to ensure the integrity and stability of the recorded images and reduce surrounding interference factors;

[0049] When recognizing images of rods, pay attention to the settings of the algorithm parameters to ensure that the recognized images are not distorted;

[0050] The order of the expansion of the basis functions after moment reconstruction needs to be set appropriately to avoid distortion of the results;

[0051] This method can effectively measure the spatial stiffness distribution of slender beam structures. It enables rapid identification of bending stiffness distribution without the need for traditional mechanical sensors and complex loading devices, and supports in-situ non-destructive measurement. This provides a new method for stiffness detection in composite material structures, flexible mechanical components, and other scenarios. Attached Figure Description

[0052] To more clearly illustrate the implementation method of the present invention and the existing technical solutions, the specific implementation method of the present invention will be described in detail below with reference to the accompanying drawings.

[0053] Figure 1 , Figure 2 This is a machine vision-based device for measuring the bending stiffness distribution of rods. Figure 1 middle: 1-Camera 2-Background fill light bracket 3-The support for the rod to be tested 4- Soft Light Screen 5-Background fill light 6-Camera accessories 7-Data connection cable 8-Computer 9-Data transmission cable Figure 2 middle: 1-Bar 2-Three-jaw clamp 3-Metal spheres 4- Thin rope 5-Lifting Platform

[0054] Figure 3 This is a flowchart of a machine vision-based method for measuring the bending stiffness distribution of a rod.

[0055] Figure 4 This is a schematic diagram of the improved sub-pixel edge detection algorithm;

[0056] Figure 5 This is a schematic diagram showing the coordinates of the member in its initial state and after loading. Detailed Implementation

[0057] To make the features and advantages of this patent more apparent and understandable, specific embodiments are provided below for detailed explanation. It should be noted that the following detailed descriptions are illustrative and intended to provide further explanation of this application. Unless otherwise specified, all technical and scientific terms used in this specification have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0058] The main measuring device used in this invention (its simplified layout is shown in the figure below) Figure 1 , Figure 2The system comprises: a camera system, including a camera 1-1, a camera support frame 1-4, high-speed camera accessories 1-6, and a data transmission cable 1-7, the main feature of which is that this system can be used to acquire clear images with a resolution of not less than 1024*512; a lighting system, including a background fill light 1-5, a soft light curtain 1-4, and a background fill light bracket 1-2, the main feature of which is to provide a shooting environment with a light intensity of not less than 5000 lux; a computer system, including a computer 1-8 and a data connection cable 1-9, the main feature of which is to read image information and run measurement methods, and to transmit images captured by the camera 1-1 to the computer 1-8; and a clamping system, including a three-jaw clamp 2-2, a rod 2-1, a thin rope 2-4, and a small metal ball 2-5, the main feature of which is to fix one end of the rod and apply a concentrated load to the other end.

[0059] This invention discloses a method (process as follows) Figure 3 (As shown), including the following steps:

[0060] Step S1: Align the perpendicular-coplanar spatial relationship between the rod axis and the camera main axis;

[0061] Step S2-1: Install the rod to be measured, perform spatial relationship calibration, and record the initial image before loading;

[0062] Step S2-2: Apply a concentrated load to the free end of the rod to cause deformation, and record the image after loading;

[0063] Step S2-3: Crop the image sequence and convert it to grayscale images;

[0064] Step S2-4: Image usability assessment. The images used for inversion should meet the following requirements: noise level ≤ 10, and ratio ≤ 10. ;

[0065] Step S3 includes the following sub-steps:

[0066] Step S3-1: Apply sub-pixel edge detection technology to obtain a discrete set of edge points;

[0067] Step S3-2: Establish screening criteria based on the geometric characteristics of the rods and screen edge point pairs;

[0068] Step S3-3: Divide the selected edge points into two groups, upper and lower, select one side, and convert the pixel coordinates into actual coordinates;

[0069] Step S3-4: Calculate the difference between the actual edge coordinates after loading and the initial state, and calculate the deflection distribution;

[0070] Step S4 includes the following sub-steps:

[0071] Step S4-1: Take the second derivative of the discrete deflection data;

[0072] Step S4-2: Calculate the bending moment distribution based on the cantilever beam model;

[0073] Step S4-3: Perform spatial stiffness inversion using Euler-Bernoulli beam theory;

[0074] Step S5: Discard the data of the loaded extension section to avoid numerical divergence problems caused by the bending moment at the beam end and the second derivative approaching 0.

