In-situ finite element simulation method for stress state of two-end full-frame beam based on mixed reality

Abstract

The application relates to the technical field of electronic data digital processing, and discloses a two-end full-frame support beam stress state in-situ finite element simulation method based on mixed reality. The method obtains a three-dimensional geometric model of a support beam and a support base through a mixed reality device, and utilizes a high-frame-rate camera to capture full-field displacement data under an excited state; a finite element model taking the rotation stiffness of the support base as an unknown parameter is constructed, and a real stiffness value is reversely identified through an iterative optimization algorithm; the corrected model is solved under a design load, and stress, strain and deformation nephograms are spatially registered and superimposed on the real structure through the mixed reality device. The system comprises five modules, namely, three-dimensional perception, micro-variation capture, boundary identification, model updating and mixed reality visualization. The application realizes accurate in-situ boundary condition identification, non-contact nondestructive testing and immersive visual evaluation, and significantly improves the accuracy and efficiency of structural safety analysis.

Description

Technical Field

[0001] This invention belongs to the field of electronic data digital processing technology, specifically involving an in-situ finite element simulation method for the stress state of a fully framed beam with two ends based on mixed reality. Background Technology

[0002] With the development of building structural health monitoring and intelligent operation and maintenance technologies, accurate assessment of the stress state of key load-bearing components has become a core requirement for ensuring engineering safety. In concrete or steel structure systems, beams, as the main horizontal load-bearing components, have their internal force distribution directly affected by the support boundary conditions.

[0003] Traditional finite element method (FEM) simulations typically simplify the supports of fully framed beams at both ends to ideal fixed or hinged ends. This rigidity assumption ignores the non-ideal rotational stiffness of supports caused by factors such as node construction, material aging, and construction deviations in actual engineering. While this simplification facilitates modeling, it significantly distorts the bending moment distribution, especially in the mid-span and support regions, easily leading to systematic deviations between calculated and measured values. This can mislead the assessment of structural safety margins and potentially cause over-strengthening or safety hazards.

[0004] In-situ simulation technology based on Mixed Reality (MR) has been introduced into the field of structural analysis in recent years, aiming to achieve intuitive visualization and interactive evaluation of the stress state of components through virtual-real fusion. This technology combines on-site sensor data with virtual finite element models to overlay and display physical fields such as stress and deformation in real space, providing engineers with an immersive diagnostic environment.

[0005] However, existing MR simulation systems still rely on preset boundary conditions and lack the ability to dynamically perceive the actual constraint characteristics of supports. This results in a fundamental disconnect between the virtual model and the physical entity in terms of mechanical behavior, which severely restricts the credibility of simulation results and their engineering guidance value.

[0006] Existing technologies struggle to obtain the true rotational stiffness of supports without compromising structural integrity. Conventional methods rely on empirical values ​​or offline experimental estimations, failing to reflect the time-varying characteristics of boundary conditions under service conditions. Furthermore, while deploying a high density of sensors to reconstruct full-field deformation can indirectly invert boundary parameters, it is costly and difficult to implement, making it unsuitable for rapid assessment of existing buildings. Summary of the Invention

[0007] To address the aforementioned technical problems, this invention provides an in-situ finite element simulation method for the stress state of a fully-supported beam at both ends based on mixed reality. This method aims to achieve high-precision in-situ simulation and intuitive presentation of the mechanical behavior of the beam structure under real boundary constraints through the deep integration of non-contact optical measurement, inverse boundary condition identification, and augmented reality visualization technology. Firstly, this invention uses a 3D perception system integrated into a mixed reality device to perform high-precision geometric scanning of the target beam and its support area, constructing a 3D digital twin model that completely corresponds to the physical entity.

[0008] Subsequently, a high-resolution imaging unit is used to capture the full-field micro-variation morphology of the support beam under minute perturbations, generating accurate surface displacement field data. The core of this invention lies in establishing a finite element model update framework with the support rotational stiffness as an unknown parameter. Through iterative optimization algorithms, the rotational stiffness parameter in the model is continuously adjusted until the residual between the displacement field calculated by the model and the measured displacement field converges to below a preset threshold, thereby inversely identifying the equivalent true rotational stiffness of the support.

