A building scheme generation method and system based on morphological structure cooperation

Abstract

The application discloses a building scheme generation method and system based on morphological structure cooperation. The method comprises the following steps: calculating building morphological parameters and their value ranges according to building function requirements and site constraint conditions to construct a first variable; constructing a second variable according to preset roof structure types and corresponding geometric structure parameters; the roof structure types at least include a double-layer net shell structure, a spatial truss beam structure and an arch truss combined structure; establishing a multi-objective optimization function; optimization objectives of the multi-objective optimization function include minimum building surface area, minimum structural steel consumption and minimum structural strain energy; and generating a building scheme by globally iteratively optimizing the first variable and the second variable according to the multi-objective optimization function. The embodiment of the application can support automatic comparison and selection of various structure types, and can synchronously optimize building morphological parameters and roof structure types in the same framework, thereby realizing morphological and structural parameterization cooperation.

Description

Technical Field

[0001] This invention relates to the field of architectural design technology, and in particular to a method and system for generating architectural schemes based on morphological and structural collaboration. Background Technology

[0002] In traditional architectural design, architects typically determine the building form first, and then structural engineers are involved in structural design and calculation.

[0003] In this model, when structural analysis reveals problems with the morphological scheme, significant modifications to the original scheme are often required, which prolongs the engineering design cycle and results in a certain degree of resource waste. Summary of the Invention

[0004] The present invention aims to provide a method and system for generating architectural schemes based on morphological and structural collaboration, which can support automatic comparison of multiple structural types and simultaneously optimize architectural morphological parameters and roof structure types within the same framework, thereby achieving parametric collaboration between morphology and structure.

[0005] In a first aspect, embodiments of the present invention provide a method for generating architectural schemes based on morphological and structural coordination, including: Based on the building's functional requirements and site constraints, the building's morphological parameters and their value ranges are calculated to construct the first variable; A second variable is constructed based on the preset roof structure type and its corresponding geometric parameters; the roof structure type includes at least a double-layer reticulated shell structure, a space truss beam structure, and an arch truss combination structure. A multi-objective optimization function is established; the optimization objectives of the multi-objective optimization function include minimizing the building's external surface area, minimizing the amount of steel used in the structure, and minimizing the structural strain energy. Based on the multi-objective optimization function, a global iterative optimization is performed on the first variable and the second variable to generate a building scheme.

[0006] As an improvement to the above scheme, the step of calculating building form parameters and their value ranges based on building functional requirements and site constraints to construct the first variable includes: Generate the grandstand cross-sectional geometry and roof outline based on the building's functional requirements and site constraints. Based on the grandstand cross-sectional geometry and the roof outline, the building form parameters and their value ranges are obtained; the building form parameters include roof height control parameters and roof outline boundary control parameters; The architectural morphology parameters and their value ranges are encapsulated as the first variable.

[0007] As an improvement to the above solution, the step of generating the grandstand cross-sectional geometry and roof outline based on building functional requirements and site constraints includes: Obtain building seating parameters and building site core parameters to obtain building functional requirements; the building seating parameters include the number of seats, seat type, seat row spacing and line of sight slope; Obtain the building height limit, roof covering method, site boundary, setback requirements and the location of surrounding road interfaces to obtain the site constraints; Based on the building seating parameters and the site constraints, calculate the elevation of each row of seats and generate the grandstand profile geometry. Based on the grandstand profile geometry, the building core parameters, and the site constraints, a roof profile covering the core area and the seating area is generated.

[0008] As an improvement to the above solution, the step of constructing a second variable based on a preset roof structure type and its corresponding geometric parameters includes: When the roof structure type is selected as a double-layer reticulated shell structure, the grid density, the distance between the upper and lower layers, and the node connection method are obtained as geometric structural parameters. When the roof structure type is selected as a spatial truss beam structure, the main truss spacing, main truss height, and secondary truss spacing are obtained as geometric structural parameters. When the roof structure type is selected as an arch truss combination structure, the main arch rise, main arch spacing, main arch landing height, and stabilizing truss type are obtained as geometric structural parameters; the stabilizing truss type includes horizontal support and diagonal support. Each roof structure type and its corresponding geometric parameters are encapsulated as a second variable.

