A camera for capturing a three-dimensional stress field of a gripping structure

Abstract

A camera for capturing the three-dimensional stress field of a structure includes an image capturing device and a three-dimensional stress field acquisition algorithm. The image capturing device includes a camera and a boundary condition extraction algorithm; the boundary condition extraction algorithm can accurately obtain the difference between the state space of feature points and their initial state. The three-dimensional stress field acquisition algorithm adopts a physical AI approach, acquiring the three-dimensional stress field information of the structure through changes in the state space of feature points, achieving global perception; the physical AI uses physical simulation to acquire the deformation characteristics of the structure, and subsequently exports the model as a VTK file, ultimately realizing the edge deployment of the physical AI at the camera side; the edge deployment adopts a GPU acceleration, multi-threaded parallel computing, and memory preloading scheme to ensure that the response time of the edge computing node is in the millisecond range, enabling smooth visualization of the three-dimensional stress field.

Description

Technical Field

[0001] This invention relates to the field of intelligent sensors, specifically to a camera for capturing the three-dimensional stress field of a structure. Background Technology

[0002] Strain gauges have wide applications in the aerospace field, such as in aircraft deformation monitoring, where they can acquire strain data from the surface of the aircraft to assess structural safety. However, strain gauges have limitations: they can only acquire two-dimensional strain data at the installation location, failing to comprehensively reflect the three-dimensional stress field of the aircraft. Furthermore, to accurately assess structural safety, engineers typically need to carefully select the installation locations of the strain gauges, a process reliant on extensive human experience. On the other hand, the installation and placement of strain gauges may be constrained by space and location, resulting in incomplete data acquisition and difficulty in accurately capturing changes in dynamic forces.

[0003] In recent years, cameras and their accompanying artificial intelligence algorithms have made significant progress, enabling the effective acquisition of state information of key feature points of aircraft. Meanwhile, with the development of large-scale models and high-performance computing, real-time perception based on cameras has become possible.

[0004] Based on extensive experience, our team developed a camera for capturing the three-dimensional stress field of structures. Through the camera and boundary condition extraction algorithms, the system can accurately acquire the state space of feature points. Compared to the initial state The differences are then identified. Subsequently, a physics-based AI approach is used to achieve global perception. This system can capture the distribution of three-dimensional stress in real time, providing a more comprehensive and accurate stress field analysis. Through this method, we can effectively overcome the limitations of strain gauges, obtain richer mechanical information, and provide more accurate data support for aircraft structural optimization, fault prediction, and safety assessment. Summary of the Invention

[0005] To address the above problems, this invention provides a camera for capturing the three-dimensional stress field of a structure, characterized by comprising: an image capturing device and a three-dimensional stress field acquisition algorithm.

[0006] The image capturing device includes a camera and a boundary condition extraction algorithm; the boundary condition extraction algorithm can accurately acquire the state space of feature points. Compared to the initial state The difference, in which the state space The expression is as follows:

[0007]

[0008] In the formula Represented as the three-dimensional spatial coordinates of feature points, located in the spatial domain Inside; the spatial domain Described as a structural surface, specifically:

[0009]

[0010] In the formula Represented as the initial three-dimensional spatial coordinates of the feature points;

[0011] In the formula The surface strain field, expressed as characteristic points, is as follows:

[0012] \xi \left ( {r} \right )=\left [ {{\xi}_{11}, {\xi}_{12}; {\xi}_{21}, {\xi}_{22}} \right ]

[0013] The , , , This refers to the strain amplitude;

[0014] In the formula Described as a temperature field;

[0015] The three-dimensional stress field acquisition algorithm employs a physical AI approach, utilizing the feature point state space. The changes in the structure's three-dimensional stress field information are obtained to achieve global perception. The physical AI uses physical simulation to obtain the deformation characteristics of the structure. The model is then exported as a VTK file, and the physical AI is finally deployed at the edge of the camera. The edge deployment adopts GPU acceleration, multi-threaded parallel computing and memory preloading scheme to ensure that the response time of the edge computing node is in the millisecond level, and the visualization of the three-dimensional stress field can be smoothly realized.