[0009] Finally, the identified real stiffness parameters are solidified into the finite element model, solved under the design load, and the calculated stress, strain and deformation cloud maps are accurately superimposed on the real beam structure through the display system of the mixed reality device in a spatial registration manner, providing engineers with in-situ, real-time and visualized structural health status assessment basis.

[0010] This invention provides an in-situ finite element simulation method for the stress state of a fully framed beam with two ends based on mixed reality. The method specifically includes the following steps: Acquire the on-site three-dimensional geometric information of the target beam structure and its support connection area, and generate an initial finite element mesh model based on the three-dimensional geometric information; Real-time capture of the full-field surface displacement data of the target support beam structure under excited state to form a measured displacement field; Establish a parametric finite element analysis model that includes the rotational stiffness parameters of the support to be identified; The difference function between the simulated displacement field calculated by the parameterized finite element analysis model and the measured displacement field is set as the objective function. The rotational stiffness parameter of the support is adjusted by an iterative optimization algorithm until the value of the objective function is less than the preset convergence criterion, and the actual rotational stiffness value of the support is determined. The actual rotational stiffness value is updated to the initial finite element mesh model to construct the corrected finite element model; Under the preset design load conditions, the modified finite element model is solved to obtain the stress distribution, strain distribution and deformation data of the target beam structure. The stress distribution, strain distribution, and deformation data are visualized to generate an augmented reality overlay view that is precisely aligned with the target support beam structure in physical space, and then output through a mixed reality display device.

[0011] As one embodiment of the present invention, obtaining the on-site three-dimensional geometric information of the target beam structure and its support connection area specifically includes: A pre-defined coded structured light pattern is projected onto the surface of the target beam and support using a structured light 3D scanning unit integrated into a mixed reality head-mounted device. The binocular infrared camera array within the structured light 3D scanning unit synchronously captures distorted structured light images modulated by the target beam and support surface. The distorted structured light image is decoded using phase deflection measurement or spatial coding methods to calculate the three-dimensional spatial coordinates of each pixel on the surface of the target beam and support, generating high-density three-dimensional point cloud data. The iterative nearest point algorithm is used to stitch and register the local point cloud data obtained from multiple scans to form a complete global point cloud model; The Poisson surface reconstruction algorithm is applied to the global point cloud model to generate a closed three-dimensional mesh surface model with a continuous topological structure. This model is the geometric basis of the initial finite element mesh model.

[0012] As one embodiment of the present invention, the real-time capture of the full-field surface displacement data of the target beam structure under excited state specifically includes: Using a high-resolution monochrome industrial camera mounted on a tripod or fixed platform, focus is made on the observation area of ​​the target support beam, and a reference image is acquired in a static, undisturbed state. Apply an instantaneous impact load to the target support beam or use environmental vibration to induce an excited state; Multiple frames of images of the target beam under excited state were continuously acquired at a frame rate five times higher than the natural vibration frequency of the structure. The reference image is divided into several non-overlapping rectangular query sub-regions; For each query sub-region, in each frame of the sequence image, a search strategy based on the zero-mean normalized cross-correlation algorithm is used to locate its best matching position in the deformed image. By fitting a bivariate quadratic function surface around the peak of the cross-correlation function, subpixel-level precise positioning of the matching position can be achieved; Calculate the displacement vector of the center point of each query sub-region between the reference image and the sequence image. All displacement vectors together constitute the discretized full-field displacement data covering the observation area, i.e., the measured displacement field.

[0013] As one embodiment of the present invention, adjusting the rotational stiffness parameter of the support through an iterative optimization algorithm specifically includes: In the parametric finite element analysis model, the boundary conditions of the two end supports are set as linear torsion spring elements with unknown rotational stiffness coefficients. The objective function is defined as the sum of squares of the differences between the measured displacement field and the simulated displacement field at all measurement nodes. The covariance matrix adaptive evolution strategy is selected as the iterative optimization algorithm. In each iteration, the covariance matrix adaptive evolution strategy updates the covariance matrix based on the distribution of excellent individuals in the previous generation population, in order to generate a new generation of candidate solutions that follow a multivariate normal distribution, i.e., a new set of support rotation stiffness parameters. Substitute each new set of support rotational stiffness parameters into the parameterized finite element analysis model to solve for the corresponding simulated displacement field and calculate the objective function value. Sort all individuals in the current generation population according to the objective function value, and update the covariance matrix and distribution mean based on the sorting results, until the objective function value is less than 1. If the number of iterations reaches a preset upper limit of 500, the corresponding support rotational stiffness parameter is the actual rotational stiffness value.