[0009] As an improvement to the above scheme, the establishment of the multi-objective optimization function includes: Geometric analysis is used to calculate the building's exterior area and spatial volume of the proposed building scheme. Based on the stated spatial volume and the preset minimum spatial volume, establish volume constraints. The structural performance parameters of the building scheme to be optimized are calculated using finite element analysis; the structural performance parameters include the amount of steel used in the structure, the structural strain energy, and the maximum vertical displacement. Based on the strain energy of the structure, the stress ratio is calculated, and strength constraint conditions based on the stress ratio are established; Based on the maximum vertical displacement and the preset deflection limit, establish displacement constraint conditions; By combining the volume constraint, the strength constraint, and the displacement constraint, the constraint conditions are obtained; The building's external surface area, the amount of steel used in the structure, and the structural strain energy are each taken as optimization objects. With the goal of minimizing each of the optimization objects, and in combination with the constraints, a multi-objective optimization function is established for the building scheme to be optimized.

[0010] As an improvement to the above scheme, the step of generating a building scheme by performing global iterative optimization on the first and second variables according to the multi-objective optimization function includes: An initial population is generated based on the first variable and the second variable; the building morphology parameters of each individual in the initial population are randomly generated based on the first variable, and the roof structure type of each individual is generated based on the building morphology parameters and the second variable. Before the preset genetic iteration stopping condition is reached, the fitness of each individual in the current population is calculated according to the multi-objective optimization function. Based on the fitness, through non-dominated sorting, female and male individuals are selected from the current population, and the female and male individuals are cross-crossed to generate offspring individuals; The offspring individuals are mutated, and based on the mutated offspring individuals, an elite strategy is adopted to update the current population; When the preset genetic iteration stopping condition is met, the building scheme is obtained based on the optimal non-dominated solution set of the current population.

[0011] As an improvement to the above scheme, the step of generating the initial population based on the first variable and the second variable includes: Based on the first variable, several building form parameters are randomly generated; Based on the architectural morphology parameters and the second variable, select the roof structure type; Generate a roof geometric model based on the building morphology parameters and the corresponding roof structure type; Based on the roof geometry model, the maximum vertical displacement and the stress ratio of each component are obtained; Calculate the support position based on the maximum vertical displacement, and add supports in the roof geometry model based on the support position; Based on the stress ratio, the component's cross-section is optimized in the roof geometry model by cross-section scaling or material replacement. Based on the aforementioned architectural morphology parameters and the corresponding roof geometric structure model, several individuals are generated to form an initial population.

[0012] As an improvement to the above scheme, the step of calculating the fitness of each individual in the current population according to the multi-objective optimization function before the preset genetic iteration stopping condition is reached includes: Before the preset genetic iteration stopping condition is met and the number of generations is not greater than the preset generation threshold, the first function value is obtained by performing finite element analysis on the individuals in the current population and calculating the multi-objective optimization function of each individual. Add the individuals in the current population and their corresponding first function values ​​to the training sample database; A Kriging surrogate model is constructed for each optimization objective to predict the multi-objective optimization function for each objective; the hyperparameter updates of the Kriging surrogate model respond to the updates of the training sample database. Before reaching the preset genetic iteration stopping condition and after the number of generations is greater than the preset generation threshold, the multi-objective optimization function of each individual in the current population is predicted through the surrogate model to obtain the second function value and its uncertainty. Based on the second function value and the uncertainty, calculate the expected improvement of aggregation for each individual in the current population; If the expected improvement of the aggregate for the individual is not less than the expected adaptive improvement threshold, then the second function value of the individual is updated through finite element analysis, and the second function value is added to the training sample database; the expected adaptive improvement threshold is calculated based on the number of iterations. Calculate the relative error of the second function value before and after the update. If the relative error is greater than a preset accuracy threshold, perform finite element analysis on all individuals in the current population and update the surrogate model based on the finite element analysis results. The fitness of each individual in the current population is obtained based on the first function value or the second function value.

