A lofting curved surface shell structure optimization design method based on hybrid sensitivity

By introducing hybrid sensitivity analysis and surface lofting techniques into the shell structure optimization design, and using the cross-sectional curve to control the vertex offset as a design variable, the problems of difficulty in preserving design intent and complexity of sensitivity analysis in shell structure optimization are solved, achieving efficient and smooth shell structure optimization.

CN122452236APending Publication Date: 2026-07-24BEIHANG UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202610604103.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-05-06
Publication Date
2026-07-24

AI Technical Summary

Technical Problem

Existing technologies struggle to effectively preserve design intent in shell structure optimization design, and the complexity of sensitivity analysis leads to poor optimization results.

Method used

A lofted surface shell structure optimization design method based on hybrid sensitivity is adopted. The coordinate offset of the control vertex of the cross-section curve is used as the design variable. Combined with the hybrid sensitivity analysis strategy, the initial lofted surface is generated by surface lofting, and then refined and structural response analysis is performed. Iterative optimization is carried out until the convergence condition is met.

Benefits of technology

It achieves high-quality and efficient shell structure optimization design, generates smooth and manufacturable curved surfaces, reduces the design space dimension, improves the feasibility and efficiency of sensitivity calculation, and reduces optimization calculation cost.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122452236A_ABST
    Figure CN122452236A_ABST
Patent Text Reader

Abstract

The application discloses a lofted curved surface shell structure optimization design method based on mixed sensitivity. The method takes the offset amount of the initial section curve control vertex coordinates as the design variable, sorts the section curve, unifies the parameter direction, inserts nodes and the like, and generates an initial lofted curved surface through lofting. The initial lofted curved surface is refined, and a refined lofted curved surface model for structure analysis is constructed. Structural response analysis is carried out based on the equal geometry Reissner-Mindlin shell theory. The derivative of the objective function to the design variable is decomposed into the accompanying analytical sensitivity term and the finite difference sensitivity term. Under the geometric constraint, the minimum flexibility is taken as the optimization target for iteration, and the optimized section curve, the lofted curved surface and the corresponding structural response are output. The application can retain the design intention driven by the section characteristic curve, reduce the design space dimension, improve the realizability of the sensitivity analysis and the optimization efficiency, and realize the high-quality and efficient optimization design of the shell structure.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to the fields of computer-aided design and computer-aided engineering, and specifically relates to a method for optimizing the design of lofted curved surface shell structures based on hybrid sensitivity. Background Technology

[0002] Shell structures are widely used in engineering fields such as aircraft, ship hulls, automobile bodies, and building roofs due to their lightweight, high load-bearing capacity, and aesthetic appeal. The geometry of a shell structure has a significant impact on its mechanical behavior and visual effect; therefore, determining a reasonable structural shape and thickness distribution under given load and boundary conditions is a key challenge in shell design.

[0003] Structural shape optimization offers an effective approach to solving this problem. Traditional shape optimization based on the finite element method typically associates design variables directly with mesh nodes, resulting in a large number of design variables and difficulty in effectively describing smooth geometry. With the development of isogeometric analysis, the same spline basis functions are used to simultaneously express the geometric and analytical models, achieving integrated design and analysis. However, most existing isogeometric structural shape optimizations use predefined surface patches as optimization objects and directly use surface control vertices as design variables. This approach struggles to retain the design intent driven by characteristic curves during actual modeling, and too many control vertices can lead to local numerical oscillations and undesirable curvature fluctuations, hindering the acquisition of high-quality optimized surfaces. Surface lofting is a common modeling technique that generates smooth surfaces from multiple cross-sectional curves, effectively reflecting the modeling logic driven by cross-sectional characteristic curves. Therefore, it is necessary to introduce surface lofting into the optimization loop to drive shell shape design with cross-sectional characteristic curves. However, the introduction of surface lofting also increases the difficulty of sensitivity analysis. Due to the complex algorithmic operations involved in the lofting process, the relationship between design variables and the objective function exhibits strong nonlinearity and implicitness, making it difficult to achieve direct and efficient solutions using a single analytical method.

[0004] Therefore, there is an urgent need to develop a lofted surface shell structure optimization design method based on hybrid sensitivity to achieve high-quality and efficient optimization design of shell structures. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention proposes a lofted surface shell structure optimization design method based on hybrid sensitivity. This method incorporates the surface lofting process into the geometric optimization loop of the shell structure, uses the control vertex coordinate offset of the cross-sectional curve as the design variable, and combines it with a hybrid sensitivity analysis strategy to achieve high-quality and efficient optimization design of the shell structure.