[0014] As one embodiment of the present invention, the visualization processing of the stress distribution, strain distribution, and deformation data, and the output through a mixed reality display device, specifically includes: Using the simultaneous localization and mapping system built into the mixed reality display device, the device’s six-degree-of-freedom pose in physical space is tracked in real time, and a sparse or dense three-dimensional map of the surrounding environment is constructed. The coordinate system of the modified finite element model is rigidly transformed and aligned with the global coordinate system of the physical space by manually selecting at least three non-collinear corresponding points or by automatically identifying structural features, thus establishing a spatial anchoring relationship. The calculated scalar field data, such as von Mises equivalent stress or principal strain, are mapped into pseudo-color textures using a preset color lookup table. The pseudo-color texture is applied to the surface of the modified finite element model; The calculated vector field data, such as displacement, is visualized by drawing scaled arrows on model nodes or by proportionally offsetting the coordinates of model vertices. The graphics rendering pipeline of the mixed reality display device renders the finite element model with pseudo-color textures and deformation effects, and uses depth sensor data to achieve correct occlusion with the real physical environment, ultimately forming an overlay view in the user's field of view that seamlessly integrates the virtual simulation results with the real support beam structure.

[0015] According to another aspect of the present invention, an in-situ finite element simulation system for the stress state of a fully framed beam with two ends based on mixed reality is provided, the system comprising: The three-dimensional geometric reality perception module is used to acquire the on-site three-dimensional geometric information of the target beam structure and its support connection area, and generate an initial finite element mesh model based on the three-dimensional geometric information. The micro-deformation field real-time capture module is used to capture the surface full-field displacement data of the target support beam structure under excited state in real time, and form a measured displacement field. The boundary condition reverse identification module is used to establish a parameterized finite element analysis model containing the rotational stiffness parameters of the support to be identified, and to set the difference function between the simulated displacement field calculated by the parameterized finite element analysis model and the measured displacement field as the objective function. The rotational stiffness parameters of the support are adjusted through an iterative optimization algorithm until the value of the objective function is less than the preset convergence criterion, thereby determining the actual rotational stiffness value of the support. The finite element model real-time update and solution module is used to update the actual rotational stiffness value determined by the boundary condition reverse identification module to the initial finite element mesh model, construct the corrected finite element model, and solve the corrected finite element model under the preset design load condition to obtain the stress distribution, strain distribution and deformation data of the target beam structure. The mixed reality augmented visualization module is used to visualize the stress distribution, strain distribution and deformation data obtained by the solution module in real time from the finite element model update, and generate an augmented reality overlay view that is precisely aligned with the target support beam structure in physical space.

[0016] As one embodiment of the present invention, the three-dimensional geometric reality perception module specifically includes: The coded structured light projection unit consists of an infrared laser diode with a center wavelength of 850 nanometers, a diffractive optical element, and a projection lens, and is used to project a preset de Bruin sequence coded light spot pattern onto the surface of the target structure. The binocular infrared imaging unit consists of two infrared image sensors with a resolution of 1280×1024 pixels, along with matching infrared filters and optical lenses, used to synchronously acquire coded light spot images modulated by the structure surface. The geometric reconstruction processing unit contains algorithms for decoding encoded spot images to calculate depth information, an iterative nearest-point algorithm for point cloud registration, and a Poisson surface reconstruction algorithm for generating a 3D mesh model from the point cloud.

[0017] As one embodiment of the present invention, the real-time micro-variable field capture module specifically includes: The high frame rate imaging unit is a global shutter complementary metal-oxide-semiconductor image sensor with a resolution of 2048×2048 pixels and a maximum acquisition frame rate of 500 frames per second. The digital image correlation calculation unit is equipped with a field-programmable gate array (FPGA) for performing zero-mean normalized cross-correlation operations and sub-pixel displacement interpolation calculations, enabling high-speed computation of full-field displacement data from continuously acquired image sequences. The input of the digital image correlation calculation unit is connected to the data output of the high frame rate imaging unit, and the output measured displacement field data is transmitted to the boundary condition inverse identification module.