[0013] As an improvement to the above scheme, the step of mutating the offspring individuals and updating the current population using an elite strategy based on the mutated offspring individuals includes: Extract the morphological features of each individual in the current population, and divide the current population into several morphological sub-regions based on the morphological features; the morphological features include the roof sag-to-span ratio, the roof plan aspect ratio, and the roof curvature variation coefficient; Every preset probability update cycle, the number of selected roof structure types in each morphological sub-region of the current non-dominated solution set is counted to obtain the Pareto contribution statistics. Based on the Pareto contribution statistics, calculate the adaptive selection probability of each roof structure type in each of the morphological sub-regions; Based on the adaptive selection probability, the roof structure type of the offspring individuals is mutated; The mutated offspring, mother, and father individuals are merged, and an elite strategy is adopted to update the current population.

[0014] Secondly, embodiments of the present invention also provide a building scheme generation system based on morphological structure collaboration, comprising: The first variable construction module is used to calculate the building form parameters and their value range based on the building's functional requirements and site constraints, and to construct the first variable. The second variable construction module is used to construct a second variable based on the preset roof structure type and its corresponding geometric parameters; the roof structure type includes at least a double-layer reticulated shell structure, a space truss beam structure and an arch truss combination structure. The optimization function definition module is used to establish a multi-objective optimization function; the optimization objectives of the multi-objective optimization function include minimizing the building's external surface area, minimizing the amount of structural steel used, and minimizing the structural strain energy. The building scheme generation module is used to perform global iterative optimization on the first variable and the second variable according to the multi-objective optimization function to generate building schemes.

[0015] Compared with existing technologies, this invention discloses a method and system for generating architectural schemes based on morphological-structural synergy. It constructs a first variable by calculating architectural morphological parameters and their value ranges according to architectural functional requirements and site constraints; constructs a second variable based on preset roof structure types and their corresponding geometric parameters; the roof structure types include at least double-layer reticulated shell structures, spatial truss beam structures, and arch-truss composite structures; establishes a multi-objective optimization function; the optimization objectives of the multi-objective optimization function include minimizing the building's external surface area, minimizing structural steel consumption, and minimizing structural strain energy; and performs global iterative optimization on the first and second variables according to the multi-objective optimization function to generate architectural schemes. Using this invention, automatic comparison of multiple structural types can be supported, and architectural morphological parameters and roof structure types can be simultaneously optimized within the same framework, achieving parametric synergy between morphology and structure. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating the steps of a method for generating architectural schemes based on morphological and structural collaboration, as provided in an embodiment of the present invention. Figure 2 This is a schematic diagram of a building scheme generation system based on morphological structure collaboration provided in an embodiment of the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] In the description and claims, it should be understood that the terms "first," "second," etc., used in the description and claims are only for the purpose of distinguishing the description of the same technical features, and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated, nor necessarily the order of description or chronological order. The terms are interchangeable where appropriate. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature.

[0019] Current architectural design methods often involve architects determining the initial form of a design, followed by structural designers selecting and optimizing the structure based on that form. This often results in an overly simplistic choice of structural system, and the judgment of whether a structural solution is reasonable is limited to the optimization of a single structural system.

[0020] Furthermore, under this traditional design model, when structural analysis reveals problems with the morphological scheme, significant modifications to the original scheme are often required, thereby extending the engineering design cycle and causing a certain degree of resource waste.