[0016] Furthermore, the camera for capturing the three-dimensional stress field of a structure is characterized in that the feature point state space... The difference is expressed as: the displacement amplitude of the feature point. Local strain amplitude at characteristic points Temperature values ​​of feature points The For feature points The three-dimensional spatial coordinates.

[0017] Furthermore, the camera for capturing the three-dimensional stress field of a structure is characterized in that the boundary condition extraction algorithm includes two implementation schemes: intelligent sensor perception and artificial intelligence algorithm calculation; the intelligent sensor perception uses micro-strain gauges to obtain the strain amplitude of local feature points. Subsequently, the current strain value of the feature point is used as a boundary condition and passed to the physical AI model to finally obtain the three-dimensional stress field of the structure. The artificial intelligence algorithm is based on image data and obtains the displacement amplitude of the feature point relative to the initial state through a convolutional neural network. The following will As boundary conditions, these conditions are passed to the physical AI model to ultimately obtain the three-dimensional stress field of the structure.

[0018] Furthermore, the camera for capturing the three-dimensional stress field of a structure is characterized in that the physical AI deployed at the edge of the camera side employs a model reduction method, recording feature points in multiple state spaces through multiple sets of VTK files. Based on the global information, and ultimately through data mapping, the real-time perception of the structure's three-dimensional deformation and stress state is achieved.

[0019] The advantages of this invention are:

[0020] 1. Acquiring high-dimensional information: This system is applied to measurement and testing in the aerospace field, and can effectively acquire three-dimensional stress field data of structures. Compared with traditional strain gauges, this system not only avoids the problem of inferring the accuracy of macroscopic deformation through discrete acquisition nodes, but also comprehensively captures the deformation characteristics inside the structure, providing more accurate and richer mechanical information.

[0021] 2. Fast Response Speed: Traditional strain gauge measurement methods require manual post-processing of experimental data, a cumbersome and time-consuming process. This system acquires structural deformation characteristics through physical simulation, then exports the model as a VTK file, ultimately enabling edge deployment of physical AI at the camera side. Edge deployment employs GPU acceleration, multi-threaded parallel computing, and memory preloading to ensure millisecond-level response times for computing nodes, achieving smooth 3D stress field visualization.

[0022] 3. High cost-effectiveness: As strain gauges are disposable consumables, traditional experiments typically require a large number of strain gauges to improve sensing accuracy. This system, through image acquisition and intelligent algorithms, uses a camera to obtain three-dimensional stress field data of the structure, exhibiting high reusability, significantly reducing experimental costs, and improving testing efficiency and accuracy. Attached Figure Description

[0023] The accompanying drawings, which are provided to further illustrate embodiments of the invention and constitute a part of this invention, are not intended to limit the scope of the invention.

[0024] Figure 1 A camera system framework for capturing the three-dimensional stress field of a structure.

[0025] Figure 2: Scenarios for structural testing in the aerospace field.

[0026] Figure 3 Application examples of strain gauges in the aerospace field.

[0027] Figure 4 A computer-based webpage display for capturing the three-dimensional stress field of a structure using a camera.

[0028] Figure 5 Example of a camera application for capturing the three-dimensional stress field of a structure. Detailed Implementation

[0029] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the embodiments and accompanying drawings. The illustrative embodiments and descriptions of this invention are only for explaining this invention and are not intended to limit this invention.

[0030] Example 1

[0031] Appendix Figure 1 This paper presents a camera system framework for capturing the three-dimensional stress field of a structure, characterized by comprising: an image capturing device and a three-dimensional stress field acquisition algorithm.

[0032] The image capturing device includes a camera and a boundary condition extraction algorithm; the boundary condition extraction algorithm can accurately acquire the state space of feature points. Compared to the initial state The difference, in which the state space The expression is as follows:

[0033]

[0034] In the formula Represented as the three-dimensional spatial coordinates of feature points, located in the spatial domain Inside; the spatial domain Described as a structural surface, specifically:

[0035]

[0036] In the formula Represented as the initial three-dimensional spatial coordinates of the feature points;

[0037] In the formula The surface strain field, expressed as characteristic points, is as follows:

[0038] \xi \left ( {r} \right )=\left [ {{\xi}_{11}, {\xi}_{12}; {\xi}_{21}, {\xi}_{22}} \right ]

[0039] The , , , This refers to the strain amplitude;

[0040] In the formula Described as a temperature field;

[0041] The three-dimensional stress field acquisition algorithm employs a physical AI approach, utilizing the feature point state space. The changes in the structure's three-dimensional stress field information are obtained to achieve global perception. The physical AI uses physical simulation to obtain the deformation characteristics of the structure. The model is then exported as a VTK file, and the physical AI is finally deployed at the edge of the camera. The edge deployment adopts GPU acceleration, multi-threaded parallel computing and memory preloading scheme to ensure that the response time of the edge computing node is in the millisecond level, and the visualization of the three-dimensional stress field can be smoothly realized.

