Structure internal polygon defect back analysis identification method
Through the combination of manifold method and optimization algorithm, the improved gray wolf algorithm is used to identify polygon defects within the structure, which solves the problem of difficult to identify complex polygon defects in the existing technology, and achieves high-precision and efficient defect inversion.
Patent Information
- Application Number
- CN202411351518.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-26
- Publication Date
- 2024-11-29
AI Technical Summary
The prior art is difficult to effectively identify and invert polygon defects inside structures, especially polygon defects in complex geometric feature designs such as triangles and quadrilaterals, and lacks mature solutions.
The manifold method is combined with optimization algorithm, and the initial defect parameters are randomly generated by building a calculation model, and the defect position and shape parameters are iteratively optimized. The defect inversion is used using the improved gray wolf algorithm to reduce computational meshing, and the recognition accuracy and efficiency are improved.
It improves the recognition accuracy of polygon defects inside the structure, enhances the global convergence and robustness of inversion, reduces the calculation overhead, and can accurately identify complex polygon defects.
Smart Images

Figure CN119047205A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to an anti-analysis identification method, in particular to an anti-analysis identification method for internal polygonal defects of a structure, and belongs to the technical field of engineering structure safety monitoring. Background Art
[0002] Cracks, holes, inclusions and other defects are common defects in structures. These defects will affect the strength of the structure to a certain extent and affect its service life. However, compared with surface defects of the structure, internal defects of the structure are more difficult to be observed by people.
[0003] With the improvement of non-destructive defect detection technology, there are technical solutions in the existing technology that use numerical calculation methods such as finite element and extended finite element combined with optimization algorithms or deep learning to locate and detect internal defects in structures. However, the inversion is currently mainly performed on simple shapes such as circles and ellipses that are expressed by geometric functions. There is no mature solution for the identification of polygonal defects expressed by piecewise functions of geometric features such as triangles and quadrilaterals. Summary of the invention
[0004] Based on the above background, the purpose of the present invention is to provide a method for reverse analysis and identification of polygonal defects inside a structure, by constructing a calculation model corresponding to the actual structure, randomly generating initial defect parameters, and performing defect inversion based on the continuous-discontinuous field function calculation capability of the manifold element method combined with an optimization algorithm to solve the problems described in the background technology.
[0005] In order to achieve the above-mentioned object of the invention, the present invention provides the following technical solutions:
[0006] A method for reverse analysis and identification of internal polygonal defects of a structure, the method comprising the following steps:
[0007] The present invention provides a method for reverse analysis and identification of polygonal defects inside a structure, the method comprising the following steps: arranging monitoring points on a structure containing polygonal defects to obtain actual response monitoring values of the structure;
[0008] According to the possible distribution range of defects, a certain number of defect samples are randomly generated, and a calculation model corresponding to the actual structure is constructed based on them. The response value of the calculation model is obtained by the manifold element method.
[0009] The optimization algorithm is used for defect inversion, with the purpose of making the calculated response value of the manifold element calculation model close to the actual response value, and iterating the position and shape parameters of the defect; a new manifold element calculation model is generated according to the iterative result, and compared with the actual monitoring value, so as to evaluate the accuracy of the iterative result. The above steps are repeated. If the accuracy meets the preset requirements, the iterative defect position and shape parameters are determined to be the inversion values of the structural polygon defect position and shape, and the defect inversion operation is terminated.
[0010] Preferably, the structure is a concrete specimen, a rock block or a metal plate, and the response value is a displacement value, a strain value or an acceleration value.
[0011] Preferably, the constructing of a calculation model corresponding to the actual structure specifically includes:
[0012] Establish a geometric model based on the size parameters of the actual structure;
[0013] Material parameters are specified for the geometric model, boundary conditions are applied at corresponding positions of the geometric model according to the stress state and constraint conditions of the actual structure, and monitoring points for extracting structural response information are arranged on the geometric model according to the monitoring point arrangement positions of the actual structure.