[0018] In one embodiment of the present invention, the boundary condition inverse identification module and the finite element model real-time update and solution module are integrated in the computational processing core. The computational processing core receives the 3D mesh model generated by the 3D geometric reality perception module and the measured displacement field data generated by the micro-deformation field real-time capture module. The computational processing core internally runs a master control optimization program, which implements a covariance matrix adaptive evolution strategy algorithm and repeatedly calls the finite element solver. During each call, the master control optimization program passes a set of candidate support rotational stiffness parameters to the finite element solver. The finite element solver performs static analysis based on these parameters and the 3D mesh model and returns the simulated displacement field. The master control optimization program compares the simulated displacement field with the measured displacement field to calculate the objective function value and adjusts the generation strategy of the next generation of candidate parameters according to this value until the final actual rotational stiffness value is identified.

[0019] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This invention achieves accurate in-situ identification of support boundary conditions. Through reverse identification technology, it directly calculates the actual rotational stiffness of the structure from its actual response, completely eliminating the idealized assumptions about boundary conditions in traditional analysis. This ensures that the input parameters of the finite element model are highly consistent with physical reality, fundamentally guaranteeing the accuracy of the simulation results.

[0020] 2. Improved reliability of structural safety assessment. Since the boundary conditions used in the model are the results of on-site measurements, the calculated bending moment, shear force, and stress distribution can truly reflect the internal force state of the structure under actual constraints, effectively avoiding serious overestimation or underestimation of the structural bearing capacity and safety margin due to incorrect boundary condition assumptions.

[0021] 3. It provides a new non-contact, non-destructive testing paradigm. The entire testing process relies solely on optical measurements, eliminating the need for installing any sensors on the structure or performing destructive sampling, thus ensuring the integrity of the tested structure. It is suitable for rapid and low-cost health assessments of existing critical structures.

[0022] 4. An intuitive and immersive way of interacting with analysis results has been created. Through mixed reality technology, abstract finite element simulation data cloud maps are directly and accurately superimposed onto real physical structures, enabling engineers to intuitively observe stress concentration areas and potential weak points on-site, greatly enhancing the depth of understanding of structural mechanical behavior and decision-making efficiency. Attached Figure Description

[0023] Figure 1 This is a schematic diagram of the overall technical solution architecture proposed in this invention; Figure 2 This is a schematic diagram of the core principle framework of the reverse identification of support rotational stiffness and finite element model update in this invention. Figure 3 This is a logical flowchart of the three-dimensional geometric reality perception and initial finite element mesh model construction in this invention; Figure 4 This is a flowchart illustrating the logical process of real-time capture of micro-variable morphological fields and generation of measured displacement fields in this invention. Figure 5 This is a schematic diagram of the spatial registration and data overlay principle framework of the mixed reality augmented visualization module in this invention; Figure 6 This is a schematic diagram of the multi-level interaction relationship and data flow between the mixed reality device and the back-end computing system in this invention. Detailed Implementation

[0024] Please refer to Figures 1 to 6 This invention provides an in-situ finite element simulation method for the stress state of a fully framed beam with two ends based on mixed reality, aiming to solve the problems of bending moment calculation errors and misjudgment of structural safety margins caused by idealized assumptions of support boundary conditions in existing structural analyses. This method achieves high-precision in-situ simulation and intuitive presentation of the mechanical behavior of the beam under real boundary constraints by integrating non-contact optical measurement, parametric finite element model updating, and mixed reality visualization technology. The specific implementation steps of this method will be described in detail below.

[0025] The method first performs step S1: acquiring the on-site three-dimensional geometric information of the target beam structure and its support connection area, and generating an initial finite element mesh model based on the three-dimensional geometric information. This step is completed by a structured light three-dimensional scanning unit integrated into a mixed reality head-mounted device. The structured light three-dimensional scanning unit includes an infrared laser diode with a center wavelength of 850 nanometers, diffractive optical elements, and a projection lens, used to project a preset de Bruin sequence-coded light spot pattern onto the surface of the target beam and support. Simultaneously, the binocular infrared imaging unit consists of two infrared image sensors with a resolution of 1280×1024 pixels, equipped with a dedicated infrared filter and optical lens, synchronously acquiring the distorted coded light spot image modulated by the structural surface.