[0021] Based on the above considerations, this invention provides a method for generating architectural schemes based on morphological and structural collaboration. Please refer to... Figure 1 In this embodiment, the architectural scheme generation method based on morphological structure collaboration is specifically executed through steps S1 to S4: S1. Based on the building's functional requirements and site constraints, calculate the building's morphological parameters and their value ranges to construct the first variable; S2. Construct a second variable based on the preset roof structure type and its corresponding geometric parameters; the roof structure type includes at least a double-layer reticulated shell structure, a space truss beam structure, and an arch truss combination structure. S3. Establish a multi-objective optimization function; the optimization objectives of the multi-objective optimization function include minimizing the building's external surface area, minimizing the amount of structural steel used, and minimizing the structural strain energy; S4. Based on the multi-objective optimization function, perform global iterative optimization on the first variable and the second variable to generate a building scheme.

[0022] Building functional requirements refer to the usage requirements that a building must meet. Taking a stadium as an example, building functional requirements include the size requirements for multi-functional venues, the size requirements for competition venues, seating requirements, and sightline quality requirements. Site constraints refer to the external limitations of the building site, such as the site boundary, building height restrictions, and setback requirements.

[0023] The architectural form parameters calculated based on functional requirements and site constraints can control the external shape of the building. It should be noted that these architectural form parameters are the starting point for architectural design, and to a certain extent determine basic information such as the roof's span, height, and outline.

[0024] Each building form parameter has its own range of values, that is, the maximum and minimum values ​​that the parameter is allowed to vary. The range of values ​​constitutes the feasible region of the parameter, and the specific building form parameter is selected within this feasible region during subsequent global iterative optimization.

[0025] Roof structures for large-span buildings vary widely, with significant differences in adaptability to building form, steel consumption, and construction difficulty. Current design processes lack the ability to systematically compare and select from multiple structural types within the same parametric framework.

[0026] This invention provides three main types of roof structures. The double-layer reticulated shell structure is suitable for curved roofs, consisting of two layers of nodal grids connected by web members. The spatial truss beam structure is suitable for roofs spanning one or two directions, consisting of a main truss and secondary trusses forming an orthogonal or oblique system. The arch truss composite structure is suitable for large-span, one-way roofs, consisting of a main arch and stabilizing trusses.

[0027] Different roof structure types correspond to different structural description parameters, that is, geometric structural parameters. In this embodiment of the invention, the roof structure type and its corresponding geometric structural parameters are encapsulated as a second variable, enabling the synchronous generation and adjustment of the corresponding geometric structural parameters when different roof structure types are selected.

[0028] The embodiments of this invention integrate three mainstream structural types—double-layer reticulated shell, spatial truss beam, and arch truss combination—within the same parametric framework. It supports roof structure type as a discrete optimization variable to participate in the optimization process, and can obtain the optimal building scheme under the synergy of morphology and structure.

[0029] Existing design methods typically treat carbon emission assessment as a post-design evaluation step, rather than a pre-design constraint. It should be noted that this invention transforms carbon emission control into a calculable optimization objective, achieving the quantitative embedding of carbon emission targets.

[0030] During the building operation phase, the operational carbon emission potential is positively correlated with the building surface area, expressed as: ; in, The building surface area; The heat transfer coefficient of the building envelope; This refers to the number of heating days. This is an energy carbon emission factor. Therefore, minimizing the building's external surface area is equivalent to minimizing its operational carbon emission potential.

[0031] During the building construction phase, implicit carbon emissions are represented as follows: ; in, This refers to the amount of steel used in the structure. The carbon emission factor for steel (approximately 2.0-2.5 kg) / kg). Therefore, minimizing the amount of steel used in structures is equivalent to minimizing implicit carbon emissions.

[0032] The optimization objectives of the multi-objective optimization function in this invention include minimizing the building's external surface area, minimizing the amount of structural steel used, and minimizing the structural strain energy. Minimizing the building's external surface area corresponds to minimizing operational carbon emission potential, minimizing the amount of structural steel used corresponds to minimizing implicit carbon emissions, and minimizing the structural strain energy corresponds to optimal structural efficiency.

[0033] The first variable corresponds to the building form, and the second variable corresponds to the structural design. By jointly iteratively optimizing the first and second variables, and using a multi-objective optimization function to evaluate the generated schemes during the optimization process, the optimal form-structure combination with multi-objective joint performance can be automatically searched in the global design space.

