Deformed mesh quality optimization method based on adjoint gradient
By constructing the grid deformation quality optimization objective function and deriving the accompanying gradient, the problem of unit quality deterioration during grid deformation is solved, the improvement of grid quality and the stability of the optimization process is achieved, and the grid deformation in the existing technology is solved The problem of distortion of calculation results caused by low quality of the rear unit.
Patent Information
- Application Number
- CN202411298130.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-18
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2044-09-18
AI Technical Summary
The prior art is prone to problems such as excessive inclination of local grid units and negative volume during grid deformation, resulting in a decrease in grid quality, increasing calculation errors and divergence in the optimization process, affecting the accuracy and reliability of the design.
The deformation grid quality optimization method based on the accompanying gradient is adopted, and the accompanying gradient is derived by constructing the mesh deformation quality optimization objective function, and the quasi-Newtonian algorithm is used to solve the optimal solution of the stiffness of the grid element, interpolation is used to optimize the stiffness value into the viscous grid. , regenerate the grid node distribution and improve the grid quality.
It effectively improves the grid quality, reduces calculation errors, improves the stability of the optimization process, and solves the problem of distortion of calculation results caused by the low unit quality after grid deformation.
Smart Images

Figure CN118821329B_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of mesh deformation optimization design, and in particular to a deformation mesh quality optimization method based on adjoint gradient. Background Art
[0002] In the aerospace field based on computational fluid dynamics (CFD), mesh deformation is an essential step for the optimization design of aerodynamic shape of aircraft, multi-body system motion, aeroelastic problems, and control surface motion problems. The ability of mesh deformation directly affects the accuracy and stability of numerical simulation. At present, commonly used mesh deformation methods include algebraic method, spring analogy method, radial basis function method, explicit interpolation method, and linear elastic method.
[0003] Among them, the linear elastic method is one of the most widely used methods. Based on the assumption of linear elastic body mechanics, the displacement field is calculated by solving a large-scale linear equation set to achieve mesh deformation. However, there are some problems in the prior art. For example, in the face of large deformation and large displacement, the inclination rate of the local mesh unit may be too large, or even negative volume may appear, which cannot meet the simulation requirements. The problems existing in the prior art mainly include the decline of mesh quality, the increase of calculation discrete error, and the convergence to false optimum. Especially in the process of aerodynamic shape optimization design based on CFD, due to the negative volume and other problems caused by mesh deformation, the optimization process is easy to diverge, affecting the accuracy and reliability of the design. Therefore, a new technology that can improve mesh quality, reduce calculation errors, and improve the stability of the optimization process is needed to solve these problems. Summary of the invention
[0004] The present invention aims to solve at least one of the technical problems existing in the prior art.
[0005] To this end, the present invention provides a deformation grid quality optimization method based on adjoint gradient.
[0006] The present invention proposes a deformation grid quality optimization method based on adjoint gradient, which is applied to the linear elastic method in grid deformation and includes the following steps:
[0007] Meshing the target object to form a viscous mesh, deforming the viscous mesh to obtain a first deformed mesh, and determining whether the quality of the first deformed mesh meets subsequent use requirements;
[0008] When the first deformed mesh does not meet the subsequent use requirements, the target object is re-meshed by using a triangulation method to form a non-viscous mesh, and the non-viscous mesh is obtained by using the same mesh deformation method to obtain a second deformed mesh;
[0009] Constructing a mesh deformation quality optimization objective function based on the non-viscous mesh, wherein the optimization objective function is constructed according to the product of the mesh quality ratios of all mesh units in the non-viscous mesh before and after deformation; the design variables of the optimization objective function include the elastic modulus coefficient of the mesh unit;
[0010] Derivation of an adjoint gradient of the optimization objective function relative to the design variable; the adjoint gradient is related to a grid node displacement vector;
[0011] Obtaining an optimal solution for the stiffness of the grid unit according to the optimization objective function and the accompanying gradient, that is, optimizing the stiffness value;
[0012] The optimized stiffness value is interpolated into the viscous grid, the stiffness equation of the viscous grid system is solved, the grid node distribution based on the optimized stiffness is obtained, and the first deformed grid is regenerated.