[0026] The geometric reconstruction processing unit receives the aforementioned image data and first decodes the coded pattern using phase deflection measurement to calculate the three-dimensional spatial coordinates of each pixel, forming a local high-density point cloud. Since a single scan can only cover a limited field of view, the system automatically controls the user's mobile device's viewing angle, acquiring local point clouds from different angles multiple times. Subsequently, an iterative nearest-point algorithm is used to rigidly register all local point clouds, eliminating coordinate deviations introduced by changes in device pose, ultimately stitching them together into a complete global point cloud model. Based on this, a Poisson surface reconstruction algorithm is applied to implicitly fit the global point cloud, generating a closed three-dimensional mesh surface model with a continuous topological structure. This model preserves the geometric details of the connection area between the main beam and the support, including cross-sectional contours, chamfers, bolt holes, and other features, serving as the geometric basis for subsequent finite element modeling. An adaptive tetrahedral meshing strategy is used to ensure higher mesh density in stress concentration areas such as support corners, while maintaining appropriate sparseness in straight sections, balancing computational efficiency and accuracy.

[0027] The S2 step then proceeds: real-time capture of the full-field surface displacement data of the target beam structure under excited conditions, forming a measured displacement field. This step relies on an independently deployed high-frame-rate imaging system, the core of which is a global shutter complementary metal-oxide-semiconductor image sensor with a resolution of 2048×2048 pixels and a maximum acquisition frame rate of 500 frames per second. The system first acquires a static reference image under undisturbed conditions as a baseline for digital image correlation. Next, an instantaneous impact load is applied to the target beam, such as by gently tapping the mid-span of the beam with a small electromagnetic hammer, or by naturally exciting the structural response using environmental micro-vibrations. Under excited conditions, the high-frame-rate imaging unit continuously acquires multiple image sequences at a sampling frequency no less than five times the lowest-order natural frequency of the structure.

[0028] After receiving the image sequence, the digital image correlation calculation unit divides the reference image into several non-overlapping rectangular query sub-regions, typically 32×32 pixels in size. For each query sub-region, in each frame of deformed image, a zero-mean normalized cross-correlation algorithm is used to search for the best matching region. This algorithm determines a coarse matching position by maximizing the normalized cross-correlation coefficient, and then constructs a bivariate quadratic function surface in the neighborhood of that position. Subpixel-level interpolation is performed on the cross-correlation peak to obtain the precise displacement components of the sub-region center point in the horizontal and vertical directions. The displacement vectors of all query sub-regions together constitute a discrete full-field displacement dataset covering the entire observation area, i.e., the measured displacement field. This displacement field contains time dimension information, but this invention only extracts the maximum displacement amplitude in the steady-state or quasi-static response stage as the optimization objective to simplify the inverse identification process.

[0029] Next, step S3 is performed: a parametric finite element analysis model containing the rotational stiffness parameters of the support to be identified is established. This model is based on the three-dimensional mesh generated in step S1 and constructed within a commercial finite element solver framework. The material properties of the support beam are set according to the design drawings or on-site sampling test results, including elastic modulus, Poisson's ratio, and density.

[0030] The key lies in the parameterization of the boundary conditions: the constraints at both supports are defined as linear torsional spring elements with rotational stiffness coefficients. and The model is introduced as an unknown design variable. Initial values ​​can be set to one-tenth to ten times the theoretical fixed-end stiffness, covering the possible range from approximately hinged to fully fixed. The model is discretized using second-order tetrahedral elements to ensure high-order accuracy in bending deformation. The loading condition does not apply actual design loads initially; instead, it simulates the same small perturbation excitation as in step S2, such as applying a unit concentrated force at mid-span, to calculate a simulated displacement response comparable to the measured displacement field.

[0031] Next, step S4 is executed: The difference function between the simulated displacement field calculated by the parameterized finite element analysis model and the measured displacement field is set as the objective function. The rotational stiffness parameter of the support is adjusted using an iterative optimization algorithm until the objective function value is less than a preset convergence criterion, thus determining the actual rotational stiffness value of the support. The objective function J is defined as the simulated displacement at all measurement nodes. Compared with the measured displacement The mathematical expression for the sum of squares of the differences between corresponding components is: ; Where N is the total number of valid measurement nodes. and These represent the horizontal and vertical directions, respectively. The optimization algorithm employs a covariance matrix adaptive evolution strategy. This strategy initializes a population of 20 individuals, with each individual forming a group. Parameter Combination. In each iteration, the master optimization program sequentially passes all individual parameters to the finite element solver, which returns the corresponding simulated displacement field and calculates the objective function value. The population is then sorted according to the objective function value, and the top 40% of the best individuals are selected to update the mean vector and covariance matrix of the multivariate normal distribution.