[0006] To achieve the above-mentioned technical effects, the present invention adopts the following technical solution:

[0007] This invention first discloses a method for optimizing the design of lofted curved surface shell structures based on hybrid sensitivity, comprising the following steps:

[0008] Step 1: Select a set of cross-sectional curves, use the offset of the control vertex coordinates of the cross-sectional curves as the design variable, and generate the initial lofted surface through lofting;

[0009] Based on the modeling requirements of the shell structure, a set of initial cross-sectional curves for surface lofting is selected, among which,

[0010] The control vertex of the initial cross-sectional curve is denoted as The offset of the control vertex coordinates of the initial cross-section curve is defined as a design variable. All cross-sectional curves are sorted and their parameter directions are unified. The control vertices of each cross-sectional curve are represented as homogeneous coordinates. The polynomial degree of all cross-sectional curves is unified by increasing the order, and the node vectors and the number of control vertices of all cross-sectional curves are unified by node insertion. The degree is selected in the lofting direction and the corresponding node vector is constructed to determine the parameter values ​​of each cross-sectional curve in the lofting direction. The homogeneous control vertex sequences corresponding to the same control vertex number are interpolated to generate the initial lofting surface.

[0011] Step 2: Refine the initial lofted surface to construct a refined lofted surface model;

[0012] The initial lofted surface is refined using a node insertion or order-increasing algorithm to obtain a refined lofted surface model for structural analysis, and control vertices of the initial lofted surface are established. Control Vertex of Refined Lofted Surface The mapping relationship between them;

[0013] Step 3: Based on the refined lofted surface model, structural response analysis is performed using the isogeometric shell theory;

[0014] Based on the refined lofted surface model, the isogeometric Reissner-Mindlin shell theory is used to perform isogeometric structural response analysis on the shell structure under given boundary conditions and load conditions.

[0015] Step 4: Calculate the derivative of the objective function with respect to the design variables, and decompose the derivative into analytical sensitivity terms and finite difference sensitivity terms;

[0016] The hybrid sensitivity described in this invention specifically refers to decomposing the derivative of the objective function with respect to the design variables into analytical sensitivity terms and finite difference sensitivity terms. Specifically, the objective function... For design variables The derivative decomposition is the objective function with respect to the control vertices of the initial lofted surface. The derivative and the control vertex of the initial lofted surface For design variables The product of the derivatives, i.e.

[0017]

[0018] in, The adjoint analytical sensitivity method was used for calculation. The calculation is performed using the finite difference method;

[0019] Step 5: Under constraints, perform iterative optimization based on the optimization objective. Repeat steps 1 to 4 according to the updated design variables until the convergence condition is met, and output the optimized lofted surface and its structural response. Under geometric constraints, with minimum compliance as the optimization objective, update the design variables using the moving asymptote method. The process repeats steps 1 to 4 based on the updated design variables until the convergence condition is met or the maximum number of iterations is reached, and outputs the optimized cross-sectional curve, lofted surface and its corresponding structural response.

[0020] Preferably, in step 4, the hybrid sensitivity analysis will use the objective function For design variables The derivative is decomposed into an analytical sensitivity term and a finite-difference sensitivity term, expressed as:

[0021]

[0022] Wherein, the objective function f is the structural response function of the shell structure optimization objective, which is preferably the compliance in this invention. The derivative of the objective function f with respect to the control vertex P of the initial lofted surface is calculated using the adjoint analytical sensitivity method. The derivative of the initial lofted surface control vertex P with respect to the design variable h is calculated using the finite difference method.

[0023] Preferably, in step 4, the analytical sensitivity term The adjoint analytical sensitivity method is used for calculation. The vertices are controlled based on the initial lofted surface. Control Vertex of Refined Lofted Surface Mapping relationship between Find the control vertices of the refined lofted surface. Control vertices of the initial lofted surface The derivative of the objective function, and combined with the objective function Control vertices for refining lofted surfaces Accompanying sensitivity This allows us to obtain the derivative of the objective function with respect to the control vertices of the initial lofted surface. .

