A method for calculating linear elastic problems of thin plate structures

Through the Voronoi structure vertex numbering and tiny offset method, combined with the generalized strain smooth stability method, the node strain and stress of the thin plate structure are directly obtained, which solves the inconvenient post-processing and insufficient calculation accuracy in the linear elasticity problem of the C1 natural unit method. The problem of , realizes efficient numerical solution.

CN116090084BActive Publication Date: 2024-11-22BEIJING INST OF STRUCTURE & ENVIRONMENT ENG
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202211449626.1
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-11-18
Publication Date
2024-11-22
Estimated Expiration
2042-11-18

AI Technical Summary

Technical Problem

The existing C1 natural unit method has problems of inconvenient post-processing in solving the linear elastic problem of thin plate structures, and the calculation accuracy is insufficient.

Method used

The counterclockwise order of Voronoi structure vertex numbering and slight offset coordinate method are used, combined with the generalized strain smooth stability method, the smooth generalized strain-displacement relationship matrix of the triangle subdomain is constructed, and the generalized strain and stress on the node are directly obtained through node integration.

Benefits of technology

It significantly improves the calculation accuracy, simplifies the post-processing of numerical results, avoids the complexity of background grid integral, and realizes efficient solution to the linear elastic problem of thin plate structure.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN116090084B_ABST
    Figure CN116090084B_ABST
Patent Text Reader

Abstract

A method for calculating linear elastic problems of thin plate structures based on the improved C 1 The natural element method is used to solve the linear elastic problem of the thin plate structure, including: S1, establishing a geometric model of the thin plate structure, obtaining the information of all discrete nodes of the thin plate structure, generating Delaunay triangle information based on the discrete nodes, the numbering of the structure boundary nodes and their counterclockwise arrangement order, the node numbering of the load, preparing the geometric dimensions, material parameters, load size information of the structure, and setting the offset coefficient; S2, preparing calculation data; S3, generating the stiffness matrix corresponding to all triangular subdomains, and finally integrating the overall stiffness matrix of the thin plate structure; integrating the overall equivalent load array of the structure according to the equivalent load array of each node; S4, solving the control equation of the thin plate linear elastic problem, and obtaining the node displacement vector of the thin plate structure; S5, finally determining the smooth generalized strain and smooth generalized stress of each node of the thin plate according to the geometry, area and constitutive relations respectively.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to a method for calculating linear elastic problems of thin plate structures, and belongs to the field of numerical calculation methods such as meshless method and computational mechanics. Background Art

[0002] Thin plate structure is a kind of structure widely used in engineering fields such as aviation, aerospace, nuclear energy and shipbuilding. In the engineering design involving thin plate structure, it is usually necessary to adopt a structure that meets C 1 The plate unit with continuity requirement is used to solve the linear elastic problem. 1 Although the thin plate finite element method and meshless method that meet the continuity requirements have excellent performances such as high calculation accuracy, good numerical stability, and flexible use, these methods usually use background grids for numerical integration, which often have the disadvantages of large calculation volume and inconvenient post-processing. 1 It is very necessary to have a numerical method with continuity requirements, high calculation accuracy and convenient post-processing, and apply it to the solution and analysis of linear elastic problems of thin plate structures.

[0003] C 1 Natural Unit Method (C 1 -NEM) is a method that is different from the finite element method and meets C 1 A novel meshless method suitable for analyzing thin plate structures with continuity requirements. 1 The natural neighbor interpolation constructs the shape function. The calculation format is relatively simple. It only requires node information but not node connection information. It does not involve complex matrix operations and any artificial parameters. 1 Shape functions have strict interpolation characteristics for node function values ​​and node gradient values, and can impose essential boundary conditions as conveniently and accurately as finite elements. 1 The background grid integration scheme used in the natural element method has both advantages and disadvantages of this integration scheme and is not an optimal integration scheme. Summary of the invention

[0004] The technical problem to be solved by the present invention is: the present invention proposes a method for calculating the linear elastic problem of thin plate structure, based on the improved C 1 The natural element method solves the linear elastic problem of thin plate structure and solves the C using the background grid integration. 1 The natural element method overcomes the disadvantage of inconvenient post-processing when solving linear elastic problems of plate structures, and significantly improves the calculation accuracy.