[0075] Furthermore, in step S2-1, the rod is selected as a composite material rod, and the camera's acquisition range covers the entire length of the rod;

[0076] Furthermore, in step S2-2, metal balls are used as weights and suspended from the free end of the rod with thin lines to apply a concentrated load; the mass of a single metal ball is 4.15g, and the load is adjusted by changing the number of metal balls; each time a shot is taken, the recording is performed only after the rod has stabilized under the load.

[0077] Furthermore, in steps S2-3, the effective area centered on the rod is retained, while background interference areas such as clamps and loads are removed;

[0078] Furthermore, in step S3-1, the sub-pixel edge detection algorithm is improved. The improved principle is as follows: the image is read as a pixel grayscale matrix, the direction with the largest gradient in the image is set as the Y-axis, and a coordinate system is established with the origin as the center; the edge coordinates c in each pixel are estimated by taking the upper and lower three pixels, and the total weighted intensity formula of the three pixels is:

[0079] Formula 2

[0080] Furthermore, in the formula and Represents the grayscale value of the region. and They belong to and The area of ​​the region's three pixels; Let c represent the total area; then the formula for the edge coordinate c is:

[0081] Formula 3

[0082] Furthermore, in the formula It is estimated from the average gray value of pixels in the neighboring region outside the edge.

[0083] Furthermore, in step S3-2, given that the rod is a slender structure, its two boundary points are usually paired and the spacing between them is within a fixed range, and the normal vectors are approximately collinear in opposite directions, a dual screening criterion is established:

[0084] Formula 4

[0085] Furthermore, and For the extracted sub-pixel edge points, and The corresponding local external normal vectors at each time; This represents the maximum value of the coordinate distance corresponding to the beam, measured in pixels. and For the two calibrated thresholds, where Take 15%~20%, Take a value of 0.98 to 0.99;

[0086] Furthermore, in step S3-3, the formula for converting pixel coordinates to actual coordinates is as follows:

[0087] Formula 5

[0088] Furthermore, These are the edge coordinates at the pixel scale. These are the edge coordinates at a real-world scale. This is the proportionality coefficient;

[0089] Further, in step S4-1, the deflection coordinates are fitted using the Taylor basis function with least squares method and Tikhonov regularization term, and the calculation formula is as follows:

[0090] Formula 6

[0091] In formula 4, Denotes the Euclidean norm. The deflection vector. The coefficient matrix, As a regularization factor;

[0092] Further, in step S4-1, the generalized cross-validation (GCV) method is used to determine the regularization factor and minimize the validation function formula:

[0093] Formula 7

[0094] Furthermore, in Formula 5, It is a unit vector. The trace of the matrix;

[0095] Furthermore, in step S4-2, based on the cantilever beam mechanical model, a concentrated load is applied to the free end of the member. When, the formula for the bending moment distribution function is:

[0096] Formula 8

[0097] The present invention adopts the above technical solution and designs a machine vision-based rod bending stiffness distribution measurement system. The method has high measurement accuracy and non-contact in-situ detection capability. The device is simple to operate and can effectively measure the spatial stiffness distribution of variable stiffness structures.

[0098] It should be noted that, in this document, the terms “comprising,” “including,” or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0099] This article uses specific examples to illustrate the principles and implementation methods of the present invention. The above examples are only for the purpose of helping to understand the method and core ideas of the present invention. The above are only preferred embodiments of the present invention. It should be noted that due to the limitations of textual expression, and the existence of an infinite number of specific structures, those skilled in the art can make several improvements, modifications, or changes without departing from the principles of the present invention, and can also combine the above technical features in an appropriate manner. These improvements, modifications, changes, or combinations, or the direct application of the inventive concept and technical solution to other situations without modification, should all be considered within the scope of protection of the present invention.

Claims

1. A machine vision-based system for measuring the bending stiffness distribution of a rod, characterized in that... The measuring device and the object to be measured include a camera system, a lighting system, a computer system, and a clamping system, characterized in that: The camera system includes a camera, a camera support frame, camera accessories, and a data transmission cable. Its main feature is that it can be used to capture clear images with a resolution of not less than 1024*512. The lighting system includes fill lights, a support frame, and a soft light curtain. Its main feature is that the fill lights and the support frame provide a light intensity of not less than 5000. The shooting environment features a soft-light screen that provides a high-contrast background. The computer system includes a computer and a data connection line. Its main features are that it reads image information and runs a measurement method, and transmits images captured by the camera to the computer, processes edge data, and performs member stiffness inversion. The clamping system includes a three-jaw clamp and a weight of known weight. Its main feature is that the two ends of the object to be measured are defined as a fixed end and a free end, respectively; one end of the three-jaw clamp is the fixed end, which is subject to fixed constraints; the other end is the free end, where the weight is suspended to apply a concentrated load.