[0013] The deformed grid quality optimization method based on the accompanying gradient according to the above technical solution of the present invention may also have the following additional technical features:
[0014] In the above technical solution, in the non-viscous grid, the method for calculating the quality of the grid cells includes:
[0015]
[0016] in, represents the quality of the mesh element, represents the vector area of the grid cell, , , Represent the three side lengths of the triangular mesh unit.
[0017] In the above technical solution, in the linear elastic method, the original elastic modulus of the grid unit is multiplied by a coefficient k as the elastic modulus of the spring system, and k is defined as the design variable in the optimization objective function.
[0018] In the above technical solution, the optimization objective function is:
[0019]
[0020] in, represents the optimization objective function, represents the mass of the i-th grid element after deformation with design variable k, represents the mass of the i-th grid cell before deformation.
[0021] In the above technical solution, the value space of the design variable k is 0.5~1.5.
[0022] In the above technical solution, the accompanying gradient of the optimization objective function relative to the design variable is:
[0023]
[0024] in, represents the adjoint gradient of the optimization objective function in the i-th grid cell with respect to the design variable, That is, the design variable corresponding to the i-th grid unit; Indicates the node coordinates after mesh deformation; represents the total stiffness matrix, which is a linear assembly of all mesh element stiffness matrices. The mesh element stiffness is a linear function of the elastic modulus; represents the original elastic modulus of the i-th grid element; Represents the mesh node displacement vector.
[0025] In the above technical solution, the optimal solution of the grid unit stiffness is obtained according to the optimization objective function and the accompanying gradient, including:
[0026] The optimization objective function and the accompanying gradient are solved by using a quasi-Newton algorithm to obtain the optimal value of the design variable of each mesh unit, and the optimal solution of the stiffness of each mesh unit is determined according to the selected design variables.
[0027] In the above technical solution, interpolating the optimized stiffness value into the viscous grid includes:
[0028] In the interpolation data of the viscous grid and the non-viscous grid, the stiffness value is considered to be at the center of the grid unit, that is, the interpolation is performed based on the grid center;
[0029] The optimized stiffness values are interpolated to the viscous grid using a gradient-based grid center interpolation method.
[0030] In the above technical solution, the gradient-based grid center interpolation method is used to interpolate the optimized stiffness value to the viscous grid, including:
[0031]
[0032] in, represents the stiffness corresponding to the center of the interpolation target grid, that is, the stiffness corresponding to the center of the viscous grid; represents the stiffness corresponding to the center of the interpolation source grid, that is, the stiffness corresponding to the center of the non-viscous grid, that is, the optimized stiffness value; represents the gradient of the stiffness corresponding to the cell center in the non-viscous grid; Represents the center vector of the sticky and non-sticky grids.
[0033] In the above technical solution, the Green-Gauss method is used to solve the gradient of the stiffness corresponding to the grid center in the non-viscous grid. ,in:
[0034]
[0035] in, represents the gradient of the stiffness at the center of the grid; when the grid is deformed in three-dimensional space, V represents the volume of the grid unit, and when the grid is deformed in two-dimensional space, V represents the area of the grid unit); Represents the average stiffness on the unit surface; when the mesh is deformed in three-dimensional space, Represents the area of the grid unit. When the grid is deformed in two-dimensional space, Represents the side length of the grid cell.
[0036] In summary, due to the adoption of the above technical features, the beneficial effects of the present invention are:
[0037] The present invention provides a method for optimizing the quality of mesh units in a process of calculating mesh deformation. The method refines mesh quality optimization into an optimization problem, takes mesh quality index as an objective function, derives the accompanying gradient of the objective function, and solves the optimization problem through a gradient-based optimization algorithm, thereby improving mesh quality and solving problems such as distortion of calculation results or termination of the calculation process caused by low unit quality after mesh deformation in engineering calculations. The method aims at the problem of unit quality deterioration caused by applying a linear elastic method to the mesh deformation process. The constructed objective function can effectively reflect the quality of the mesh, and the accompanying gradient can accurately express the sensitivity of the objective function to the design variable. Both are easy to implement in programming, and are a simple and effective method for improving the quality of deformed meshes.