[0005] The technical solution adopted by the present invention is: a method for calculating the linear elastic problem of a thin plate structure, comprising:

[0006] S1. Establish a geometric model of the thin plate structure, obtain the numbers and spatial coordinates of all discrete nodes of the thin plate structure, generate the numbers of Delaunay triangles based on the discrete nodes and the numbers of the three nodes that make up each Delaunay triangle vertex, the numbers of the boundary nodes of the thin plate structure and their counterclockwise arrangement order, the numbers of the nodes on which the load acts, prepare the geometric dimensions, material parameters, and load size information of the thin plate structure, and set the offset coefficient;

[0007] S2, prepare calculation data to form a smooth generalized strain-displacement relationship matrix corresponding to all triangular subdomains;

[0008] S3, generating stiffness matrices corresponding to all triangular subdomains, and finally integrating the overall stiffness matrix of the plate structure; integrating the overall equivalent load matrix of the structure according to the equivalent load matrix of each node;

[0009] S4, solving the governing equations of the linear elastic problem of the thin plate structure, and obtaining the node displacement vectors of the thin plate structure;

[0010] S5. According to the geometric relationship, area relationship and constitutive relationship, the smooth generalized strain of the triangular subdomain, the smooth generalized strain of each node and the smooth generalized stress are determined in turn.

[0011] Furthermore, the S2, prepares calculation data to form a smooth generalized strain-displacement relationship matrix corresponding to all triangular subdomains, including:

[0012] S21, generating the numbers and spatial coordinates of all the nodes of the Voronoi structure vertex, and determining the counterclockwise arrangement order of the Voronoi structure vertex numbers according to the angle size;

[0013] S22, using a Voronoi structure vertex coordinate micro-shift method to slightly shift the coordinates of all Voronoi structure vertices;

[0014] S23, divide the Voronoi structure of all nodes into triangular subdomains, use the generalized strain smoothing stability method on each subdomain, and use C 1 The shape function is constructed by natural neighbor interpolation, and the smooth generalized strain-displacement relationship matrix corresponding to all triangular subdomains is generated by numerical integration.

[0015] Further, the S21 generates the numbers and spatial coordinates of all the nodes of the Voronoi structure vertex, and determines the counterclockwise arrangement order of the Voronoi structure vertex numbers according to the angle, including:

[0016] S21a, matching the numbers of all nodes with the node numbers of all Delaunay triangle vertices in sequence, determining and counting the number of times each node is shared by Delaunay triangles and the numbers of the Delaunay triangles that share the node;

[0017] S21b, according to the spatial coordinates of the three nodes constituting the Delaunay triangle, calculate the spatial coordinates of the circumcenters of all Delaunay triangles, and sequentially number the circumcenters in turn; if the circumcenter coordinates of two triangles are the same, their numbers are the circumcenter numbers of the first numbering;

[0018] S21c, according to the spatial coordinates of the two adjacent displacement boundary nodes, the midpoint coordinates of the two adjacent boundary nodes on all displacement boundaries are calculated in sequence, and then the circumcenter numbering in S21b is followed to complete the sequential numbering of all midpoint coordinates;

[0019] S21d, generate the numbers and coordinates of the vertices of the Voronoi structure of all nodes: the vertices of the Voronoi structure of non-boundary nodes are the circumcenters of the Delaunay triangles that share the node, and their coordinates and numbers are generated according to S21a and S21b; the vertices of the Voronoi structure of boundary nodes include the circumcenters of the Delaunay triangles that share the node and the midpoints of the lines connecting the nodes adjacent to the node on the boundary, and their coordinates and numbers are generated according to S21a to S21c;

[0020] S21e, according to the angle between each node Voronoi vertex and the line connecting the node and the horizontal right direction, determine the counterclockwise order of the vertices numbering of the Voronoi structure of all nodes.

[0021] Furthermore, the S22, using a Voronoi structure vertex coordinate micro-shift method to slightly shift the coordinates of all Voronoi structure vertices, includes:

[0022] x′ i =x i +α x ×β×γ (1)

[0023] y′ i =y i +α y ×β×γ (2)

[0024] Among them, (x i ,y i ) is the original coordinate of the Voronoi structure vertex i, (x′ i ,y′ i ) is the offset coordinate of vertex i; α x and α yis the characteristic size of the structure; β is a random number between [-1,1]; γ is the offset coefficient.

[0025] Further, the S23, the Voronoi structure of all nodes is divided into triangular subdomains, and the generalized strain smoothing stability method is used in each subdomain, and C 1 The shape function is constructed by natural neighbor interpolation, and the smooth generalized strain-displacement relationship matrix corresponding to all triangular subdomains is generated by numerical integration, including:

[0026] S23a, in counterclockwise order, select two vertices of the Voronoi structure of each node in turn to form a triangular subdomain with the node: node x r The Voronoi domain is S r , area A r , the triangular subdomain is S rs ,and RS is the total number of triangular subdomains, r = 1 ~ NP, NP is the total number of nodes; then according to the node x r And the spatial coordinates of the two vertices, calculate the triangle subdomain S rs Area A rs , each boundary Γ rs The normal vector n1 and tangent vector n2;

[0027] S23b, using the empty circle criterion in turn, determine the triangle subdomain S rs The three boundaries of Γ rs The number of natural neighboring nodes of each integral point is recorded as np, and the C of each natural neighboring node I is calculated 1 The first derivative of the natural neighbor shape function x rs corresponds to the triangular subdomain S rs Virtual nodes, I = 1 ~ np; where superscript w, θ x ,θ y They represent the deflection of the plate, the rotation in the x direction, and the rotation in the y direction respectively;

[0028] S23c, calculate the smooth generalized strain-displacement relationship matrix corresponding to each integral point's naturally adjacent node I=1~np:

[0029]

[0030] S23d, integrated triangular subdomain S rs All integral points naturally adjacent to the corresponding nodes Matrix, forming a triangular subfield S rs The corresponding smooth generalized strain-displacement relationship matrix

[0031]

[0032] in, It is the total number of nodes after merging all naturally adjacent nodes with the same integral points.