[0024] Preferably, in step 4, the finite difference sensitivity term The calculation is performed using the finite difference method, expressed as follows:

[0025]

[0026] in, For finite difference step size, This indicates that when the design variable takes the value h, the initial lofted surface controls the vertices obtained by lofting the updated cross-sectional curve. This indicates applying a finite difference step size to the current design variable. Then, the vertices are controlled by the initial lofted surface obtained by lofting the corresponding updated cross-sectional curve.

[0027] Preferably, the finite difference step size δh is scaled according to the physical dimensions of the model.

[0028]

[0029] Where L represents the maximum side length of the initial configuration of the lofting model. Compared with the prior art, the beneficial effects of the present invention are as follows:

[0030] (1) This invention integrates the surface lofting process into the geometric shape optimization loop of shell structure, etc., and drives the shell shape design with the cross-sectional feature curve. This is beneficial to retain the design intent in the modeling process, significantly reduce the design space dimension, generate high-quality and manufacturable surfaces, and achieve more seamless and compact CAD / CAE integration.

[0031] (2) The present invention adopts a hybrid sensitivity analysis strategy based on finite difference and adjoint analysis, which improves the feasibility and computational efficiency of sensitivity solution while ensuring the accuracy of sensitivity calculation.

[0032] (3) The present invention adopts a multi-layer geometric representation that combines a coarse design model and a fine analysis model, which reduces the optimization calculation cost while ensuring the accuracy of structural analysis, and is conducive to realizing efficient and high-precision optimization design of shell structure. Attached Figure Description

[0033] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the embodiments will be briefly introduced below. The features and advantages of the present invention can be more clearly understood by referring to the accompanying drawings. The accompanying drawings are schematic and should not be construed as limiting the present invention in any way. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0034] Figure 1 This is a flowchart of a lofted surface shell structure optimization design method based on hybrid sensitivity.

[0035] Figure 2 This is a schematic diagram of the surface lofting modeling process, where a, b, c, and d represent: the initial cross-sectional curve diagram used for lofting, the cross-sectional curve diagram with unified degree, node vector, and number of control vertices after parameter direction unification, order increase, and node insertion, the lofting surface diagram with degree 3×1 constructed based on the unified cross-sectional curve, and the lofting surface diagram with degree 3×2 constructed based on the unified cross-sectional curve.

[0036] Figure 3 This is a schematic diagram of the main workflow for optimizing the geometry of shell structures and other objects based on surface lofting technology.

[0037] Figure 4 This is a schematic diagram of the roof model, where a, b, and c represent: a schematic diagram of the cross-sectional curves constituting the roof model and their design variables, a schematic diagram of the initial roof surface model generated by the cross-sectional curves through lofting, and a schematic diagram of the boundary conditions and load conditions of the roof surface, respectively.

[0038] Figure 5 This is a schematic diagram of the roof model optimization process, where a and b represent: the curves of the compliance of the analysis model at different refinement levels as a function of the number of iterations, and the schematic diagrams of the evolution of the roof model geometry and its displacement response under several typical iteration steps.

[0039] Figure 6 This is a schematic diagram of the roof model optimization results, where a, b, and c represent: a schematic diagram comparing the cross-sectional curves before and after optimization, a schematic diagram of the optimized lofted roof surface, and a schematic diagram comparing the displacement response of the roof structure before and after optimization, respectively. Detailed Implementation

[0040] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments of the present invention and the features thereof can be combined with each other.

[0041] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and therefore the scope of protection of the invention is not limited to the specific embodiments disclosed below.

[0042] This invention provides a method for optimizing the design of lofted surface shell structures based on hybrid sensitivity, such as... Figure 1As shown, this method integrates cross-sectional curve design, surface lofting, model refinement, isogeometric analysis, sensitivity analysis, and optimization iteration into a unified computational framework to achieve high-quality and efficient optimization design of shell structures. Specifically, design variables are associated with the control vertices of the cross-sectional curve, and the cross-sectional curve is updated accordingly; an initial lofted surface is generated through surface lofting, and the initial lofted surface is refined to establish an analysis model; the isogeometric Reissner-Mindlin shell theory is used for structural response analysis; the derivative of the objective function with respect to the design variables is calculated through sensitivity analysis; the design variables are updated using the optimization algorithm, and the above process is repeated until the convergence condition is met or the maximum number of iterations is reached, outputting the optimized cross-sectional curve, lofted surface, and their corresponding structural response.