[0042] Furthermore, the camera for capturing the three-dimensional stress field of a structure is characterized in that the feature point state space... The difference is expressed as: the displacement amplitude of the feature point. Local strain amplitude at characteristic points Temperature values ​​of feature points The For feature points The three-dimensional spatial coordinates.

[0043] Furthermore, the camera for capturing the three-dimensional stress field of a structure is characterized in that the boundary condition extraction algorithm includes two implementation schemes: intelligent sensor perception and artificial intelligence algorithm calculation; the intelligent sensor perception uses micro-strain gauges to obtain the strain amplitude of local feature points. Subsequently, the current strain value of the feature point is used as a boundary condition and passed to the physical AI model to finally obtain the three-dimensional stress field of the structure. The artificial intelligence algorithm is based on image data and obtains the displacement amplitude of the feature point relative to the initial state through a convolutional neural network. The following will As boundary conditions, these conditions are passed to the physical AI model to ultimately obtain the three-dimensional stress field of the structure.

[0044] Furthermore, the camera for capturing the three-dimensional stress field of a structure is characterized in that the physical AI deployed at the edge of the camera side employs a model reduction method, recording feature points in multiple state spaces through multiple sets of VTK files. Based on the global information, and ultimately through data mapping, the real-time perception of the structure's three-dimensional deformation and stress state is achieved.

[0045] Example 2

[0046] Appendix Figure 2-3 This section showcases a scenario for structural testing in the aerospace field and application examples of strain gauges. Engineers use experience to determine the coordinates of the sampling points and place resistance strain gauges at sensitive points such as flange root connectors, rib gaps, and opening transitions. In the straight plate area, uniaxial / triaxial fancy strain gauges are used to determine the principal strain, while in the circular shaft or sleeve area, pairs of gauges arranged at ±45° are used to measure torsion.

[0047] The strain gauges were mounted by grinding and degreasing, marking, pasting and curing, welding, and wire fixing. A Wheatstone bridge with a 5V bridge voltage was used for the measurement link. The sampling frequency was set to 10–100 Hz for static and quasi-static sampling, with anti-aliasing filtering and clock synchronization implemented. Zero-point and temperature drift checks and calibrations were performed before loading, and sensitivity and linearity were verified. During the experiment, the strain-time / load curves at each point were recorded in real time, providing data for stress conversion at key sections, allowable stress verification, and fatigue life assessment.

[0048] Example 3

[0049] Appendix Figure 4 This demonstrates a web-based interface for capturing the 3D stress field of a structure using a camera. The menu bar contains five main options: File, Data Source, Visualization, Rendering Acceleration, and About. The File menu is primarily used for deploying and configuring the VTK model to achieve real-time simulation of specific scenarios.

[0050] The data source mainly includes:

[0051] 1. Camera configuration: Real-time images are acquired via RTSP video stream, and the state space of feature points is subsequently obtained through a boundary condition extraction algorithm;

[0052] 2. Structural Deformation Feature Display (Linear Loading): Through the built-in code of the program, the linear transformation of the state space of feature points is realized, and the three-dimensional stress features of the structure are obtained in real time.

[0053] 3. Historical data backtracking: The program code connects to the MySQL database to obtain the historical state information of feature points. Subsequently, by using the user-specified sampling frequency, the structural deformation characteristics are backtracked to obtain the evolution data of three-dimensional forces.

[0054] Visualization mainly includes:

[0055] Background color: Users can choose the background color;

[0056] Grid Display Button: Allows users to choose whether to display the grid frame;

[0057] Axis display button: Allows users to choose whether to display the axis direction in real time;

[0058] Legend display: Allows users to choose whether to display legends, and control the size, name and image position of legends.