[0033] Furthermore, in S3, the stiffness matrices corresponding to all triangular subdomains are generated, and finally the overall stiffness matrix of the plate structure is integrated; the overall equivalent load matrix of the structure is integrated according to the equivalent load matrix of each node, including:

[0034] S31, generate each triangular subdomain S in sequence rs The corresponding stiffness matrix The overall stiffness matrix K of the final integrated plate structure is:

[0035]

[0036]

[0037] Where D is the elastic relationship matrix, D0 = Eh0 3 / [12(1-v 2 )] is the bending stiffness of the plate, E is Young's modulus, h0 is the plate thickness, and v is Poisson's ratio;

[0038] S32. For the uniformly distributed force, according to the Voronoi domain S r Area A r The uniformly distributed force is equivalent to the lateral force acting on the node; for the concentrated force, it is applied to the acting node; according to the equivalent load array F of each node r The overall equivalent load array F of the integrated structure:

[0039]

[0040] Furthermore, in S4, the control equations for solving the linear elastic problem of the thin plate structure include:

[0041] The displacement boundary conditions are introduced by direct substitution method, diagonal element change to 1 method or diagonal element multiplication by large number method, and the overall stiffness matrix K and the overall load matrix F are modified. The Gaussian sequential elimination method, triangular decomposition method or Gauss-Jordan elimination method are used to solve the control equation Ka=F of the linear elastic problem of the thin plate structure, and the node displacement vector a of the thin plate structure is obtained:

[0042] a=[w1 θ x1 θ y1 w2θ x2 θ y2 … w NP θx(NP) θ y(NP) ] T (8)

[0043] Where w1 is the deflection of node 1, θ x1 is the rotation of node 1x, θ y1 is the rotation of node 1 in the y direction; w2 is the deflection of node 2, θ x2 is the rotation of the node in the 2x direction, θ y2 is the rotation of node 2 in the y direction; ...; w NP is the deflection of node NP, θ x(NP) is the rotation of the node NP in the x direction, θ y(NP) is the rotation of the node NP in the y direction.

[0044] Furthermore, the step S5 determines the smooth generalized strain of the triangular subdomain, the smooth generalized strain of each node and the smooth generalized stress in turn according to the geometric relationship, the area relationship and the constitutive relationship, including:

[0045] According to the geometric relationship, the triangular subdomain S is determined by the node displacement vector and the smooth generalized strain-displacement relationship matrix. rs Smooth generalized strain

[0046]

[0047] According to each triangular subdomain S rs With Voronoi domain S r The area relationship of each node is determined by the smooth generalized strain

[0048]

[0049] According to the constitutive relationship, the smooth generalized stress of each node is calculated

[0050]

[0051] The advantages of the present invention compared with the prior art are:

[0052] (1) The method for determining the counterclockwise order of the Voronoi structure vertex numbers and the method for slightly offsetting the coordinates of the Voronoi structure vertices adopted by the present invention are simple to implement and effectively ensure the sequential division of the triangular subdomains and the integral point C on the edge of the Delaunay triangle. 1 Smooth solution of natural neighboring shape functions.

[0053] (2) The Voronoi structure triangle subdomain partitioning method proposed in the present invention and the generalized strain smoothing and stabilization method used in the subdomain effectively ensure the stability of numerical calculation, significantly improve the calculation accuracy, and only need to calculate C 1 The first derivative of the shape function is naturally approached, without the need for its second derivative.

[0054] (3) The present invention adopts the node integration method, which can directly obtain the generalized strain and generalized stress at the node, without the need for C based on background grid integration. 1 Like the natural element method and the finite element method, it is necessary to interpolate and extrapolate the strains and stresses at the integration points to obtain the corresponding results at the nodes, which greatly facilitates the post-processing of the numerical results.

[0055] (4) C proposed by the present invention 1 The nodal natural element method is highly versatile and can be further applied to nonlinear problems of thin plate structures, providing an effective, accurate, convenient and flexible numerical method for the computational research of thin plate structures. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 Schematic diagram of triangular subdomain division;

[0057] Figure 2 For the circular thin plate with support, Figure 2 (a) is the action diagram of uniformly distributed force q, (b) is the action diagram of concentrated force P, (c) is the schematic diagram of 801 discrete nodes, and (d) is the diagram of 1536 Delaunay triangles.

