Wing regular metal structure stress simulation data dimension reduction and correction method

Abstract

The invention discloses a wing regular metal structure stress simulation data dimensionality reduction and correction method, which comprises the following steps: firstly, carrying out wing regular metal structure finite element simulation, respectively obtaining data such as grid nodes and node stress values under dense grid and sparse grid conditions, and completing data preparation; carrying out data clustering analysis by utilizing a distance-based clustering algorithm, calculating the distance between each nearest neighbor point and the central point, and obtaining a plurality of nearest neighbor points with the nearest distance and node stress values corresponding to the nearest neighbor points; calculating the stress value of the target point according to the distance weight and the node stress of each nearest neighbor point; correcting the stress value of the target point by using the maximum value of the stress value of the target point and the relative error of the concentrated stress to obtain a corrected stress value; and finally, calculating a relative error between the concentrated stress and the maximum stress of the structure as a second relative error, and correcting the corrected stress value again. According to the method, dimension reduction and correction of large-scale and high-dimension finite element simulation data can be realized, and the accuracy degree is high.

Description

Technical Field

[0001] The present invention belongs to the field of digital twin virtual model construction, and specifically relates to a dimensionality reduction and correction method for stress simulation data of a regular metal structure of a wing, a computer device, a computer-readable storage medium, and a computer program product. Background Art

[0002] The digital twin virtual model is composed of a geometric model, a simulation model, a rule model, and a data model (a proxy model or a reduced-order model obtained through training with simulation or test data), among which the data model is the basis for building a virtual model. When establishing a data model of a wing regular metal structure (such as a wall panel structure, a lap ear, a long stringer structure, and other structures with relatively uniform overall stress distribution, fewer stress concentration sites, or a lighter stress concentration), some structures contain holes and other features that require mesh refinement or the overall mesh division of the simulation model is relatively dense to ensure the accuracy of the calculation, resulting in a large amount of simulation data. A large amount of simulation data can ensure the fidelity of the virtual model and improve the predictive ability of the virtual model, but when the amount of data exceeds a certain level, not only will the performance of the virtual model decline, but the operating speed of the entire system will also be affected, so it is necessary to reduce the dimension of the data used to establish the data model. However, simply replacing dense grids with sparse grids cannot guarantee the accuracy of finite element simulation results. Therefore, dimensionality reduction and correction of finite element simulation data of wing regular metal structure is the most basic and important link in constructing the wing regular metal structure data model. It is necessary to reduce the dimensionality of stress simulation data while ensuring the accuracy of calculation results, and reduce the dense grid to a sparse grid to achieve data compression and optimization.

[0003] The existing technologies mainly related to data dimension reduction / order reduction have the following problems:

[0004] (1) Most of the technologies that use clustering algorithms as data classification algorithms are combined with eigenvalue decomposition / principal component analysis. These methods are relatively mature and have high computational efficiency, but they have certain limitations when processing large-scale, high-dimensional data.

[0005] (2) The technology that uses neural networks as the main data dimensionality reduction method requires a relatively obvious linear / nonlinear relationship between the input set and the output set. However, in the dimensionality reduction of finite element stress simulation data, the load conditions are used as the input set and the stresses of multiple nodes are used as the output set. This is a single-input and multiple-output system. It is difficult to train a more accurate model through the neural network algorithm, and the dimensionality reduction results have large errors.

[0006] (3) In distance-based weighted interpolation, the commonly used method at home and abroad is the inverse distance weighted interpolation algorithm (IDW), which is a spatial interpolation technology used in geographic information systems and environmental sciences to estimate the missing value of any location based on the values ​​of nearby locations. Since geographic spatial coordinates have a strong physical meaning, there are almost no coordinate points with a distance of 0. However, in the dimensionality reduction of finite element stress simulation data, there may be coordinate points with a distance of 0 in dense grids and sparse grids. When calculating the weight, the denominator is 0, resulting in errors in the entire algorithm. Summary of the invention

[0007] The purpose of the present invention is to provide a method for dimensionality reduction and correction of stress simulation data of regular metal structures of wings, a computer device, a computer-readable storage medium and a computer program product, which can realize dimensionality reduction and correction of large-scale, high-dimensional finite element simulation data.