[0032] The next generation of candidate solutions is generated by random sampling from this distribution, ensuring that the search direction shifts towards the region with low objective function values. The iterative process continues until the objective function value is less than [a certain value]. Or the number of iterations reaches the upper limit of 500. The final converged result is... This is the actual rotational stiffness value of the support.

[0033] Then, step S5 is executed: the actual rotational stiffness value is updated to the initial finite element mesh model, and the corrected finite element model is constructed. This process will remove the values ​​identified in step S4. and The actual parameters of the support boundary conditions are solidified, replacing the initial assumptions in the original model. The corrected model fully retains the original geometry, material, and mesh information, with only the boundary constraints precisely calibrated, thus ensuring that it can accurately reflect the mechanical properties of the structure under actual support conditions.

[0034] Next, step S6 is executed: Under the preset design load conditions, the modified finite element model is solved to obtain the stress distribution, strain distribution, and deformation data of the target beam structure. The design load is set according to engineering specifications or actual usage scenarios, such as uniformly distributed load, concentrated load, or combined load. The finite element solver performs static analysis and outputs nodal displacements, element stress tensors, and strain tensors across the entire model range. Key results include von Mises equivalent stress, principal strain directions and magnitudes, and the overall deflection curve. These data constitute the complete mechanical response spectrum of the structure under real boundary constraints.

[0035] Finally, step S7 is executed: the stress distribution, strain distribution, and deformation data are visualized to generate an augmented reality overlay view that is precisely aligned with the target beam structure in physical space, and then output through a mixed reality display device. This process first relies on the simultaneous localization and mapping system built into the mixed reality device to track the device's six-degree-of-freedom pose in physical space in real time and construct a three-dimensional map of the environment. Subsequently, the corrected finite element model coordinate system is aligned with the global coordinate system of the physical space.

[0036] Alignment can be achieved manually by selecting three or more non-collinear corresponding points, such as beam ends, mid-span, and support feature points; or by using an automatic feature recognition algorithm to match geometric features between the model and the real-world environment. After establishing spatial anchoring relationships, scalar field data, such as von Mises stress, are pseudo-color mapped using a preset color lookup table to generate texture maps and apply them to the model surface. For displacement vector fields, a proportional offset method for vertex coordinates is used to visualize geometric deformation, with the offset scaling factor dynamically adjusted based on the maximum deflection to ensure visibility. The graphics rendering pipeline receives the above data and, combined with depth information from a depth sensor, correctly handles the occlusion relationship between the virtual model and the real structure. Finally, users observe an immersive view through a mixed reality display device, where stress cloud maps and deformation patterns are precisely superimposed on the real beam, achieving in-situ, intuitive structural condition assessment.

[0037] Throughout the methodology, strict data dependencies and feedback mechanisms exist between each step. The geometric model generated in S1 forms the basis for modeling in S3; the measured displacement field acquired in S2 is the target for optimization in S4; the stiffness parameters identified in S4 directly drive model correction in S5; and the solution results in S6 provide a visualization data source for S7. The system ensures a high degree of consistency between the simulation model and the physical entity through closed-loop iteration. An anomaly handling mechanism is also embedded in each step: if point cloud stitching fails, the user is prompted to rescan the missing area; if the signal-to-noise ratio of the displacement field is too low, the acquisition time is automatically extended or the excitation intensity is increased; if the optimization process diverges, it reverts to the previous generation's optimal solution and narrows the search range. All data streams are exchanged through a unified intermediate format to ensure seamless integration between modules.

[0038] The method described in this embodiment fully realizes the entire process from on-site data acquisition, boundary condition identification, model correction to augmented reality visualization, solves the fundamental problem of boundary condition distortion in traditional structural analysis, and provides a high-precision, non-contact, and intuitive technical means for the safety assessment of existing building structures.