[0058] Figure 3 The figure is a comparison between the numerical solution and the theoretical solution of the deflection w of the clamped circular thin plate;

[0059] Figure 4 is the rotation of the circular plate in the x direction θ x Comparison diagram between numerical solution and theoretical solution;

[0060] Figure 5 is the rotation of the circular plate in the y direction θ y Comparison diagram between numerical solution and theoretical solution;

[0061] Figure 6 The deflection and generalized strain cloud diagram of a circular thin plate clamped by a uniformly distributed force: Figure 6 In the figure, (a) is the cloud diagram of the deflection w; (b) is the rotation θ in the x direction x (c) is the y-direction rotation θ y (d) is the cloud diagram of the generalized strain in the x direction. (e) is the cloud diagram of the generalized strain in the y direction (f) is the cloud diagram of the generalized strain in the xy direction Cloud map.

[0062] Figure 7 The deflection and generalized strain cloud diagram of a circular thin plate supported by a concentrated force: Figure 7 In the figure, (a) is the cloud diagram of the deflection w; (b) is the rotation θ in the x direction x (c) is the y-direction rotation θ y (d) is the cloud diagram of the generalized strain in the x direction. (e) is the cloud diagram of the generalized strain in the y direction (f) is the cloud diagram of the generalized strain in the xy direction Cloud map. DETAILED DESCRIPTION

[0063] The present invention will be further described below in conjunction with the accompanying drawings.

[0064] In addition to the implementation cases described below, the present invention can also be applied to the implementation of other cases. Therefore, the present invention is not limited to the structures described in the following description or shown in the drawings. Of course, only one implementation case is introduced here, and the claims are not limited to this implementation case.

[0065] The present invention improves C 1 Natural unit method (named C 1 Nodal natural element method, C 1 -nodal-NEM), and based on this method, a method for solving the linear elastic problem of thin plate structure is proposed. In this method, C 1 The shape function of the thin plate structure is constructed by natural neighbor interpolation. Based on the strain smoothing stability and node integration technology, the triangle subdomain stable compatible integration scheme (TSCI) is developed and used for numerical integration. The integration scheme first determines the order of the numbering of each vertex of the Voronoi structure counterclockwise according to the angular relationship between the discrete nodes of the plate structure and each vertex of its Voronoi structure; then, a small offset is made to the coordinates of each Voronoi structure vertex to effectively overcome the C error caused by the integration point being located on the edge of the Delaunay triangle. 1 The difficulty of calculating the shape function of the natural neighbor is that the Voronoi structure is divided into a series of triangular subdomains consisting of the nodes surrounded by the Voronoi structure and any two adjacent vertices of the Voronoi structure. The generalized strain smoothing and numerical integration are used on each triangular subdomain to form the smooth generalized strain-displacement relationship matrix corresponding to all triangular subdomains required for numerical calculation. Finally, according to the calculated thin plate node displacement, as well as the geometric relationship, area relationship and constitutive relationship, the smooth generalized strain and smooth generalized stress of each node of the thin plate are directly calculated.

[0066] The present invention proposes a method for calculating the linear elastic problem of a thin plate structure, and the specific steps are as follows:

[0067] Step 1: Prepare model data

[0068] 1.1) Establish a geometric model of the thin plate structure and discretize it to obtain the numbers of all discrete nodes (the total number of nodes is denoted as NP, which is a positive integer) and spatial coordinates;

[0069] 1.2) Generate Delaunay triangle data according to the spatial coordinates of the discrete nodes, and obtain the Delaunay triangle number and the numbers of the three nodes that make up the triangle vertices;

[0070] 1.3) Extract the node numbers of the displacement boundary of the thin plate structure (arranged in counterclockwise order) and the node numbers of the load action.

[0071] 1.4) Prepare the geometric dimensions (radius, plate thickness), material parameters (Young's modulus, Poisson's ratio), load size information of the thin plate structure, and set the offset coefficient.

[0072] Step 2: Prepare the calculation data

[0073] 2.1) Generate the numbers and spatial coordinates of all the nodes of the Voronoi structure, and determine the counterclockwise order of these numbers based on the angle.

[0074] a) Match the numbers of all nodes with the node numbers of all Delaunay triangle vertices in sequence, determine and count the number of times each node is shared by Delaunay triangles and the numbers of the Delaunay triangles that share the node.

[0075] b) Based on the spatial coordinates of the three nodes that make up the Delaunay triangle, calculate the spatial coordinates of the circumcenters of all Delaunay triangles, and number the circumcenters in sequence; if the circumcenter coordinates of two triangles are the same, their numbers are the circumcenter numbers of the first number.