[0008] One aspect of the present invention provides a method for reducing and correcting dimensionality of stress simulation data of a regular metal structure of a wing, comprising:

[0009] Step S1, dividing the finite element model of the regular metal structure of the wing into a dense grid to obtain a first grid node coordinate set and performing stress simulation analysis to obtain a node stress value of each node in the dense grid, dividing the finite element model of the regular metal structure of the wing into a sparse grid to obtain a second grid node coordinate set and performing stress simulation analysis to obtain a node stress value of each node in the sparse grid;

[0010] Step S2, taking each node in the second grid node coordinate set as the center point, obtaining the nearest neighbor point of each center point in the first grid node coordinate set by a cluster analysis algorithm, calculating the distance between each nearest neighbor point and the center point, and obtaining multiple nearest neighbor points with the shortest distance and their corresponding node stress values;

[0011] Step S3, taking the center point as the target point, calculating the sum of the distances from the target point to each of the multiple nearest neighbor points as the total distance, calculating the distance weight according to the total distance and the distance from the target point to each nearest neighbor point, and calculating the stress value of the target point according to the distance weight and the node stress of each nearest neighbor point;

[0012] Step S4, calculating the nominal stress according to the mechanical mechanism formula, obtaining the concentrated stress according to the nominal stress, calculating the relative error between the maximum value of the stress value of the target point and the concentrated stress as the first relative error, when the first relative error is greater than the first predetermined value, reducing the number of the nearest neighbor points obtained in step S2 and repeating step S3 until the first relative error is greater than or equal to the first predetermined value, and correcting the stress value of the target point using the first relative error to obtain a corrected stress value;

[0013] Step S5, affixing a strain sensor to the regular metal structure of the wing, calculating the maximum stress of the structure in the measured stress field according to the measured stress at the affixing point of the strain sensor, the stress value corresponding to the affixing position coordinate of the strain sensor in the sparse grid, and the maximum stress value of the target point, calculating the relative error between the concentrated stress and the maximum stress of the structure as the second relative error, when the second relative error is greater than the second predetermined value, correcting the corrected stress value again according to the second relative error, and taking the corrected stress value as the full-field stress value, when the second relative error is less than or equal to the second predetermined value, directly taking the corrected stress value as the full-field stress value.

[0014] Another aspect of the present invention provides a computer device, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the above method.

[0015] Another aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, which implements the steps of the above method when executed by a processor.

[0016] Another aspect of the present invention provides a computer program product, comprising a computer program, which implements the steps of the above method when executed by a processor.

[0017] The dimensionality reduction and correction method, computer device, computer readable storage medium and computer program product of wing regular metal structure stress simulation data according to the above aspects of the present invention can achieve dimensionality reduction and correction of large-scale, high-dimensional finite element simulation data. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solution of the present invention, the following briefly introduces the drawings used in the description of the embodiments of the present invention. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative work:

[0019] Figure 1 It is a flow chart of a method for reducing dimension and correcting stress simulation data of a regular metal structure of a wing according to an embodiment of the present invention;

[0020] Figure 2 It is a schematic diagram of a method for reducing and correcting the dimensionality of a wing regular metal structure stress simulation data according to an embodiment of the present invention;

[0021] Figure 3 is a schematic diagram of a finite element simulation dense grid according to an embodiment of the present invention;

[0022] Figure 4is a schematic diagram of a finite element simulation sparse grid according to an embodiment of the present invention;

[0023] Figure 5 It is a structural diagram of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0024] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0025] An embodiment of the present invention provides a method for reducing and correcting the stress simulation data of a regular metal structure of a wing, which is mainly applied to reducing and correcting the stress simulation data of metal elements with regular geometric shapes and simple structures in the wing (such as wall panels, lap ears, long stringers, etc., with relatively uniform overall stress distribution, fewer stress concentration sites or less stress concentration) when building a digital twin model. Figure 1 As shown, the method of the embodiment of the present invention includes steps S1 to S5. Figure 2 The schematic diagram of the embodiment of the present invention is used to describe in detail each step of the method.