Claims

1. An in-situ finite element simulation method for the stress state of a fully framed beam with two ends based on mixed reality, characterized in that: include: Acquire the on-site three-dimensional geometric information of the target beam structure and its support connection area, and generate an initial finite element mesh model based on the three-dimensional geometric information; Real-time capture of the full-field surface displacement data of the target support beam structure under excited state to form a measured displacement field; Establish a parametric finite element analysis model that includes the rotational stiffness parameters of the support to be identified; The difference function between the simulated displacement field calculated by the parameterized finite element analysis model and the measured displacement field is set as the objective function. The rotational stiffness parameter of the support is adjusted by an iterative optimization algorithm until the value of the objective function is less than the preset convergence criterion, and the actual rotational stiffness value of the support is determined. The actual rotational stiffness value is updated to the initial finite element mesh model to construct the corrected finite element model; Under the preset design load conditions, the modified finite element model is solved to obtain the stress distribution, strain distribution and deformation data of the target beam structure. The stress distribution, strain distribution, and deformation data are visualized to generate an augmented reality overlay view that is precisely aligned with the target support beam structure in physical space, and then output through a mixed reality display device.

2. The in-situ finite element simulation method for the stress state of a fully framed beam with two ends based on mixed reality according to claim 1, characterized in that, Acquire the on-site three-dimensional geometric information of the target beam structure and its support connection area, and generate an initial finite element mesh model based on the three-dimensional geometric information, including: A pre-defined coded structured light pattern is projected onto the surface of the target beam and support using a structured light 3D scanning unit integrated into a mixed reality head-mounted device. The binocular infrared camera array within the structured light 3D scanning unit synchronously captures distorted structured light images modulated by the target beam and support surface. The distorted structured light image is decoded using phase deflection measurement or spatial coding methods to calculate the three-dimensional spatial coordinates of each pixel on the surface of the target beam and support, generating high-density three-dimensional point cloud data. The iterative nearest point algorithm is used to stitch and register the local point cloud data obtained from multiple scans to form a complete global point cloud model; The Poisson surface reconstruction algorithm is applied to the global point cloud model to generate a closed three-dimensional mesh surface model with a continuous topological structure. This model is the geometric basis of the initial finite element mesh model.

3. The in-situ finite element simulation method for the stress state of a fully framed beam with two ends based on mixed reality according to claim 2, characterized in that, Real-time capture of the full-field surface displacement data of the target beam structure under excited state to form a measured displacement field, including: Using a high-resolution monochrome industrial camera mounted on a tripod or fixed platform, focus is made on the observation area of ​​the target support beam, and a reference image is acquired in a static, undisturbed state. Apply an instantaneous impact load to the target support beam or use environmental vibration to induce an excited state; Multiple frames of images of the target beam under excited state were continuously acquired at a frame rate five times higher than the natural vibration frequency of the structure. The reference image is divided into several non-overlapping rectangular query sub-regions; For each query sub-region, in each frame of the sequence image, a search strategy based on the zero-mean normalized cross-correlation algorithm is used to locate its best matching position in the deformed image. By fitting a bivariate quadratic function surface around the peak of the cross-correlation function, subpixel-level precise positioning of the matching position can be achieved; Calculate the displacement vector of the center point of each query sub-region between the reference image and the sequence image. All displacement vectors together constitute the discretized full-field displacement data covering the observation area, i.e., the measured displacement field.

4. The in-situ finite element simulation method for the stress state of a fully framed beam with two ends based on mixed reality according to claim 3, characterized in that, Establish a parametric finite element analysis model that includes the rotational stiffness parameters of the support to be identified, including: In the parametric finite element analysis model, the boundary conditions of the two end supports are set as linear torsion spring elements with unknown rotational stiffness coefficients. The material properties of the target support beam are set according to the design drawings or on-site sampling test results, including elastic modulus, Poisson's ratio and density; The initial finite element mesh model is discretized using second-order tetrahedral elements to ensure high-order accuracy of bending deformation; A unit concentrated force is applied at the mid-span position as a small perturbation excitation to generate a simulated displacement response comparable to the measured displacement field.