[0076] c) According to the spatial coordinates of two adjacent displacement boundary nodes, the midpoint coordinates of two adjacent boundary nodes on all displacement boundaries are calculated in sequence, and then the circumcenter numbering in step b) of step 2.1 is followed to complete the sequential numbering of all midpoint coordinates.

[0077] d) Generate numbers and coordinates of all node Voronoi structure vertices: The vertices of the non-boundary node Voronoi structure are the circumcenter of the Delaunay triangle sharing the node, and their coordinates and numbers are generated according to steps a) and b) in step 2.1; The vertices of the boundary node Voronoi structure include the circumcenter of the Delaunay triangle sharing the node and the midpoint of the line connecting the node adjacent to the node on the boundary, and their coordinates and numbers are generated according to steps a), b), and c) in step 2.1.

[0078] e) According to the angle between each node Voronoi vertex and the line connecting the node and the horizontal right direction, determine the counterclockwise order of the vertex numbers of all the nodes Voronoi structure.

[0079] 2.2) Use the Voronoi structure vertex coordinates slight offset method to slightly offset the coordinates of all Voronoi structure vertices:

[0080] x′ i =x i +α x ×β×γ (1)

[0081] y′ i =y i +α y ×β×γ (2)

[0082] In formula (1)-(2), (x i ,y i ) is the original coordinate of the Voronoi structure vertex i, (x′ i ,y′ i ) is the coordinate of the vertex after offset; α x and α y is the characteristic size of the structure. For rectangular plates, it is recommended to take the length of the plate structure in the x and y directions respectively; for circular plates, it is recommended to take the radius R of the plate; β is a random number between [-1,1]; γ is the offset coefficient, which is usually taken as 1.0×10 -7 ~1.0×10 -4 .

[0083] 2.3) Divide the Voronoi structure of all nodes into triangular subdomains, use the generalized strain smoothing stability method on each subdomain, and use C 1 The shape function is constructed by natural neighbor interpolation, and the smooth generalized strain-displacement relationship matrix corresponding to all triangular subdomains is generated by numerical integration.

[0084] a) In counterclockwise order, select two vertices of the Voronoi structure of each node in turn to form a triangular subdomain with the node (with Figure 1 Node x inr (r=1~NP) as an example, its Voronoi domain is S r , area A r , the triangular subdomain is S rs ,and RS is the total number of triangular subdomains); then according to the node x r And the spatial coordinates of the two vertices, calculate the triangle subdomain S rs Area A rs , each boundary Γ rs Normal vector n1, tangent vector n2;

[0085] b) Use the empty circle criterion in turn to determine the triangular subdomain S rs The three boundaries of Γ rs The natural neighboring nodes (number is np) of each integral point (each edge uses the 2-point integration rule, 3 edges have a total of 6 integral points) on the network, and calculate each natural neighboring node I (I = 1 ~ np) C 1 The first derivative of the natural neighbor shape function x rs corresponds to the triangular subdomain S rs Virtual nodes. Superscript w, θ x ,θ y They represent the deflection of the thin plate, the rotation in the x direction, and the rotation in the y direction respectively. The subscripts x and y represent the first-order derivatives in the x direction and y direction respectively.

[0086] c) Calculate the smooth generalized strain-displacement relationship matrix corresponding to each natural neighboring node of each integration point (I = 1 ~ np):

[0087]

[0088] d) Integrate the triangular subdomain S according to the “matching” method rs All integral points naturally adjacent to the corresponding nodes ( The total number of nodes after merging the 6 integral points with the same natural adjacent nodes) matrix, thus forming a triangular subdomain S rs The corresponding smooth generalized strain-displacement relationship matrix

[0089]

[0090] Step 3: Integrate the global stiffness matrix and the global equivalent load matrix

[0091] 3.1) Generate each triangular subdomain S in turn rs The corresponding stiffness matrix The overall stiffness matrix K of the final integrated plate structure is:

[0092]

[0093]

[0094] In formula (5), D is the elastic relationship matrix, and in formula (6), D0 = Eh0 3 / [12(1-v 2 )] is the bending stiffness of the plate, E is Young's modulus, h0 is the plate thickness, and v is Poisson's ratio.

[0095] 3.2) For the uniformly distributed force, according to the Voronoi domain S r Area A r The uniformly distributed force is equivalent to the lateral force acting on the node; for the concentrated force, it is directly applied to the node. Finally, according to the equivalent load array F of each node r The overall equivalent load array F of the integrated structure:

[0096]

[0097] Step 4: Solve the governing equations

[0098] The displacement boundary conditions are introduced by direct substitution method, diagonal element change to 1 method or diagonal element multiplication by large number method, and the overall stiffness matrix K and the overall load matrix F are modified. Then the Gaussian sequential elimination method, triangular decomposition method or Gauss-Jordan elimination method are used to solve the control equation Ka=F of the thin plate linear elastic problem, and the node displacement vector a of the thin plate structure is obtained:

[0099] a=[w1 θ x1 θ y1 w2θ x2 θ y2 … w NP θ x(NP) θ y(NP) ] T (8)

[0100] Where w1 is the deflection of node 1, θ x1 is the rotation of node 1x, θ y1 is the rotation of node 1 in the y direction; w2 is the deflection of node 2, θ x2 is the rotation of the node in the 2x direction, θ y2 is the rotation of node 2 in the y direction; ...; w NP is the deflection of node NP, θ x(NP) is the rotation of the node NP in the x direction, θ y(NP) is the rotation of the node NP in the y direction.