[0026] Step S1 is a data preparation step. In step S1, the finite element model of the wing regular metal structure is first divided into dense grids (such as Figure 3 As shown, the number of grids is determined according to the actual simulation requirements) to obtain the grid node coordinate set A, where A i (i = 1, 2, 3, ...) represents the nodes in set A, and performs stress simulation analysis to obtain the node stress simulation results under dense grids, and obtain the node stress value of each node. Then, the finite element model of the wing regular metal structure is divided into sparse grids (such as Figure 4 As shown in the figure, the number of grids is determined according to the compatibility of the digital twin visualization platform with the number of grids) to obtain the grid node coordinate set B, where B i (i=1, 2, 3, ...) represents the nodes in set B, and stress simulation analysis is performed to obtain the node stress value of each node under the sparse grid, completing the data preparation.

[0027] Step S2 is a cluster analysis step. In step S2, each node B in set B is i is the center point, A is the data set to be clustered, and B is obtained in A through cluster analysis algorithm. i A set of (e.g. 15) nearest neighbor points. According to formula (1), calculate each nearest neighbor point A iWith B i and obtain the nearest multiple (e.g., 8) nearest neighboring points and their corresponding node stress values;

[0028]

[0029] Where, L i A is the nearest neighbor i With center point B i The Euclidean distance of

[0030] x Ai ,y Ai 、z Ai is the coordinate of the nearest neighbor point Ai;

[0031] x Bi ,y Bi 、z Bi is the coordinate of the center point Bi.

[0032] Step S3 is a distance-based weighted interpolation step. In step S3, according to formula (2), the center point B i For the target point, calculate the distance L from the target point to multiple (for example, 8) nearest neighbor points i And the total distance L; L = L1 + L2 + L3 + ... + L8 (2)

[0033] According to formula (3), the total distance L and the distance L from the target point to each nearest neighbor point i Calculate the distance weight d i ;

[0034]

[0035] According to formula (4), the distance weight di and the node stress σ of each nearest neighbor i Calculate target point B i The stress value σ Bi

[0036] σ Bi =d i *σ i (4)

[0037] Step S4 is a step of evaluating the interpolation calculation results and correcting the stress based on the mechanism formula. In step S4, the nominal stress σ is calculated according to the mechanical mechanism formula. M = load / bearing area, based on nominal stress σ M Calculation of concentrated stress σ Max =K*σ M (K is the stress intensity factor).

[0038] According to formula (5), the relative error c (first relative error) between the maximum value of the target point stress value (the maximum value of the target point stress) and the concentrated stress is calculated;

[0039]

[0040] In the formula, σ Bimax The target point B i The maximum value of stress value;

[0041] σ Max For concentrated stress.

[0042] When the first relative error c> the first predetermined value (for example, 2%), it indicates that the target point B obtained in step S2 i For example, among the eight nearest neighboring points, there are coordinates of the stress concentration site, which affects the stress calculation of the target point. Therefore, the search range is narrowed to retain, for example, six points for calculation, and if the relative error c is still greater than the first predetermined value, four points are retained for calculation, and so on (generally, the relative error c is less than or equal to 2%, and this operation is only to eliminate the influence of data errors caused by abnormal file saving); when the first relative error c≤the first predetermined value, the stress value based on the mechanism formula is corrected according to formula (6);

[0043] σ Bi(修) =σ Bi *(1-c) (6)

[0044] Step S5 is a step of correcting stress based on measured data. In step S5, Figure 5 As shown in the figure, a strain sensor is attached to the regular metal structure of the wing, and the measured stress σ at the sensor attachment point is calculated using formula (7) (测) :

[0045] σ (测) =ε*E (7)

[0046] In the formula, σ (测) is the measured stress at the sensor attachment point;

[0047] ε is the measured strain at the sensor attachment point;

[0048] E is the elastic modulus of the structural material.