[0038] Additional aspects and advantages of the present invention will become apparent from the following description or may be learned by practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS
[0039] The above and / or additional aspects and advantages of the present invention will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which:
[0040] Figure 1 is a flow chart of a deformed mesh quality optimization method based on adjoint gradient according to an embodiment of the present invention;
[0041] Figure 2 It is a schematic diagram of a viscous mesh near the entire airfoil of an aircraft in a deformed mesh quality optimization method based on adjoint gradient according to an embodiment of the present invention;
[0042] Figure 3It is a schematic diagram of a viscous mesh near the trailing edge of an aircraft airfoil in a deformed mesh quality optimization method based on adjoint gradient according to an embodiment of the present invention;
[0043] Figure 4 It is a schematic diagram of a non-viscous mesh near the entire airfoil of an aircraft in a deformed mesh quality optimization method based on adjoint gradient according to an embodiment of the present invention;
[0044] Figure 5 It is a schematic diagram of a non-viscous mesh near the trailing edge of an aircraft airfoil in a deformable mesh quality optimization method based on an adjoint gradient according to an embodiment of the present invention;
[0045] Figure 6 is a schematic diagram of rightward translation of inner boundary nodes in a deformed mesh quality optimization method based on adjoint gradient according to an embodiment of the present invention;
[0046] Figure 7 is Figure 6 A schematic diagram comparing the calculation results of the grid quality relative to the grid unit stiffness of the gradient solution method and the finite difference method of the present invention;
[0047] Figure 8 is a schematic diagram of optimal stiffness distribution based on a non-viscous grid in a deformable grid quality optimization method based on an adjoint gradient according to an embodiment of the present invention;
[0048] Fig. 9 It is a schematic diagram of interpolating the optimized stiffness value to the viscous mesh in the deformed mesh quality optimization method based on the adjoint gradient according to an embodiment of the present invention;
[0049] Fig.10 Schematic diagram of stiffness distribution of an interpolation source grid in a deformed grid quality optimization method based on adjoint gradient according to an embodiment of the present invention;
[0050] Fig.11 Schematic diagram of stiffness distribution of an interpolated target mesh in a deformed mesh quality optimization method based on adjoint gradient according to an embodiment of the present invention;
[0051] Fig.12 It is a schematic diagram of a viscous mesh near the entire airfoil of a certain aircraft after optimization in a deformed mesh quality optimization method based on accompanying gradient according to an embodiment of the present invention;
[0052] Fig.13 It is a schematic diagram of a viscous mesh near the trailing edge of an aircraft airfoil after optimization in a deformable mesh quality optimization method based on accompanying gradients according to an embodiment of the present invention. DETAILED DESCRIPTION
[0053] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.
[0054] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited to the specific embodiments disclosed below.
[0055] Refer to the following Figures 1 to 13 The following describes a deformed mesh quality optimization method based on adjoint gradients provided according to some embodiments of the present invention.
[0056] Some embodiments of the present application provide a method for optimizing quality of a deformed mesh based on an adjoint gradient.
[0057] like Figure 1 As shown, the first embodiment of the present invention proposes a deformation grid quality optimization method based on accompanying gradient, which is applied to the linear elastic method in grid deformation and includes the following steps S1-S6.
[0058] S1. Grid the target object to form a viscous grid. After the viscous grid is deformed, a first deformed grid is obtained, and it is determined whether the quality of the first deformed grid meets the subsequent use requirements. Specifically, in step S1, the traditional linear elastic method is directly used to deform the viscous grid in a set manner, and the deformation method can be rotation, translation, etc. The method of determining whether the quality of the first deformed grid meets the subsequent use requirements is well known to those skilled in the art, and is not an improved content of the present disclosure, so it will not be elaborated here. It can be understood that when the first deformed grid can meet the subsequent use requirements, subsequent analysis is directly carried out on the basis of the existing first deformed grid to form an output result. In some embodiments, the first deformed grid that meets the subsequent use requirements is defined as a high-quality grid, and the first deformed grid that does not meet the subsequent use requirements is defined as a low-quality grid.