[0101] Step 5: Post-processing of calculation results

[0102] 5.1) According to the geometric relationship, the triangular subdomain S is determined by the node displacement vector and the smooth generalized strain-displacement relationship matrix rs Smooth generalized strain

[0103]

[0104] 5.2) According to each triangular subdomain S rs With Voronoi domain S r The area relationship of each node is determined by the smooth generalized strain

[0105]

[0106] 5.3) According to the constitutive relationship, calculate the smooth generalized stress of each node

[0107]

[0108] Example:

[0109] Clamped by uniform force q and concentrated force P Round thin plate ( Figure 2 Taking the linear elasticity problem in (a) and (b) as an example, the implementation method is as follows:

[0110] Step 1: Establish the geometric model of the circular thin plate and obtain the data of discrete nodes (see Figure 2 In (c), the total number of nodes is NP = 801), and Delaunay triangle data is generated from it (see Figure 2 (d), the total number is 1536);

[0111] Step 2: Extract the node numbers of the structure boundary ( Figure 2 The nodes on the arc in (c) are arranged in counterclockwise order) and the node number data of the load (when the uniformly distributed force q acts Figure 2 All nodes in (c) are Figure 2 The node at the center of the circle in (c).

[0112] Step 3: Prepare the radius, thickness, Young's modulus, Poisson's ratio, uniformly distributed force q and concentrated force P data of the thin plate structure, and set the offset coefficient γ.

[0113] Step 4: Determine the shared information of the Delaunay triangle of each node, the circumcenter information of the Delaunay triangle, and the midpoint information of two adjacent boundary nodes on the displacement boundary in turn, and generate the numbers and coordinates of the vertices of the Voronoi structure of all nodes based on this.

[0114] Step 5: According to the angle relationship, determine the counterclockwise order of the vertex numbers of the non-boundary nodes and boundary nodes of the Voronoi structure.

[0115] Step 6: Use the Voronoi structure vertex coordinate micro-offset method to slightly offset the coordinates of all Voronoi structure vertices.

[0116] Step 7: Calculate the area of ​​each triangular subdomain of the Voronoi structure of each node, the normal vector and the tangent vector of each triangular boundary; determine the natural adjacent nodes of the integral point on each triangular boundary according to the empty circle criterion, and calculate the first-order derivatives of the shape functions of these natural adjacent nodes.

[0117] Step 8: Integrate the smooth generalized strain-displacement relationship matrix corresponding to each triangular subdomain in a “matching” manner.

[0118] Step 9: Integrate the global stiffness matrix and the global equivalent load matrix.

[0119] Step 10: Introduce displacement boundary conditions to modify the overall stiffness matrix and the overall equivalent load matrix, solve the control equations of the thin plate structure, and obtain the node displacement vector.

[0120] Step 11: Based on the geometric relationship, area relationship and constitutive relationship, the smooth generalized strain and smooth generalized stress of each node are directly obtained.

[0121] Table 1 Comparison of errors and calculation time of different numerical methods

[0122]

[0123] Table 1 shows the method of the present invention (C 1 Nodal natural element method, C 1 -nodal-NEM) to calculate the displacement error between the numerical solution and the theoretical solution of the thin plate structure d (see formula (12)) and the computation time spent, and C 1 Natural Unit Method (C 1 -NEM) calculated displacement error e d The calculation time is compared.

[0124]

[0125] In formula (12): The superscripts n and t denote numerical and theoretical solutions, respectively.

[0126] It can be seen from Table 1 that: (1) When the offset coefficient γ takes 7 different values, the calculation accuracy of the present invention (through the displacement error e d (2) The calculation accuracy of the present invention is significantly higher than that of the C method using background grid integration. 1 - NEM, calculation time and C 1 -NEM equivalent.

[0127] Figure 3 to Figure 5 The invention and C 1 -The plate structure calculated by NEM is located at y=0.0,x∈[-R,R]( Figure 2 (c) The deflection w and rotation θ of the nodes between the horizontal solid lines x and θ y Comparison of numerical and theoretical solutions.