[0049] According to the measured stress σ at the point where the strain sensor is pasted (测) , strain sensor pasting position coordinates (x, y, 0), stress value σ corresponding to (x, y, 0) in the sparse grid B(x，y，0) , the maximum stress value of the target point σ Bimax And formula (8) calculates the maximum structural stress in the measured stress field:

[0050]

[0051] According to formula (9), the second relative error d between the concentrated stress and the maximum stress of the structure is calculated;

[0052]

[0053] When the second relative error d> the second predetermined value (for example, 2%), according to formula (10), Bi(修) Correct again, with σ Bi(实测修正) As the full-field stress value; when the relative error d≤ the second predetermined value, directly use σ Bi(修) As the full-field stress value, the dimension reduction and correction of the regular metal structure stress simulation model data are completed.

[0054] σ Bi(实测修正) =σ Bi(修) *(1-d) (10)

[0055] The method of the embodiment of the present invention is further described in detail below through a specific example.

[0056] In a specific embodiment, taking the hole-containing part of the wing stringer as an example, a finite element simulation model is constructed, and a dense grid (the number of grid nodes is 64181) is as follows: Figure 3 As shown in the figure, the sparse grid (the number of grid nodes is 7773) is as follows Figure 4 As shown in the figure, mesh subdivision is performed at the hole position. The specific steps of data dimension reduction are as follows:

[0057] Step S1: First, export the dense grid to obtain the grid node coordinate set A, and then export the sparse grid to obtain the grid node coordinate set B, thus completing data preparation;

[0058] Step S2: i As the center point, A i For the data set to be clustered, obtain B in Ai through the clustering analysis algorithm i According to formula (1), the relationship between each nearest neighbor point and B is calculated. i The distance is calculated and the 8 nearest neighboring points and their corresponding node stress values ​​are obtained;

[0059]

[0060] Step S3: According to formula (2), i For the target point, calculate the distance L to the 8 nearest neighbor points i and total distance L;

[0061] L=L1+L2+L3+...+L8 (2)

[0062] According to formula (3), the total distance L and the distance L from the target point to the eight nearest neighboring points i Calculate the distance weight d i ;

[0063]

[0064] According to formula (4), the distance weight d i and the nodal stresses σ of the 8 nearest neighboring points i Calculate target point B i The stress value σ Bi ;

[0065] σ Bi =d i *σ i (4)

[0066] Step S4: With a bearing area of ​​0.0004473m 2 , taking load 100kN as an example, the nominal stress σ is calculated according to the mechanical mechanism formula M = load / bearing area = 223563603.85Pa, then the concentrated stress σ Max =2.8*σ M =625978090.8Pa (K=2.8, obtained by theoretical analysis);

[0067] σ Bimax =620000000Pa, according to formula (5), the maximum value of the target point stress value and the first relative error of the concentrated stress c = -0.96%;

[0068]

[0069] When the relative error c>2%, it means that the B obtained in step S2 i The coordinates of the stress concentration point exist in the 8 nearest neighboring points, which affects the stress calculation of the target point. The search range is narrowed to retain 6 points for calculation. If the relative error c is still greater than 2%, 4 points are retained for calculation, and so on (generally the relative error c is less than or equal to 2%, and this operation is only to eliminate the influence of data errors caused by abnormal file saving); when the relative error c≤2%, the stress value based on the mechanism formula is corrected according to formula (6);

[0070] σ Bi(修) =σ Bi *(1-c)=625952000Pa (6)