[0014] Preferably, the step of calculating the response value of the calculation model by the manifold element method specifically includes:
[0015] Performing computational meshing on the geometric model, and establishing multiple manifold units, multiple mathematical covers and multiple physical covers based on the mesh according to the manifold element method;
[0016] Identifying manifold units containing defective boundaries from the manifold units, and dividing the manifold units containing the boundaries according to a set resolution to obtain a plurality of integral voxels;
[0017] Determine the voxel value of the integral voxel containing the boundary as the first voxel value; determine the voxel value of the integral voxel not containing the boundary as the second voxel value; and mark the physical coverage corresponding to the manifold unit containing the boundary as the physical coverage to be updated;
[0018] Dividing the physical cover to be updated according to the integral voxels corresponding to the second voxel value in the physical cover to be updated, and determining the connected domain included in the physical cover to be updated;
[0019] Obtaining an updated physical coverage using a numerical manifold method according to the connected domain;
[0020] According to the mathematical cover corresponding to the manifold unit containing the boundary and the updated physical cover, a global continuous-discontinuous test function is established using manifold elements;
[0021] According to the global continuous-discontinuous test function, a linear equation group is established by adopting the minimum potential energy principle, and the displacement field distribution of the calculation model is obtained by solving the equation group;
[0022] Determine the strain, stress, and acceleration distribution of the analysis object using continuum mechanics theory according to the displacement distribution of the analysis object;
[0023] Input the coordinate value of the measuring point and obtain the response value of the monitoring point from the displacement, strain, stress, and acceleration field functions.
[0024] Preferably, the defect inversion is performed using an optimization algorithm, specifically comprising:
[0025] Define the objective function to quantify the deviation between the calculated response value and the actual response value, and set the convergence threshold;
[0026] Randomly generate multiple groups of defect parameters as the initial population of the optimization algorithm;
[0027] For each set of defect parameters, the geometric information of the calculation model is updated according to the defect position and shape parameters represented by it, and the response value of the measurement point of the updated calculation model is calculated, and the objective function value is calculated accordingly;
[0028] Sort multiple groups of defect parameters according to the objective function value, intercept the defect parameter combination with the smaller deviation between the calculated response value and the actual response value, and narrow the defect parameter search range;
[0029] Randomly generate multiple groups of defect parameters within the reduced defect parameter search range, and cyclically calculate their response values and objective function values until the objective function value is less than the set convergence threshold;
[0030] The defect parameter with the smallest objective function is the defect inversion value of the actual structure.
[0031] Compared with the prior art, the present invention has the following advantages:
[0032] The invention discloses a method for inverse analysis and identification of polygonal defects inside a structure, which can flexibly simulate polygonal defects and improve the identification accuracy of polygonal defects inside the structure; an improved grey wolf algorithm is introduced to perform defect inversion, effectively exploring the solution space of defect parameters and improving the global convergence and robustness of the inversion; in the iterative process, each time a new structural calculation model is generated, there is no need to re-divide the grid, which reduces unnecessary calculation overhead and improves calculation efficiency. BRIEF DESCRIPTION OF THE DRAWINGS
[0033] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.
[0034] Figure 1 It is a structural schematic diagram of a method for reverse analysis and identification of internal polygonal defects of a structure according to the present invention;
[0035] Figure 2 is an example diagram of the manifold element calculation model in the present invention;
[0036] Figure 3 This is an example diagram comparing the structural calculation model and the actual structure in the present invention. DETAILED DESCRIPTION
[0037] The technical solution of the present invention is further described in detail below through specific embodiments and in conjunction with the accompanying drawings. It should be understood that the implementation of the present invention is not limited to the following embodiments, and any form of modification and / or change made to the present invention will fall within the protection scope of the present invention.
[0038] In the present invention, unless otherwise specified, all parts and percentages are weight units, and the equipment and raw materials used can be purchased from the market or are commonly used in the art. The methods in the following embodiments, unless otherwise specified, are conventional methods in the art. The components or equipment in the following embodiments, unless otherwise specified, are universal standard parts or components known to those skilled in the art, and their structures and principles are known to those skilled in the art through technical manuals or conventional experimental methods.
[0039] The embodiments of the present invention are described in detail below in conjunction with the accompanying drawings. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present invention. However, one or more embodiments may also be implemented by those skilled in the art without these specific details.