[0059] In a specific embodiment, the rotation deformation of the airfoil of a certain type of aircraft is taken as an example, that is, the target object is the airfoil of the aircraft, and it is assumed that the mesh deformation is a necessary step in an engineering calculation process. The rotation mode of the airfoil is set to rotate 80° clockwise around the leading edge point, such as Figure 2 and Figure 3 As shown in Figure 1, the mesh quality near the trailing edge of the deformed airfoil is extremely poor.
[0060] S2. When the first deformed mesh does not meet the subsequent use requirements, a triangulation (Delaunay mesh) method is used to re-mesh the target object to form a non-viscous mesh, and the non-viscous mesh uses the same mesh deformation method to obtain a second deformed mesh.
[0061] Among them, Delaunay triangulation is a geometric structure introduced by Boris Delaunay in 1934 and is widely used in computational geometry, data fitting, and unstructured grid generation. Its core feature is the "hollow circle" property, that is, for any edge, a circle can be found that contains the two endpoints of the edge but does not contain any other points.
[0062] It is understandable that, relatively speaking, the non-viscous mesh is a coarser mesh and the viscous mesh is a finer mesh. In step S2, the non-viscous mesh is deformed in the same deformation manner as step S1, and the mesh quality will be lower than that of the viscous mesh.
[0063] In a specific embodiment, based on the airfoil provided in the above embodiment, a triangulation method is used to regenerate the non-viscous mesh of the airfoil before deformation, such as Figure 4 and Figure 5 As shown in the figure, after the same rotational deformation, the mesh quality of the non-viscous mesh is lower than that of the viscous mesh, the mesh of the lower surface of the airfoil trailing edge becomes a negative volume, the number of mesh cells with a mesh quality less than 0 is 1, and the number of mesh cells with a mesh quality less than 0.05 is 4.
[0064] S3. Constructing a mesh deformation quality optimization objective function based on the non-viscous mesh, wherein the optimization objective function is constructed according to the product of the mesh quality ratios of all mesh units in the non-viscous mesh before and after deformation; and the design variables of the optimization objective function include the elastic modulus coefficients of the mesh units.
[0065] The mesh deformation quality optimization objective function is a function of the quality of the mesh after deformation. In some embodiments, in the non-viscous mesh, the method for calculating the quality of the mesh unit includes:
[0066]
[0067] in, represents the quality of the mesh element, represents the vector area of the grid cell, , , Represent the three side lengths of the triangular mesh unit.
[0068] Generally speaking, the number of cells whose mesh quality changes significantly after deformation is small. If the objective function of mesh deformation quality optimization is defined as the sum of the qualities of all mesh cells, the objective function numerically includes the contribution of a large proportion of high-quality meshes (far away from the deformed surface) and a small proportion of low-quality meshes (close to the deformed surface). The impact of design variables on the objective function is small in magnitude, which is not conducive to optimizing the objective function. To avoid the above problems, for mesh cells whose quality deteriorates after deformation, the ratio of the mesh quality before and after deformation is is much smaller than 1, while for mesh elements that retain their quality after deformation, the ratio is close to 1. The product of the mesh quality ratios of all elements is It can react more sensitively to less mesh quality degradation. However, in some cases, the quality of the mesh elements is generally degraded after deformation. The value of may also be very small, which is not suitable as an objective function. Therefore, in the present disclosure, the optimization objective function is:
[0069]
[0070] in, represents the optimization objective function, represents the mass of the i-th grid element after deformation with design variable k, represents the mass of the i-th grid cell before deformation.