[0034] The above scheme can support automatic comparison of multiple structural types and simultaneously optimize building form parameters and roof structure type under the same framework, breaking the traditional design process of separating architecture and structure and realizing parametric collaboration between form and structure.

[0035] As a preferred implementation, step S1 involves calculating the building form parameters and their value ranges based on the building's functional requirements and site constraints, constructing the first variable, and then executing steps S11-S13: S11. Generate the grandstand cross-sectional geometry and roof outline based on the building's functional requirements and site constraints. S12. Based on the grandstand cross-sectional geometry and the roof outline, obtain the building form parameters and their value ranges; the building form parameters include roof height control parameters and roof outline boundary control parameters; S13. Encapsulate the building form parameters and their value range as a first variable.

[0036] In this preferred embodiment of the invention, a stadium / gymnasium architectural design is used as an example. The basic form of the building is determined by the sectional geometry of the stands and the roof outline, and then the architectural form parameters are further extracted.

[0037] Grandstand sectional geometry refers to the shape of the grandstand in the vertical section of the building, which can indicate the position of each row of seats; the roof outline can indicate the projection boundary of the entire building on the horizontal plane, which covers all the area covered by the building.

[0038] After determining the basic grandstand cross-sectional geometry and roof outline, the building's geometry still has adjustable components. For example, the roof can be raised or lowered as a whole while meeting coverage requirements, and the roof boundary can expand outwards or contract inwards within the land boundary line. In this embodiment of the invention, this adjustable component is quantified as building morphology parameters, and the adjustable range of each parameter is clearly defined, providing a clear search boundary for subsequent optimization steps.

[0039] Specifically, the roof height control parameter is used to control the vertical shape of the roof. Understandably, the supported roof structure types may differ in different roof height scenarios; the roof outline boundary control parameter is used to control the planar shape of the roof.

[0040] The aforementioned architectural form parameters and their value ranges are encapsulated as the first variable, representing the design freedom at the architectural form level. It should be noted that the first variable is a continuous variable, which can be adjusted infinitely within its value range.

[0041] Preferably, in the global iterative optimization, the first variable is encoded as a real number.

[0042] In the above scheme, the architectural form design is transformed into a quantifiable parameter expression, clarifying the adjustable range of each parameter and providing a clear search boundary for subsequent optimization steps. Furthermore, the encapsulated set of variables can be reused in different stadium projects or different optimization scenarios, laying the foundation for the coordinated optimization of the building's form and structure.

[0043] Further, preferably, step S11, generating the grandstand cross-sectional geometry and roof outline based on building functional requirements and site constraints, includes: Obtain building seating parameters and building site core parameters to obtain building functional requirements; the building seating parameters include the number of seats, seat type, seat row spacing and line of sight slope; Obtain the building height limit, roof covering method, site boundary, setback requirements and the location of surrounding road interfaces to obtain the site constraints; Based on the building seating parameters and the site constraints, calculate the elevation of each row of seats and generate the grandstand profile geometry. Based on the grandstand profile geometry, the building core parameters, and the site constraints, a roof profile covering the core area and the seating area is generated.

[0044] In a preferred embodiment of the invention, the building functional requirements include building seating parameters and building field core parameters. The building seating parameters include the number of seats, seating type (fixed / movable), seat row spacing, and line-of-sight slope. The building field core parameters include the length and width of the competition area and the size requirements of the multi-functional venue, which determine the core area that the roof must cover.

[0045] As a preferred implementation method, obtain building seating parameters. and building site core parameters .in, To maintain a fixed number of seats, For the number of movable seats, Seat spacing, To create a slope for the line of sight; For the core length, The width of the field core.

[0046] Site constraints limit the adjustable range of the building's form, and building height restrictions and setback requirements also affect the highest point of the grandstand. The generated grandstand cross-sectional geometry must simultaneously meet the requirements for visual quality and site constraints.