[0043] like Figure 2 As shown, to illustrate the modeling process of surface lofting in this invention, an example of constructing a lofted surface using four NURBS cross-sectional curves is given. The four cross-sectional curves are denoted as follows: , , and Their polynomial degrees are respectively The corresponding node vectors are respectively , , , After order enhancement and node insertion, all cross-sectional curves have the same degree. and the same node vector Using these unified cross-sectional curves, in Number of times selected in each direction and Two sheets with exponents of 1 and 2 can be constructed. and The lofted surface.

[0044] like Figure 3 As shown, the optimization process of this invention uses cross-sectional curves to control the vertices. The design variables are composed of coordinate offsets. Starting from, among which, This represents the control vertex of the initial cross-section curve before sorting, unified parameter direction, order increase, and node insertion processing. In each optimization iteration, the cross-section curve is updated according to the current design variables, and then an initial lofted surface is generated through surface lofting. Its control vertex is denoted as... The initial lofted surface is refined through node insertion or order-promotion operations to construct a refined lofted surface model for structural analysis, with its control vertices denoted as... Initial lofted surface control vertices Control Vertex of Refined Lofted Surface They satisfy a mapping relationship

[0045]

[0046] in, To control the vertex transformation matrix. The control vertex transformation matrix... The refinement operation from the initial lofted surface to the refined lofted surface is determined. When the refined lofted surface is obtained from the initial lofted surface through node insertion or order increase, the refined control vertex Q can be represented as the control vertex P of the initial lofted surface obtained through linear transformation. The transformation matrix is ​​formed by inserting the corresponding nodes or combining the higher-order transformation matrices in the refinement order. Based on the refined lofted surface model, structural analysis is performed using the isogeometric Reissner-Mindlin shell theory to obtain the structural displacement response. The objective function is then... For design variables The derivative is decomposed into

[0047]

[0048] in, The adjoint analytical sensitivity method was used for calculation. The finite difference method is used for calculation. The finite difference part is specifically represented as follows:

[0049]

[0050] in, The step size is finite difference. This indicates that when the design variable takes the value h, the initial lofted surface, obtained by lofting the updated initial cross-section curve, controls the vertices. This indicates applying a finite difference step size to the current design variable. Then, the vertices are controlled by the initial lofted surface obtained from the updated cross-sectional curve. After obtaining the sensitivity, the design variables are updated using the moving asymptote method, and the above process is repeated until the optimization is complete.

[0051] The effectiveness of the method proposed in this invention will be illustrated below using the optimization of the flexibility of a roof structure as an example.

[0052] like Figure 4 As shown, the roof model consists of six cross-sectional curves. Formed through surface lofting. The six cross-sectional curves are all cubic curves, uniformly distributed along the y-direction, and about... Planar symmetry. Cross-sectional curve. and Each has 5 control vertices, and the cross-sectional curves are... and Each has 7 control vertices, and the cross-sectional curves are... and Each section curve has 6 control vertices, for a total of 36 control vertices. A bicubic NURBS roof surface with 54 control vertices is generated by lofting this set of section curves. The roof surface is fixedly constrained on both sides and subjected to uniformly distributed vertical loads. The vertical coordinate offset of the control vertices within each cross-sectional curve is taken as design variables, totaling 24 design variables, with a value range of [missing value]. Geometric constraints are applied by limiting the range of values ​​for each design variable. To improve the accuracy of structural analysis, the initial roof surface is refined by inserting 3, 4, 5, 6, and 7 uniform nodes in each node interval, thus forming analysis models with different refinement levels.

[0053] Figure 5 The paper presents the compliance changes corresponding to different refinement levels of the analysis model, as well as the design model and displacement response evolution process corresponding to several iterations at representative refinement levels. In this embodiment, the absolute value of the compliance change between two adjacent iterations being less than a preset threshold is used as the convergence criterion, where the preset threshold is set to 10. -8 As iterations proceed, the compliance at each refinement level decreases significantly. Taking the level 4 refinement model as an example, the optimization ends at the 96th iteration after meeting the convergence condition, with the compliance decreasing from 1373.41J to 47.45J. The iterative evolution process shows that the geometry obtained in the 32nd and 80th iterations is essentially the same, indicating that the solution has reached a stable state.

[0054] Figure 6 The paper presents the initial and optimized cross-sectional curves, the optimized lofted roof, and a comparison of the roof's displacement response before and after optimization. After optimization using the method of this invention, the structural displacement response is significantly reduced, demonstrating that this invention can effectively reduce the flexibility of the roof structure and improve its mechanical performance while preserving the modeling logic driven by the cross-sectional characteristic curve.