[0071] Specifically, in the linear elastic method, the original elastic modulus of the grid unit is multiplied by a coefficient k as the elastic modulus of the spring system, and k is defined as the design variable in the optimization objective function. In a specific embodiment, in order to make the design variables of the optimization problem have the same scale, the value space of the design variable k is 0.5~1.5.
[0072] S4, deriving the adjoint gradient of the optimization objective function relative to the design variable; the adjoint gradient is related to the grid node displacement vector;
[0073] According to the chain derivation principle, the adjoint gradient of the objective function with respect to the design variable It can be expressed as:
[0074]
[0075] in, represents the adjoint gradient of the optimization objective function in the i-th grid cell with respect to the design variable, That is, the design variable corresponding to the i-th grid unit; Represents the mesh node displacement vector.
[0076] When the mesh is deformed, the spatial mesh node displacement vector is affected by the unit elastic modulus, the initial position of the node, and the displacement of the boundary node. The deformation mesh quality optimization problem designed in the present disclosure is aimed at a specific deformation process. The initial position of the mesh node and the displacement of the boundary node are known. Therefore, the mesh node displacement is only a function of the unit elastic modulus, that is, , where E represents the original elastic modulus of the mesh element.
[0077] The initial node coordinates of the grid are expressed as , the node coordinates after mesh deformation are expressed as x, then:
[0078]
[0079] The objective function F is defined as an explicit function of the grid node coordinates. It can be obtained by simple derivation. The dimension of is the same as q. The displacement of the boundary grid points of the computational domain contained in q is zero, so The corresponding value of is treated as zero.
[0080] The finite element equation of the linear elastic method is:
[0081]
[0082] in, represents the total stiffness matrix; f represents the elastic force on the node in the spring system.
[0083] In the linear elastic method finite element equation, both ends are The derivative is:
[0084]
[0085] in, represents the original elastic modulus of the i-th mesh element.
[0086] Since the total stiffness matrix K is a linear assembly of all mesh element stiffness matrices, and the mesh element stiffness is a linear function of the elastic modulus, K has no significant effect on the individual elements. The partial derivative matrix of may be non-zero only at the corresponding position of the element (the number of non-zero elements is equal to the dimension of the mesh element stiffness matrix), and the rest of the values are zero. The node displacement vector q is solved in each optimization iteration, so the first term on the left side of the equation is solved.
[0087] After the same analysis, the value of q at the boundary point is not affected by the elastic modulus, These values of are also zero. Therefore, the linear elastic method finite element equation can be transformed into:
[0088]
[0089] The solvable matrix equation is obtained , whose computational complexity is similar to solving the finite element equation The same, but the number of equations to be solved for one mesh quality optimization iteration is the same as the number of mesh elements (that is, the number of design variables), and the total amount of computation for iteratively solving the equations is large.
[0090] Therefore, considering The values on the boundaries are zero, but the matrix The elements in are non-zero only at the corresponding index of the node of the current grid unit, so K and The boundary corresponding element, the new stiffness matrix at this time is a non-singular matrix, and then Decomposition or inverse solution , and fill the solution vector with zeros to obtain .
[0091] Finally, the adjoint gradient of the objective function with respect to the design variable can be expressed as:
[0092]
[0093] That is, the adjoint gradient of the objective function is obtained.
[0094] In a specific embodiment, Figure 6 A schematic diagram of a simple grid system with boundary nodes shifted to the right is given. The numbers 1-52 in the figure are used to represent the grid unit numbers. Figure 7 Zhongzai Figure 6 Based on the results of the calculation of the grid quality of the gradient solution method of the present invention and the finite difference method relative to the design variable gradient, it can be seen that the results of the gradient solution method of the present invention are in good agreement with the results of the finite difference method, which proves the effectiveness of the present method; and the design parameters of the present invention are adopted. The adjoint gradient of the objective function obtained is related to the elastic modulus Compared with the adjoint gradient of the objective function obtained as the design parameter, the mesh quality of the present invention is obviously better.
[0095] S5. Obtaining the optimal solution of the grid unit stiffness according to the optimization objective function and the accompanying gradient, that is, optimizing the stiffness value.