[0040] The embodiment of the present invention discloses a method for reverse analysis and identification of internal polygonal defects of a structure. Figure 1 As shown, the method comprises the following steps:
[0041] S1. Arrange monitoring points on the structure with polygonal defects and obtain the actual response monitoring value of the structure by applying loads;
[0042] S2. According to the possible distribution range of defects, a certain number of defect samples are randomly generated, and a calculation model corresponding to the actual structure is constructed based on the samples, and the response value of the calculation model is obtained by the manifold element method;
[0043] S3, using an optimization algorithm to perform defect inversion, with the goal of making the calculated response value of the manifold element calculation model close to the actual response value, and iterating the position and shape parameters of the defect;
[0044] S4, generating a new manifold element calculation model according to the iteration result, and comparing it with the actual monitoring value, so as to evaluate the accuracy of the iteration result;
[0045] S5, looping the above steps, if the accuracy meets the preset requirements, it is determined that the iterative defect position and shape parameters are the inversion values of the structural polygon defect position and shape, and the defect inversion operation is terminated.
[0046] The structure is a concrete specimen, a rock block or a metal plate, the load is a uniformly distributed load, a concentrated force, a displacement constraint or a vibration load, and the response value is a displacement value, a strain value or an acceleration value.
[0047] Defect detection inside a structure can be described as an inverse problem, in which the defect information can be described by a set of parameter vectors, as shown in the following formula:
[0048]
[0049] In the formula, is the real number domain, and m is the number of polygonal defect parameters to be inverted.
[0050] In the prior art, the inversion object is usually a circular or elliptical defect, but there is a lack of better polygonal defect inversion technology. The reason is that, as an inverse problem, the defect inversion process needs to continuously update the geometric dimensions and re-divide the grid, which is very computationally intensive. The extended finite element method developed in recent years can make its grid independent of the geometric boundary of the analysis object, avoiding the process of re-dividing the grid and significantly reducing the computational cost. However, the extended finite element uses the level set function to establish a discontinuous displacement field. When the geometric information of the analysis object has an analytical expression, such as circular or elliptical defects, the level set function is easy to establish. However, when the expression of the geometric information is very complex, such as polygonal defects, the process of constructing the expression of the level set function and establishing the discontinuous displacements on both sides at the same time will be particularly complicated, especially when there are two polygonal boundaries in one unit, which increases the complexity of the inversion.
[0051] Different from the level set function of the extended finite element method, the numerical manifold element method uses the covering cut method to establish the continuous-discontinuous test function and uses the voxel crack model to track the crack boundary. Therefore, it does not require a simple geometric expression for the geometric information of the analysis object and is more suitable for the inversion of polygonal defects.
[0052] After constructing the calculation model, the method of this embodiment uses the improved gray wolf algorithm to perform defect inversion. The algorithm optimizer that executes the improved gray wolf algorithm will generate predicted defect information according to the limited search range of each parameter, and then transmit the predicted defect information to the manifold element calculation model to generate a structural measurement model, and record the response estimation value of the corresponding monitoring point of the structural measurement model. The objective function is used to calculate the gap between the estimated value of the response of the monitoring point and the true value, and the gap between the two is continuously narrowed to obtain the result closest to the true defect information, that is, the optimal solution.