[0059] Rendering acceleration mainly includes:

[0060] 1. Cache pool size: By establishing a cache pool, fluctuations in camera and boundary condition extraction algorithms are avoided, thereby obtaining smooth rendering.

[0061] 2. Preloaded frames: By preloading data in memory, the system ensures good rendering effects and ultimately achieves real-time updates from camera vision to 3D force.

[0062] 3. Number of threads: Depending on the computer's configuration, users can select the number of threads. Currently, 1, 4, 8, 16, and 64 are supported. The more threads there are, the faster the rendering speed and the better the visualization effect under normal circumstances, but the higher the requirements for computer performance are.

[0063] The "About" menu primarily displays user affiliation information and the software's core functions.

[0064] Appendix Figure 5 This paper demonstrates a camera application example for capturing the three-dimensional stress field of a structure. The camera captures footage of a hand-launched aircraft, and the system obtains the deformation characteristics of the structure under gravitational loads, enabling real-time image updates. This solution greatly simplifies the acquisition of the three-dimensional force field, reduces experimental costs, and offers excellent demonstrability.

[0065] The specific implementation methods described above provide a detailed explanation of the purpose, technical solution, and beneficial effects of this invention. It should be understood that the above description is merely a specific embodiment of this invention and is not intended to limit the scope of protection of this invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A camera for capturing a three-dimensional stress field of a gripping structure, characterized in that Includes: image capture device and 3D stress field acquisition algorithm. The image capturing device comprises a camera and a boundary condition extraction algorithm; the boundary condition extraction algorithm can accurately obtain the state space of the feature point Compared with the initial state The difference, wherein the state space Is expressed as: In the formula The three-dimensional spatial coordinates of the feature points are expressed in a spatial domain ; the spatial domain is expressed as a structure surface, in particular: In the formula expressed as the initial three-dimensional spatial coordinates of the feature points; In the formula The surface strain field is expressed as a characteristic point, specifically: The , , , This refers to the strain amplitude; In the formula Described as a temperature field; The three-dimensional stress field acquisition algorithm employs a physical AI approach, utilizing the feature point state space. The system acquires 3D stress field information of the structure by observing changes in its structure, achieving global perception. The physical AI uses physical simulation to obtain the deformation characteristics of the structure, and then exports the model as a VTK file, ultimately enabling the physical AI to be deployed at the edge of the camera. The edge deployment adopts a GPU acceleration, multi-threaded parallel computing, and memory preloading scheme to ensure that the response time of the edge computing node is in the millisecond range, enabling visualization of the 3D stress field. The multi-threaded parallel computing is described as follows: by establishing a cache pool, fluctuations in the camera and boundary condition extraction algorithm are avoided, thereby obtaining the rendered image; users can select the number of threads, and the more threads, the faster the rendering speed and the better the visualization effect; by preloading data in memory, real-time updates of the 3D force from the camera's vision are achieved.

2. The camera for capturing the three-dimensional stress field of a structure according to claim 1, characterized in that, The feature point state space The difference is expressed as: the displacement amplitude of the feature point. Local strain amplitude at characteristic points and / or temperature values ​​of feature points The For feature points The three-dimensional spatial coordinates.

3. The camera for capturing the three-dimensional stress field of a structure according to claim 1, characterized in that, The boundary condition extraction algorithm includes two implementation schemes: intelligent sensor sensing and artificial intelligence algorithm calculation; the intelligent sensor sensing uses micro-strain gauges to obtain the strain amplitude of local feature points. Subsequently, the current strain value of the feature point is used as a boundary condition and passed to the physical AI model to finally obtain the three-dimensional stress field of the structure. The artificial intelligence algorithm is based on image data and obtains the displacement amplitude of the feature point relative to the initial state through a convolutional neural network. The following will As boundary conditions, these conditions are passed to the physical AI model to ultimately obtain the three-dimensional stress field of the structure.

4. A camera for capturing a three-dimensional stress field of a structure according to claim 1, characterized in that, The physical AI deployed at the edge of the camera side employs a model reduction method, recording feature points in multiple state spaces using multiple sets of VTK files. Based on the global information, and ultimately through data mapping, the real-time perception of the structure's three-dimensional deformation and stress state is achieved.

Images