[0096] In some embodiments, the optimization objective function and the accompanying gradient are solved using a quasi-Newton algorithm (BFGS) to obtain the best value of the design variable of each mesh unit, and the optimal solution of the stiffness of each mesh unit is determined according to the selected design variables.
[0097] Specifically, see again Figure 1, the mesh quality of the non-viscous mesh deformation is evaluated by optimizing the objective function, sensitivity analysis is performed using the adjoint gradient, and iterative calculations are performed using the quasi-Newton algorithm until suitable design variables are found to make the objective function converge, that is, the current design variables can make the deformed mesh quality optimal. The optimal solution of the mesh unit stiffness can be calculated by the selected design variables. The optimal stiffness distribution based on the non-viscous mesh is as follows: Figure 8 shown.
[0098] S6. Interpolate the optimized stiffness value into the viscous grid, solve the stiffness equation of the viscous grid system, obtain the grid node distribution based on the optimized stiffness, and regenerate the first deformed grid.
[0099] In some embodiments, interpolating the optimized stiffness value into the viscous grid comprises:
[0100] In the interpolation data of the viscous grid and the non-viscous grid, the stiffness value is considered to be at the center of the grid unit, that is, the interpolation is based on the grid center; the optimized stiffness value is interpolated to the viscous grid using the gradient-based grid center interpolation method. Fig. 9 As shown, we have:
[0101]
[0102] in, represents the stiffness corresponding to the center of the interpolation target grid, that is, the stiffness corresponding to the center of the viscous grid; represents the stiffness corresponding to the center of the interpolation source grid, that is, the stiffness corresponding to the center of the non-viscous grid, that is, the optimized stiffness value; represents the gradient of the stiffness corresponding to the cell center in the non-viscous grid; Represents the center vector of the sticky and non-sticky grids.
[0103] Specifically, the Green-Gauss method is used to solve the gradient of the stiffness corresponding to the grid center in the non-viscous grid: ,in:
[0104]
[0105] in, represents the gradient of the stiffness at the center of the grid; when the grid is deformed in three-dimensional space, V represents the volume of the grid unit, and when the grid is deformed in two-dimensional space, V represents the area of the grid unit); Represents the average stiffness on the unit surface; when the mesh is deformed in three-dimensional space, Represents the area of the grid unit. When the grid is deformed in two-dimensional space, Represents the side length of the grid cell.
[0106]
[0107] in, and They represent the stiffness at the center of two adjacent grid cells respectively.
[0108] Fig.10 shows the stiffness distribution of the interpolated source mesh, Fig.11 The stiffness distribution of the interpolated target mesh is shown.
[0109] On the viscous grid with interpolated stiffness, the stiffness equation of the viscous grid system is solved to obtain the grid node distribution based on the optimized stiffness, and the first deformed grid is regenerated. The regenerated first deformed grid is as follows: Fig.12 and 13 As shown. The optimized high-quality mesh is input into the subsequent analysis process of engineering calculation. It should be noted that the stiffness equation of the mesh system and its solution method are well known to those skilled in the art and will not be described in detail here.
[0110] In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.
[0111] Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.