[0053] Specifically, a computational model corresponding to the actual structure is constructed, and the response value of the computational model is obtained by calculating the manifold element method, which specifically includes:
[0054] S201, establishing a geometric model according to the size parameters of the actual structure;
[0055] S202, specifying material parameters for the geometric model, applying boundary conditions at corresponding positions of the geometric model according to the stress state and constraint conditions of the actual structure, and arranging monitoring points for extracting structural response information on the geometric model according to the monitoring point arrangement positions of the actual structure;
[0056] S203, performing computational meshing on the geometric model, and establishing multiple manifold units, multiple mathematical covers, and multiple physical covers based on the mesh according to the manifold element method;
[0057] S204, identifying manifold units containing defective boundaries from the manifold units, and dividing the manifold units containing the boundaries according to a set resolution to obtain a plurality of integral voxels;
[0058] S205, determining that the voxel value of the integral voxel containing the boundary is a first voxel value; determining that the voxel value of the integral voxel not containing the boundary is a second voxel value; and marking the physical coverage corresponding to the manifold unit containing the boundary as the physical coverage to be updated;
[0059] S206, dividing the physical cover to be updated according to the integral voxels corresponding to the second voxel value in the physical cover to be updated, and determining the connected domain included in the physical cover to be updated;
[0060] S207, obtaining an updated physical coverage using a numerical manifold method according to the connected domain;
[0061] S208, establishing a global continuous-discontinuous test function using manifold elements according to the mathematical cover corresponding to the manifold unit containing the boundary and the updated physical cover;
[0062] S209, according to the global continuous-discontinuous trial function, a linear equation system is established by adopting the minimum potential energy principle, and the displacement field distribution of the calculation model is obtained by solving the equation system;
[0063] S210, determining the strain, stress, and acceleration distribution of the analysis object using continuum mechanics theory according to the displacement distribution of the analysis object;
[0064] S211. Input the coordinate value of the measuring point, and obtain the response value of the monitoring point from the displacement, strain, stress, and acceleration field functions.
[0065] The above process is suitable for computer implementation and can achieve effective inversion of geometric parameters of two-dimensional polygonal and three-dimensional polyhedral defects, with outstanding advancement.
[0066] Specifically, the optimization algorithm is used for defect inversion, including:
[0067] S301, defining an objective function for quantifying the deviation between a calculated response value and an actual response value, and setting a convergence threshold;
[0068] S302, randomly generating multiple groups of defect parameters as initial groups for the optimization algorithm;
[0069] S303, for each set of defect parameters, according to the defect position and shape parameters represented by the defect parameters, the geometric information of the calculation model is updated, and the measurement point response value of the updated calculation model is calculated, and the objective function value is calculated accordingly;
[0070] S304, sorting multiple groups of defect parameters according to the objective function value, intercepting a defect parameter combination with a smaller deviation between the calculated response value and the actual response value, and narrowing the defect parameter search range;
[0071] S305, randomly generate multiple groups of defect parameters within the reduced defect parameter search range, and cyclically calculate their response values and objective function values until the objective function value is less than the set convergence threshold; the defect parameter with the smallest objective function is the defect inversion value of the actual structure. Among them, the optimization algorithm can be a variety of heuristic search algorithms. In this embodiment, the improved gray wolf algorithm is used as the optimization algorithm. Therefore, when using the improved gray wolf algorithm for defect inversion, it specifically includes the following steps:
[0072] Define the objective function to quantify the deviation between the calculated response value and the actual response value, and set the algorithm convergence condition;
[0073] S randomly generates defect parameter combinations as the initial population of the improved grey wolf algorithm, so that each grey wolf represents a possible defect position and shape parameter;
[0074] For each gray wolf, according to the defect position and shape parameters it represents, the geometric information in the calculation model is updated, the measurement point response value of the updated calculation model is calculated, and the objective function value is calculated accordingly;
[0075] Sort the gray wolves according to their fitness values;
[0076] Implement the search strategy of the improved grey wolf algorithm;
[0077] According to the search strategy, the defect position and shape parameters of each gray wolf representative are updated to form a new generation of gray wolf groups;
[0078] Repeat the above steps until the algorithm convergence condition is met and the final gray wolf group is obtained;
[0079] From the final gray wolf population, select the gray wolf individual with the highest fitness, and use its corresponding defect position and shape parameters as the inversion optimal solution;
[0080] According to the optimal solution, the defect position and shape in the calculation model are updated to generate a new structural calculation model.
[0081] Among them, the expression of the objective function is:
[0082]
[0083] In the formula, ε M is the measurement point response value under the real defect, ε C is the measurement point response value corresponding to each iteration result, is the defect parameter vector obtained by inversion.
[0084] Specifically, the gray wolves are sorted according to their fitness values, including:
[0085] Save the positions of the alpha, beta, and delta alpha Use the following formula to update the position vectors of all gray wolf individuals:
[0086]
[0087]
[0088]
[0089] In the formula, Represents the individual gray wolf The distance vector between the alpha, beta, and delta wolves, and is the coefficient vector, are random vectors uniformly distributed on [0,1].