2. A machine vision-based system for measuring the bending stiffness distribution of a rod, characterized in that, The measurement method includes the following steps: (a) Step S1: Calibration of the perpendicular-coplanar relationship between the rod axis and the camera main axis, the main feature of which is that... Define the central axis of the member along its length as horizontal. Axis; with the central optical axis of the camera as the main axis of the camera, it is defined as horizontal. Axis; adjust the lens angle via the support bracket, so that shaft and The spatial angle between the axes is 90°. shaft and Shaft angle deviation ≤ 0.5°; Adjustment axis, The vertical distance between the axis and the horizontal platform surface is consistent, and the deviation of the vertical distance from the horizontal platform surface is ≤0.1mm; (b) Step S2: Image acquisition and preprocessing of the rod, characterized in that S2 includes the following sub-steps: (1) Sub-step S2-1: The rod to be measured is mounted on the measuring device using a three-jaw clamp, characterized in that it satisfies the perpendicular-coplanar relationship with the main axis of the camera; the camera is used to record the initial image of the rod before loading, and the acquisition range covers the complete length of the rod; (2) Sub-step S2-2——Apply a concentrated load to the free end of the member; during the loading process, the member undergoes a certain deformation, and the image of the member after the load stabilizes is recorded by a camera each time; (3) Sub-step S2-3: Crop the image sequence, retain the effective area centered on the rod, and remove background interference areas such as clamps and loads; convert the cropped images into grayscale images. (4) Sub-step S2-4—Image usability assessment, characterized by introducing the standard deviation of image background grayscale fluctuation as a noise level estimation index to assess the noise level; defining the ratio of maximum deflection to beam length. As a measure of deflection range; (c) Step S3: Edge extraction of the rod, characterized in that S3 includes the following sub-steps: (1) Sub-step S3-1—Apply sub-pixel edge detection technology to obtain a discrete set of edge points; (2) Sub-step S3-2 - Screening of edge point pairs, characterized by establishing screening criteria based on the geometric features of the rods, eliminating falsely detected isolated edge points or background interference edges, and obtaining a pure set of edge points; (3) Sub-step S3-3 - Edge grouping and coordinate transformation, characterized in that the selected edge points are divided into upper and lower groups according to spatial proximity, representing the upper and lower edges of the rod; one side edge is selected to represent its deformation shape, and the pixel coordinates are converted into actual coordinates; (4) Sub-step S3-4 - Calculation of actual deflection of the rod, characterized in that the difference between the edge coordinates after loading and the initial state is calculated to obtain the deflection distribution of the rod caused by loading; (d) Step S4: Stiffness inversion of the rod, characterized in that step S4 includes the following sub-steps: (1) Sub-step S4-1 — Perform second derivative on discrete deflection data to invert stiffness; (2) Sub-step S4-2 - Calculation of bending moment distribution: Based on the cantilever beam mechanical model, a concentrated load is applied to the free end of the member. When, its bending moment distribution function is ; (3) Sub-step S4-3—Spatial stiffness inversion, combined with Euler-Bernoulli beam theory; substituting the moment distribution function and the second derivative of deflection, the spatial stiffness distribution is obtained by inversion; (e) Step S5: Stiffness result optimization, characterized by discarding the free end extension region and outputting the stiffness result.

3. The rod as described in claims 1 and 2, characterized in that its geometric features conform to the Euler-Bernoulli beam theory in mechanics of materials, and satisfy the geometric constraint of a length-to-thickness ratio ≥ 300 to ensure negligible shear deformation; it adopts a cantilever beam fixing method, with one end rigidly fixed by a three-jaw clamp, and the other end being a free end, extended by 20% as a loading extension section; its deflection function satisfies the following differential equation: in, Indicates the position coordinates along the length of the beam. Let be the deflection function of the beam. Let x be the bending moment distribution function along the x-direction. Let be the bending stiffness function distributed along the length of the beam.

4. A machine vision-based system for measuring the bending stiffness distribution of a rod as described in claims 1 and 2, characterized in that, The images used for inversion should meet the following requirements: noise level ≤ 10, deflection ratio ≤ 10. .

5. A machine vision-based system for measuring the bending stiffness distribution of a rod as described in claims 1-3, characterized in that, In step S4-1, the deflection coordinates are fitted using the Taylor basis function with the least squares method and the Tikhonov regularization term; the minimum regularization factor is obtained by the generalized cross-validation method, and the second derivative of the fitted basis function is taken as the second derivative of the deflection.

6. A machine vision-based system for measuring the bending stiffness distribution of a rod as described in claims 1-3, characterized in that, In step S5-1, the data of the loading extension section is discarded to avoid numerical divergence caused by the bending moment at the beam end and the second derivative tending to 0.

Images