[0128] Figure 6-Figure 7 As the offset coefficient, γ = 1.0 × 10 -5 When the deflection, rotation and generalized strain cloud diagram of the clamped circular thin plate under the action of uniform force and concentrated force are drawn, the results directly solved by the present invention are used. Figure 6 In the figure, (a) is the cloud diagram of the deflection w; (b) is the rotation θ in the x direction x (c) is the y-direction rotation θ y (d) is the cloud diagram of the generalized strain in the x direction (e) is the cloud diagram of the generalized strain in the y direction (f) is the cloud diagram of the generalized strain in the xy direction Cloud map. Figure 7 In the figure, (a) is the cloud diagram of the deflection w; (b) is the rotation θ in the x direction x (c) is the y-direction rotation θ y (d) is the cloud diagram of the generalized strain in the x direction. (e) is the cloud diagram of the generalized strain in the y direction (f) is the cloud diagram of the generalized strain in the xy direction Cloud map.

[0129] Parts of the present invention that are not described in detail belong to the well-known technologies of those skilled in the art.

Claims

1. A method for calculating linear elastic problems of thin plate structures, characterized in that: include: S1. Establish a geometric model of the thin plate structure, obtain the numbers and spatial coordinates of all discrete nodes of the thin plate structure, generate the numbers of Delaunay triangles based on the discrete nodes and the numbers of the three nodes that make up each Delaunay triangle vertex, the numbers of the boundary nodes of the thin plate structure and their counterclockwise arrangement order, the numbers of the nodes on which the load acts, prepare the geometric dimensions, material parameters, and load size information of the thin plate structure, and set the offset coefficient; S2, prepare calculation data to form a smooth generalized strain-displacement relationship matrix corresponding to all triangular subdomains; S3, generating stiffness matrices corresponding to all triangular subdomains, and finally integrating the overall stiffness matrix of the plate structure; integrating the overall equivalent load matrix of the structure according to the equivalent load matrix of each node; S4, solving the governing equations of the linear elastic problem of the thin plate structure, and obtaining the node displacement vectors of the thin plate structure; S5. According to the geometric relationship, area relationship and constitutive relationship, the smooth generalized strain of the triangular subdomain, the smooth generalized strain of each node and the smooth generalized stress are determined in turn; S2, prepare calculation data to form a smooth generalized strain-displacement relationship matrix corresponding to all triangular subdomains, including: S21, generating the numbers and spatial coordinates of all the nodes of the Voronoi structure vertex, and determining the counterclockwise arrangement order of the Voronoi structure vertex numbers according to the angle size; S22, using a Voronoi structure vertex coordinate micro-shift method to slightly shift the coordinates of all Voronoi structure vertices; S23, divide the Voronoi structure of all nodes into triangular subdomains, use the generalized strain smoothing stability method on each subdomain, and use C 1 The shape function is constructed by natural neighbor interpolation, and the smooth generalized strain-displacement relationship matrix corresponding to all triangular subdomains is generated by numerical integration.

2. A method for calculating linear elastic problems of thin plate structures according to claim 1, characterized in that: The step S21 generates the numbers and spatial coordinates of all the nodes of the Voronoi structure vertex, and determines the counterclockwise arrangement order of the Voronoi structure vertex numbers according to the angle, including: S21a, matching the numbers of all nodes with the node numbers of all Delaunay triangle vertices in sequence, determining and counting the number of times each node is shared by Delaunay triangles and the numbers of the Delaunay triangles that share the node; S21b, according to the spatial coordinates of the three nodes constituting the Delaunay triangle, calculate the spatial coordinates of the circumcenters of all Delaunay triangles, and sequentially number the circumcenters in turn; if the circumcenter coordinates of two triangles are the same, their numbers are the circumcenter numbers of the first numbering; S21c, according to the spatial coordinates of the two adjacent displacement boundary nodes, the midpoint coordinates of the two adjacent boundary nodes on all displacement boundaries are calculated in sequence, and then the circumcenter numbering in S21b is followed to complete the sequential numbering of all midpoint coordinates; S21d, generate the numbers and coordinates of the vertices of the Voronoi structure of all nodes: the vertices of the Voronoi structure of non-boundary nodes are the circumcenters of the Delaunay triangles that share the node, and their coordinates and numbers are generated according to S21a and S21b; the vertices of the Voronoi structure of boundary nodes include the circumcenters of the Delaunay triangles that share the node and the midpoints of the lines connecting the nodes adjacent to the node on the boundary, and their coordinates and numbers are generated according to S21a to S21c; S21e, determining the counterclockwise order of the vertices numbering of the Voronoi structures of all nodes according to the angle between each node Voronoi vertex and the line connecting the node and the horizontal rightward direction.

3. A method for calculating linear elastic problems of thin plate structures according to claim 2, characterized in that: The step S22, using a Voronoi structure vertex coordinate micro-shift method to micro-shift the coordinates of all Voronoi structure vertices, includes: x′ i =x i +a x ×β×γ, and' i =and i +α y ×β×γ, Among them, (x i ,y i ) is the original coordinate of the Voronoi structure vertex i, (x′ i ,y′ i ) is the offset coordinate of vertex i; α x and α y is the characteristic size of the structure; β is a random number between [-1,1]; γ is the offset coefficient.