[0090] The gray wolf algorithm divides the wolf pack into four levels: α, β, δ, and ω. The first three wolves α, β, and δ are the leaders of the wolf pack. They are more familiar with the location of the prey and guide the wolf pack to update its own position, gradually approach the prey, and finally capture the prey. This optimization method is likely to reduce the diversity of the population and cause the wolf pack to fall into a local optimum. In order to increase the diversity of the wolf pack and make the wolf pack jump out of the local optimum, this embodiment adopts an improved gray wolf algorithm, which changes the update method of the gray wolf position based on the traditional gray wolf algorithm.
[0091] Specifically, the search strategy of the improved grey wolf algorithm is implemented, including:
[0092] The following formula is used to calculate the t-th generation of gray wolves The gray wolf individuals and the gray wolf of the t+1 generation The Euclidean distance R(t) between the corresponding gray wolf individuals,
[0093]
[0094] Building a wolf pack Neighborhood, calculate The i-th wolf in the gray wolf population The Euclidean distance D between all gray wolf individuals i , and then use the following formula to select The i-th wolf in the gray wolf population Neighbors These gray wolf neighbors constitute Neighborhood
[0095]
[0096] Use the following formula to perform multi-neighborhood learning to generate a new set of gray wolf locations
[0097]
[0098]
[0099] In the formula, is the neighborhood of the i-th gray wolf The d-th dimension parameter (d∈[1,m]) of any neighbor selected in for The d-th dimension parameter of any gray wolf in the gray wolf population, r c is a random number between 0 and 1;
[0100] calculate and The fitness value of the gray wolf is updated using the following rule:
[0101]
[0102] Thus, the position of the gray wolf is generated in the same way as the traditional gray wolf algorithm. As a set of candidate solutions for the new position of the gray wolf, another set of candidate solutions for the gray wolf position was added through another new approach Compare the objective function values of the two sets of candidate solutions, that is, the fitness values, and finally determine the updated position of the gray wolf. The position of the α wolf obtained in each iteration is the optimal solution to the defect inversion problem. When the population size and the number of iterations are the same, the improved gray wolf algorithm has a better effect on defect inversion.
[0103] According to the method of this embodiment, the following Figure 2 The manifold element calculation model shown in the figure takes a structure with triangular defects as an example. The number of monitoring points is selected as 9 and the layout is shown in the figure. The defect inversion is performed by improving the gray wolf algorithm. The number of gray wolves is 50 and the number of population iterations is 300. The following is obtained: Figure 3 The new structural calculation model is shown in the figure. The solid line represents the actual defect position and the dotted line represents the inversion result. It can be seen that the inversion result can accurately reflect the actual defect position and shape of the structure.
[0104] For ease of explanation and understanding, this embodiment introduces the implementation method in a two-dimensional situation, but the method is also applicable to a three-dimensional situation. Although this embodiment uses a triangular defect as an example for explanation, the method is also suitable for inversion identification of other polygonal defects containing a sharp point, such as a quadrilateral, and the method is also suitable for inversion identification of circular and elliptical defects.
[0105] Compared with the existing technology, the reverse analysis and identification method of internal polygonal defects of the structure has the following advantages:
[0106] The prior art usually uses a simplified defect model, which is difficult to accurately characterize complex polygonal defects. The method disclosed in this embodiment adopts the manifold element covering cutting theory, which can flexibly simulate polygonal defects and improve the recognition accuracy of polygonal defects inside the structure.
[0107] The defect inversion of the prior art is prone to fall into local optimum. The method disclosed in this embodiment introduces an improved grey wolf algorithm for defect inversion, effectively explores the solution space of defect parameters, and improves the global convergence and robustness of the inversion.
[0108] In the iterative process of the method disclosed in this embodiment, the structure calculation model generated each time does not need to be re-grid, which reduces unnecessary calculation overhead and improves calculation efficiency.
[0109] The principles and implementation methods of the present invention are described in this article using specific examples. The description of the above embodiments is only used to help understand the method and core ideas of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made to the present invention without departing from the principles of the present invention, and these improvements and modifications also fall within the scope of protection of the claims of the present invention.