[0071] Step S5: The wing stringer material is 7075 aluminum alloy, and the elastic modulus is 69.2 GPa. If the measured microstrain at the strain sensor attachment point is 3.0977, the measured stress σ at the sensor attachment point is calculated using formula (7): (测) :

[0072] σ (测) =ε*R=214360840Pa (7)

[0073] According to the measured stress σ at the point where the strain sensor is attached (测) =214360840Pa, the stress value corresponding to the location where the strain sensor is pasted in the sparse grid is 214520000Pa, and the maximum stress value of the target point is σ Bimax =620000000Pa and formula (8) to calculate the maximum structural stress in the measured stress field:

[0074]

[0075] According to formula (9), the relative error d between the concentrated stress at the hole edge and the maximum stress of the structure is calculated;

[0076]

[0077] When the relative error d≤2%, directly use σ Bi(修) as the full-field stress value.

[0078] In summary, the method of the embodiment of the present invention uses data such as finite element mesh nodes and node stress values ​​in the dense stress simulation model of the regular metal structure of the wing as the data set to be classified, and takes each node in the sparse finite element mesh nodes of the regular metal structure of the wing as the center point to complete data preparation; uses a distance-based clustering algorithm to cluster the data to obtain dense finite element mesh nodes and node stress values ​​near the center point; performs node stress interpolation calculation according to the distance weight to obtain the center point stress interpolation result; and evaluates the interpolation calculation results and makes corrections based on the mechanism formula and the measured data.

[0079] According to the characteristics of the digital twin virtual model, the method of the embodiment of the present invention fuses the mechanical mechanism formula, the measured data of the sensor and the calculation results, and corrects the center point stress interpolation result. This method is based on the clustering algorithm, and replaces the principal component analysis method used in the prior art with the distance weight interpolation. It can not only reduce the dimension of less data, but also reduce the dimension of the data of the large-scale, high-dimensional wing regular metal structure stress simulation model, without the need to train the neural network model, and has a high degree of accuracy. In terms of the distance weight algorithm, the inverse distance weight interpolation algorithm of the prior art is replaced by a simpler and easier to understand distance weight interpolation, avoiding the calculation result error caused by the distance being 0. The correction method based on the measured data of the embodiment of the present invention is different from the data fusion algorithm such as Kalman filtering and Bayesian theory commonly used in digital twin technology that requires data sequences. It can timely correct the single-point stress calculation results, and at the same time, for test pieces with different sensor pasting positions, the relevant calculation parameters can be quickly modified, and it has strong applicability. Example verification shows that the time required to train the proxy model after data dimensionality reduction using the method of the embodiment of the present invention is greatly reduced compared with the prior art.

[0080] An embodiment of the present invention further provides a computer device, which may be a server, and its internal structure diagram may be as shown in FIG. Figure 5 As shown. The computer device includes a processor, a memory and a network interface connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store operating parameter data of each framework. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the steps of the method of the embodiment of the present invention are implemented.

[0081] Those skilled in the art will understand that Figure 5 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0082] An embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the method of the embodiment of the present invention are implemented.

[0083] An embodiment of the present invention further provides a computer program product, including a computer program, which implements the steps of the method of the embodiment of the present invention when executed by a processor.