[0055] In this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0056] In this invention, unless otherwise explicitly specified and limited, "above" or "below" the second feature can include direct contact between the first and second features, or contact between the first and second features through another feature between them. Furthermore, "above," "over," and "on top" of the second feature includes the first feature directly above or diagonally above the second feature, or simply indicates that the first feature is at a higher horizontal level than the second feature. "Below," "below," and "under" the second feature includes the first feature directly below or diagonally below the second feature, or simply indicates that the first feature is at a lower horizontal level than the second feature.

[0057] In this invention, the terms "first," "second," "third," and "fourth" are used for descriptive purposes only and should not be construed as indicating or implying relative importance. The term "multiple" refers to two or more unless otherwise expressly defined.

[0058] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for optimizing the design of lofted curved surface shell structures based on hybrid sensitivity, comprising the following steps: Step 1: Select a set of cross-sectional curves, use the offset of the control vertex coordinates of the cross-sectional curves as the design variable, and generate the initial lofted surface through lofting; Step 2: Refine the initial lofted surface to construct a refined lofted surface model; Step 3: Based on the refined lofted surface model, structural response analysis is performed using the isogeometric shell theory; Step 4: Calculate the derivative of the objective function with respect to the design variables, and decompose the derivative into analytical sensitivity terms and finite difference sensitivity terms; Step 5: Under constraints, perform iterative optimization based on the optimization objective. Repeat steps 1 to 4 according to the updated design variables until the convergence condition is met, and output the optimized lofted surface and its structural response.

2. The method for optimizing the design of a lofted curved surface shell structure according to claim 1, wherein: In step 1, the cross-sectional curves are sorted and the parameter directions are unified, and the control vertices of each cross-sectional curve are represented as homogeneous coordinates; the polynomial degree of all cross-sectional curves is unified by order increase, and the node vectors and the number of control vertices of all cross-sectional curves are unified by node insertion.

3. The method for optimizing the design of a lofted curved surface shell structure according to claim 2, wherein: Select the number of times in the lofting direction and construct the corresponding node vectors. Determine the parameter values ​​of each cross-sectional curve in the lofting direction. Use the homogeneous control vertex sequence corresponding to the same control vertex number as interpolation data points for interpolation to generate the initial lofting surface.

4. The method for optimizing the design of a lofted curved surface shell structure according to claim 1, wherein: In step 2, the initial lofted surface is refined using a node insertion or order-increasing algorithm to obtain the refined lofted surface model, and a mapping relationship is established between the control vertices of the initial lofted surface and the control vertices of the refined lofted surface.

5. The method for optimizing the design of a lofted curved surface shell structure according to claim 1, wherein: In step 3, the structural response analysis is performed using the isogeometric Reissner-Mindlin shell theory.

6. The method for optimizing the design of a lofted curved surface shell structure according to claim 1, wherein: In step 4, the derivative of the objective function with respect to the design variables is expressed as follows: the derivative of the objective function with respect to the design variables is decomposed into the product of the derivative of the objective function with respect to the control vertices of the initial lofting surface and the derivative of the control vertices of the initial lofting surface with respect to the design variables.

7. The method for optimizing the design of a lofted curved surface shell structure according to claim 6, wherein: The derivative of the objective function with respect to the control vertex of the initial lofted surface is calculated using the adjoint analytical sensitivity method. Based on the mapping relationship between the control vertex of the initial lofted surface and the control vertex of the refined lofted surface, the derivative of the control vertex of the refined lofted surface with respect to the control vertex of the initial lofted surface is obtained. Combined with the adjoint sensitivity of the objective function with respect to the control vertex of the refined lofted surface, the derivative of the objective function with respect to the control vertex of the initial lofted surface is obtained.

8. The method for optimizing the design of a lofted curved surface shell structure according to claim 6, wherein: The derivative of the initial lofted surface control vertex with respect to the design variables is calculated using the finite difference method.

9. The method for optimizing the design of a lofted curved surface shell structure according to claim 8, wherein: The finite difference step size in the finite difference method is scaled according to the physical size of the model.

10. The method for optimizing the design of a lofted curved surface shell structure according to claim 1, wherein: In step 5, under given geometric constraints, the design variables are updated using the moving asymptote method with minimum compliance as the optimization objective.