Claims
1. A method for reverse analysis and identification of internal polygonal defects of a structure, characterized by: The method comprises the following steps: Arrange monitoring points on the structure with polygonal defects to obtain the actual response monitoring value of the structure; According to the possible distribution range of defects, a certain number of defect samples are randomly generated, and a calculation model corresponding to the actual structure is constructed based on them. The response value of the calculation model is obtained by the manifold element method. The optimization algorithm is used to perform defect inversion, so as to make the calculated response value of the manifold element calculation model close to the actual response value, and iterate the position and shape parameters of the defect; Generate a new manifold element calculation model based on the iteration results and compare it with the actual monitoring value to evaluate the accuracy of the iteration results; The above steps are repeated. If the accuracy meets the preset requirements, the iterative defect position and shape parameters are determined to be the inversion values of the structural polygon defect position and shape, and the defect inversion operation is terminated.
2. The method for reverse analysis and identification of internal polygonal defects of a structure according to claim 1, characterized in that: The structure is a concrete specimen, a rock block or a metal plate, and the response value is a displacement value, a strain value or an acceleration value.
3. The method for reverse analysis and identification of internal polygonal defects of a structure according to claim 1, characterized in that: The constructing of a calculation model corresponding to the actual structure specifically includes: Establish a geometric model based on the size parameters of the actual structure; Material parameters are specified for the geometric model, boundary conditions are applied at corresponding positions of the geometric model according to the stress state and constraint conditions of the actual structure, and monitoring points for extracting structural response information are arranged on the geometric model according to the monitoring point arrangement positions of the actual structure.
4. The method for reverse analysis and identification of internal polygonal defects of a structure according to claim 3, characterized in that: The step of obtaining the response value of the calculation model by calculating the manifold element method specifically includes: Performing computational meshing on the geometric model, and establishing multiple manifold units, multiple mathematical covers and multiple physical covers based on the mesh according to the manifold element method; Identifying manifold units containing defective boundaries from the manifold units, and dividing the manifold units containing the boundaries according to a set resolution to obtain a plurality of integral voxels; Determine the voxel value of the integral voxel containing the boundary as the first voxel value; determine the voxel value of the integral voxel not containing the boundary as the second voxel value; and mark the physical coverage corresponding to the manifold unit containing the boundary as the physical coverage to be updated; Dividing the physical cover to be updated according to the integral voxels corresponding to the second voxel value in the physical cover to be updated, and determining the connected domain included in the physical cover to be updated; Obtaining an updated physical coverage using a numerical manifold method according to the connected domain; According to the mathematical cover corresponding to the manifold unit containing the boundary and the updated physical cover, a global continuous-discontinuous test function is established using manifold elements; According to the global continuous-discontinuous test function, a linear equation group is established by adopting the minimum potential energy principle, and the displacement field distribution of the calculation model is obtained by solving the equation group; Determine the strain, stress, and acceleration distribution of the analysis object using continuum mechanics theory according to the displacement distribution of the analysis object; Input the coordinate value of the measuring point and obtain the response value of the monitoring point from the displacement, strain, stress, and acceleration field functions.
5. The method for reverse analysis and identification of internal polygonal defects of a structure according to claim 1, characterized in that: The defect inversion using the optimization algorithm specifically includes: Define the objective function to quantify the deviation between the calculated response value and the actual response value, and set the convergence threshold; Randomly generate multiple groups of defect parameters as the initial population of the optimization algorithm; For each set of defect parameters, the geometric information of the calculation model is updated according to the defect position and shape parameters represented by it, and the response value of the measurement point of the updated calculation model is calculated, and the objective function value is calculated accordingly; Sort multiple groups of defect parameters according to the objective function value, intercept the defect parameter combination with the smaller deviation between the calculated response value and the actual response value, and narrow the defect parameter search range; Randomly generate multiple groups of defect parameters within the reduced defect parameter search range, and cyclically calculate their response values and objective function values until the objective function value is less than the set convergence threshold; The defect parameter with the smallest objective function is the defect inversion value of the actual structure.