Claims
1. A method for optimizing the quality of deformed meshes based on adjoint gradients, applied to the linear elastic method in mesh deformation, characterized in that: The following steps are involved: Meshing the target object to form a viscous mesh, deforming the viscous mesh to obtain a first deformed mesh, and determining whether the quality of the first deformed mesh meets subsequent use requirements; When the first deformed mesh does not meet the subsequent use requirements, the target object is re-meshed by using a triangulation method to form a non-viscous mesh, and the non-viscous mesh is obtained by using the same mesh deformation method to obtain a second deformed mesh; Constructing a mesh deformation quality optimization objective function based on the non-viscous mesh, wherein the optimization objective function is constructed according to the product of the mesh quality ratios of all mesh units in the non-viscous mesh before and after deformation; the design variables of the optimization objective function include the elastic modulus coefficient of the mesh unit; Derivation of an adjoint gradient of the optimization objective function relative to the design variable; the adjoint gradient is related to a grid node displacement vector; Obtaining an optimal solution for the stiffness of the grid unit according to the optimization objective function and the accompanying gradient, that is, optimizing the stiffness value; interpolating the optimized stiffness value into the viscous grid, solving the stiffness equation of the viscous grid system, obtaining the grid node distribution based on the optimized stiffness, and regenerating the first deformed grid; The adjoint gradient of the optimization objective function relative to the design variable is: in, represents the adjoint gradient of the optimization objective function in the i-th grid cell with respect to the design variable, represents the optimization objective function, That is, the design variable corresponding to the i-th grid unit; Indicates the node coordinates after mesh deformation; represents the total stiffness matrix, which is a linear assembly of all mesh element stiffness matrices. The mesh element stiffness is a linear function of the elastic modulus; represents the original elastic modulus of the i-th grid element; Represents the mesh node displacement vector.
2. The method for optimizing deformed mesh quality based on adjoint gradient according to claim 1, characterized in that: In the non-viscous grid, the method for calculating the quality of the grid cells includes: in, represents the quality of the mesh element, represents the vector area of the grid cell, , , Represent the three side lengths of the triangular mesh unit.
3. The method for optimizing deformed mesh quality based on adjoint gradient according to claim 2, characterized in that: In the linear elastic method, the original elastic modulus of the grid unit is multiplied by a coefficient k as the elastic modulus of the spring system, and k is defined as the design variable in the optimization objective function.
4. The method for optimizing deformed mesh quality based on adjoint gradient according to claim 3, characterized in that: The optimization objective function is: in, represents the mass of the i-th grid element after deformation with design variable k, represents the mass of the i-th grid cell before deformation.
5. The method for optimizing deformed mesh quality based on adjoint gradient according to claim 4, characterized in that: The value space of the design variable k is 0.5~1.
5.
6. The method for optimizing deformed mesh quality based on adjoint gradient according to claim 4, characterized in that: The step of obtaining an optimal solution for the stiffness of a mesh unit according to the optimization objective function and the accompanying gradient includes: The optimization objective function and the accompanying gradient are solved by using a quasi-Newton algorithm to obtain the optimal value of the design variable of each mesh unit, and the optimal solution of the stiffness of each mesh unit is determined according to the selected design variables.
7. The method for optimizing deformed mesh quality based on adjoint gradient according to claim 1, characterized in that: The interpolating the optimized stiffness value into the viscous grid comprises: In the interpolation data of the viscous grid and the non-viscous grid, the stiffness value is considered to be at the center of the grid unit, that is, the interpolation is performed based on the grid center; The optimized stiffness values are interpolated to the viscous grid using a gradient-based grid center interpolation method.
8. The method for optimizing deformed mesh quality based on adjoint gradient according to claim 7, characterized in that: The gradient-based grid center interpolation method is used to interpolate the optimized stiffness value to the viscous grid, including: in, represents the stiffness corresponding to the center of the interpolation target grid, that is, the stiffness corresponding to the center of the viscous grid; represents the stiffness corresponding to the center of the interpolation source grid, that is, the stiffness corresponding to the center of the non-viscous grid, that is, the optimized stiffness value; represents the gradient of the stiffness corresponding to the cell center in the non-viscous grid; Represents the center vector of the sticky and non-sticky grids.
9. The method for optimizing deformed mesh quality based on adjoint gradient according to claim 8, characterized in that: The Green-Gauss method is used to solve the gradient of the stiffness corresponding to the center of the non-viscous grid. ,in: in, represents the gradient of the stiffness at the center of the grid; when the grid is deformed in three-dimensional space, V represents the volume of the grid unit, and when the grid is deformed in two-dimensional space, V represents the area of the grid unit); Represents the average stiffness on the unit surface; when the mesh is deformed in three-dimensional space, Represents the area of the grid unit. When the grid is deformed in two-dimensional space, Represents the side length of the grid cell.