4. A method for calculating linear elastic problems of thin plate structures according to claim 3, characterized in that: S23, divide the Voronoi structure of all nodes into triangular subdomains, use the generalized strain smoothing stability method on each subdomain, and use C 1 The shape function is constructed by natural neighbor interpolation, and the smooth generalized strain-displacement relationship matrix corresponding to all triangular subdomains is generated by numerical integration, including: S23a, in counterclockwise order, select two vertices of the Voronoi structure of each node in turn to form a triangular subdomain with the node: node x r The Voronoi domain is S r , area A r , the triangular subdomain is S rs ,and RS is the total number of triangular subdomains, r = 1 ~ NP, NP is the total number of nodes; then according to the node x r And the spatial coordinates of the two vertices, calculate the triangle subdomain S rs Area rs , each boundary Γ rs The normal vector n1 and tangent vector n2; S23b, using the empty circle criterion in turn, determine the triangle subdomain S rs The three boundaries of Γ rs The number of natural neighboring nodes of each integral point is recorded as np, and the C of each natural neighboring node I is calculated 1 The first derivative of the natural neighbor shape function x rs corresponds to the triangular subdomain S rs Virtual nodes, I = 1 ~ np; where superscript w, θ x ,θ y They represent the deflection of the plate, the rotation in the x direction, and the rotation in the y direction respectively; S23c, calculate the smooth generalized strain-displacement relationship matrix corresponding to each integral point's naturally adjacent node I=1~np; S23d, integrated triangular subdomain S rs All integral points naturally adjacent to the corresponding nodes Matrix, forming a triangular subfield S rs The corresponding smooth generalized strain-displacement relationship matrix in, It is the total number of nodes after merging all naturally adjacent nodes with the same integral points.

5. A method for calculating linear elastic problems of thin plate structures according to claim 4, characterized in that: The smooth generalized strain-displacement relationship matrix corresponding to each natural neighboring node of each integration point 6. A method for calculating linear elastic problems of thin plate structures according to claim 4, characterized in that: In S3, the stiffness matrices corresponding to all triangular subdomains are generated, and finally the overall stiffness matrix of the integrated plate structure is obtained; The overall equivalent load array of the structure is integrated according to the equivalent load array of each node, including: S31, generate each triangular subdomain S in sequence rs The corresponding stiffness matrix The overall stiffness matrix K of the final integrated plate structure; Where D is the elastic relationship matrix; S32. For the uniformly distributed force, according to the Voronoi domain S r Area A r The uniformly distributed force is equivalent to the lateral force acting on the node; for the concentrated force, it is applied to the acting node; according to the equivalent load array F of each node r The overall equivalent load array F of the integrated structure:

7. A method for calculating linear elastic problems of thin plate structures according to claim 6, characterized in that: The overall stiffness matrix K of the final integrated plate structure is: Where D0 = Eh0 3 / [12(1-v 2 )] is the bending stiffness of the plate, E is Young's modulus, h0 is the plate thickness, and v is Poisson's ratio.

8. A method for calculating linear elastic problems of thin plate structures according to claim 6, characterized in that: In S4, the control equations for solving the linear elastic problem of the thin plate structure include: The displacement boundary conditions are introduced by direct substitution method, diagonal element change to 1 method or diagonal element multiplication by large number method, and the overall stiffness matrix K and the overall load matrix F are modified. The Gaussian sequential elimination method, triangular decomposition method or Gauss-Jordan elimination method are used to solve the control equation Ka=F of the linear elastic problem of the thin plate structure, and the node displacement vector a of the thin plate structure is obtained: a=[w1 θ x1 i y1 w2 θ x2 i y2 …w NP i x(NP) i y(NP) ] T Where w1 is the deflection of node 1, θ x1 is the rotation of node 1x, θ y1 is the rotation of node 1 in the y direction; w2 is the deflection of node 2, θ x2 is the rotation of the node in the 2x direction, θ y2 is the rotation of node 2 in the y direction; w NP is the deflection of node NP, θ x(NP) is the rotation of the node NP in the x direction, θ y(NP) is the rotation of the node NP in the y direction.

9. A method for calculating linear elastic problems of thin plate structures according to claim 8, characterized in that: The S5, according to the geometric relationship, area relationship and constitutive relationship, determines the smooth generalized strain of the triangular subdomain, the smooth generalized strain and the smooth generalized stress of each node in turn, including: According to the geometric relationship, the triangular subdomain S is determined by the node displacement vector and the smooth generalized strain-displacement relationship matrix. rs Smooth generalized strain According to each triangular subdomain S rs With Voronoi domain S r The area relationship of each node is determined by the smooth generalized strain According to the constitutive relationship, the smooth generalized stress of each node is calculated