5. The in-situ finite element simulation method for the stress state of a fully framed beam with two ends based on mixed reality according to claim 4, characterized in that, The difference function between the simulated displacement field calculated by the parameterized finite element analysis model and the measured displacement field is set as the objective function. The rotational stiffness parameter of the support is adjusted through an iterative optimization algorithm until the objective function value is less than a preset convergence criterion. The actual rotational stiffness value of the support is then determined, including: The objective function is defined as the sum of squares of the differences between the measured displacement field and the simulated displacement field at all measurement nodes. The covariance matrix adaptive evolution strategy is selected as the iterative optimization algorithm. In each iteration, the covariance matrix adaptive evolution strategy updates the covariance matrix based on the distribution of excellent individuals in the previous generation population, in order to generate a new generation of candidate solutions that follow a multivariate normal distribution, i.e., a new set of support rotation stiffness parameters. Substitute each new set of support rotational stiffness parameters into the parameterized finite element analysis model to solve for the corresponding simulated displacement field and calculate the objective function value. Sort all individuals in the current generation population according to the objective function value, and update the covariance matrix and distribution mean based on the sorting results, until the objective function value is less than 1. If the number of iterations reaches a preset upper limit of 500, the corresponding support rotational stiffness parameter is the actual rotational stiffness value.

6. The in-situ finite element simulation method for the stress state of a fully framed beam with two ends based on mixed reality according to claim 5, characterized in that, The actual rotational stiffness value is updated to the initial finite element mesh model to construct the corrected finite element model, including: The actual rotational stiffness value is solidified as the stiffness parameter of the linear torsion spring unit at both ends; The geometry, material properties, and mesh generation scheme of the initial finite element mesh model are retained; The model correction is completed by simply replacing the initial assumptions of the support boundary conditions in the original model.

7. The in-situ finite element simulation method for the stress state of a fully framed beam with two ends based on mixed reality according to claim 6, characterized in that, Under the preset design load conditions, the modified finite element model is solved to obtain the stress distribution, strain distribution, and deformation data of the target beam structure, including: Design loads are set according to engineering specifications or actual usage scenarios, including uniformly distributed loads, concentrated loads, or combined loads. Perform static analysis and output nodal displacements, element stress tensors, and strain tensors across the entire model. The von Mises equivalent stress, principal strain direction and magnitude, and overall deflection curve were extracted as key mechanical response data.

8. The in-situ finite element simulation method for the stress state of a fully framed beam with two ends based on mixed reality according to claim 7, characterized in that, The stress distribution, strain distribution, and deformation data are visualized to generate an augmented reality overlay view that is precisely aligned with the target beam structure in physical space, and then output through a mixed reality display device, including: Using the simultaneous localization and mapping system built into the mixed reality display device, the device’s six-degree-of-freedom pose in physical space is tracked in real time, and a sparse or dense three-dimensional map of the surrounding environment is constructed. The coordinate system of the modified finite element model is rigidly transformed and aligned with the global coordinate system of the physical space by manually selecting at least three non-collinear corresponding points or by automatically identifying structural features, thus establishing a spatial anchoring relationship. The calculated scalar field data, such as von Mises equivalent stress or principal strain, are mapped into pseudo-color textures using a preset color lookup table. The pseudo-color texture is applied to the surface of the modified finite element model; The calculated vector field data, such as displacement, is visualized by drawing scaled arrows on model nodes or by proportionally offsetting the coordinates of model vertices. The graphics rendering pipeline of the mixed reality display device renders the finite element model with pseudo-color textures and deformation effects, and uses depth sensor data to achieve correct occlusion with the real physical environment, ultimately forming an overlay view in the user's field of view that seamlessly integrates the virtual simulation results with the real support beam structure.

9. The in-situ finite element simulation method for the stress state of a fully framed beam with two ends based on mixed reality as described in claim 8, characterized in that, The coded structured light pattern is a de Bruin sequence coded light spot pattern, projected by a coded structured light projection unit consisting of an infrared laser diode with a center wavelength of 850 nanometers, diffractive optical elements, and a projection lens.

10. The in-situ finite element simulation method for the stress state of a fully framed beam with two ends based on mixed reality according to claim 9, characterized in that, The high-resolution monochrome industrial camera is a global shutter complementary metal-oxide-semiconductor image sensor with a resolution of 2048×2048 pixels and a maximum acquisition frame rate of 500 frames per second.

Images