[0084] The above description is only by way of illustration of certain exemplary embodiments of the present invention. It is undoubted that those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A method for reducing and correcting the dimensionality of stress simulation data of a regular metal structure of a wing, characterized in that: include: Step S1, dividing the finite element model of the regular metal structure of the wing into a dense grid to obtain a first grid node coordinate set and performing stress simulation analysis to obtain a node stress value of each node in the dense grid, dividing the finite element model of the regular metal structure of the wing into a sparse grid to obtain a second grid node coordinate set and performing stress simulation analysis to obtain a node stress value of each node in the sparse grid; Step S2, taking each node in the second grid node coordinate set as the center point, obtaining the nearest neighbor point of each center point in the first grid node coordinate set by a cluster analysis algorithm, calculating the distance between each nearest neighbor point and the center point, and obtaining multiple nearest neighbor points with the shortest distance and their corresponding node stress values; Step S3, taking the center point as the target point, calculating the sum of the distances from the target point to each of the multiple nearest neighbor points as the total distance, calculating the distance weight according to the total distance and the distance from the target point to each nearest neighbor point, and calculating the stress value of the target point according to the distance weight and the node stress of each nearest neighbor point; Step S4, calculating the nominal stress according to the mechanical mechanism formula, obtaining the concentrated stress according to the nominal stress, calculating the relative error between the maximum value of the stress value of the target point and the concentrated stress as the first relative error, when the first relative error is greater than the first predetermined value, reducing the number of the nearest neighbor points obtained in step S2 and repeating step S3 until the first relative error is greater than or equal to the first predetermined value, and correcting the stress value of the target point using the first relative error to obtain a corrected stress value; Step S5, affixing a strain sensor to the regular metal structure of the wing, calculating the maximum stress of the structure in the measured stress field according to the measured stress at the affixing point of the strain sensor, the stress value corresponding to the affixing position coordinate of the strain sensor in the sparse grid, and the maximum stress value of the target point, calculating the relative error between the concentrated stress and the maximum stress of the structure as the second relative error, when the second relative error is greater than the second predetermined value, correcting the corrected stress value again according to the second relative error, and taking the corrected stress value as the full-field stress value, when the second relative error is less than or equal to the second predetermined value, directly taking the corrected stress value as the full-field stress value.

2. The method according to claim 1, characterized in that In step S5, the maximum structural stress σ in the measured stress field is calculated as follows: Among them, σ (测) represents the measured stress at the point where the strain sensor is attached, σ B(x，y，0) represents the stress value corresponding to the coordinates of the strain sensor pasting position in the sparse grid, σ Bimax Indicates the maximum value of the stress value at the target point; The concentration stress σ is calculated as follows Max and the maximum structural stress σ (测max) The relative error is taken as the second relative error d: When the second relative error d> the second predetermined value, the corrected stress value σ is Bi(修) Correct again and get the corrected stress value σ Bi(实测修正) : s Bi(实测修正) =s Bi(修) *(1-d).

3. The method according to claim 2, characterized in that The measured stress σ at the sensor attachment point is calculated as follows (测) : s (测) =e*E Among them, ε represents the measured strain at the sensor attachment point, and E is the elastic modulus of the structural material.

4. The method according to any one of claims 1 to 3, characterized in that In step S4, the nominal stress σ is calculated as follows M and the concentrated stress σ Max : σ M = load / bearing area, σ Max =K*σ M , where K is the stress intensity factor; The maximum stress value of the target point is calculated as follows and the concentrated stress σ Max The relative error is taken as the first relative error c: When the first relative error c≤the first predetermined value, the stress value σ of the target point is corrected using the first relative error c as follows: Bi , and obtain the corrected stress value σ Bi(修) : s Bi(修) =s Bi *(1-c).

5. The method according to any one of claims 1 to 3, characterized in that: In step S3, the distance weight d is calculated as follows i And the stress value σ at the target point Bi : s Bi =d i *s i , Among them, L is the total distance, σ i is the nodal stress for each nearest neighbor point.

6. The method according to any one of claims 1 to 3, characterized in that In step S2, each nearest neighbor point A is calculated as follows i With center point B i The distance L i : Among them, x Ai ,y Ai 、z Ai A is the nearest neighbor i The coordinates, x Bi ,y Bi 、z Bi is the coordinate of the center point Bi.

7. The method according to any one of claims 1 to 3, characterized in that The first predetermined value and the second predetermined value are both 2%.

8. A computer device comprising a memory, a processor and a computer program stored in the memory, characterized in that: The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images