[0047] Preferably, the grandstand cross-sectional geometry can be generated based on the elevation of each row of seats, where the elevation of the i-th row of seats is... Represented as: ; ; in, This represents the height of the field core plane. The line-of-sight rise angle of the j-th row must meet the C-value (line-of-sight rise height) requirement; The height of the viewpoint focus. The horizontal distance between the (i-1)th row of seats and the focal point of sight is... The minimum C value requirement is (usually taken as 0.06 / 0.12m).

[0048] The roof profile is the shape of the roof's projected boundary on a horizontal plane. Based on the known grandstand cross-sectional geometry and building core parameters, merging the core area and seating area forms the total area that needs to be covered. Then, further considering site constraints, including the land boundary line and setback requirements, the merged area boundary is extended outwards by a certain amount of eaves projection to form the final roof profile.

[0049] Preferably, based on the grandstand profile geometry and field core range Generate a roof outline that completely covers the playing field and seating area, represented as: ; in, The overhang of the eaves.

[0050] In the above scheme, the functional requirements and site constraints of the initial design of the stadium building are transformed into a parametrically driven geometric generation process, which lays a geometric foundation for the subsequent extraction and optimization of building form parameters.

[0051] As a preferred implementation, step S2, constructing a second variable based on a preset roof structure type and its corresponding geometric parameters, includes: When the roof structure type is selected as a double-layer reticulated shell structure, the grid density, the distance between the upper and lower layers, and the node connection method are obtained as geometric structural parameters. When the roof structure type is selected as a spatial truss beam structure, the main truss spacing, main truss height, and secondary truss spacing are obtained as geometric structural parameters. When the roof structure type is selected as an arch truss combination structure, the main arch rise, main arch spacing, main arch landing height, and stabilizing truss type are obtained as geometric structural parameters; the stabilizing truss type includes horizontal support and diagonal support. Each roof structure type and its corresponding geometric parameters are encapsulated as a second variable.

[0052] A double-layer reticulated shell structure is a spatial structure suitable for curved roof forms, consisting of two layers of nodal mesh connected by web members. When a double-layer reticulated shell structure is selected as the roof structure type for the current scheme during optimization, the mesh density indicates how many mesh elements the curved surface of the reticulated shell is divided into; the spacing between the upper and lower layers reflects the overall thickness of the double-layer reticulated shell; and the nodal connection method reflects the stress performance and construction technology of the nodes. Nodal connection methods include bolted ball joints and welded ball joints.

[0053] Spatial truss beam structures are suitable for structural systems that span unidirectional or bidirectional roof forms. When a spatial truss beam structure is selected as the roof structure type for the current scheme during the optimization process, the horizontal distance between the main load-bearing components is reflected by the spacing of the main trusses; the vertical section height of the trusses is reflected by the height of the main trusses; and the arrangement density of the secondary load-bearing components is reflected by the spacing of the secondary trusses.

[0054] Arch truss composite structures are suitable for large-span, unidirectional roof structures. When an arch truss composite structure is selected as the roof structure type for the current scheme during optimization, the rise of the main arch reflects its initial height; a larger rise results in higher arch load-bearing efficiency but also increases building height. The spacing between adjacent main arches reflects the horizontal distance between them; a smaller spacing indicates a denser structure. The landing point height of the main arch reflects the elevation at the arch foot support. The out-of-plane stability of the main arch is reflected by the type of stabilizing truss, which includes horizontal and diagonal bracing.

[0055] In this embodiment of the invention, the selection variable for the roof structure type and the corresponding geometric parameters for each type are encapsulated together as a second variable. It should be noted that the second variable is a discrete variable, and the specific values ​​of the geometric parameters depend on the currently selected roof structure type.

[0056] In a preferred embodiment of the invention, the second variable is encoded as an integer variable during subsequent global iterative optimization. For example, through... The selection corresponds to three types of roof structure.

[0057] The above scheme integrates three mainstream large-span roof structure types—double-layer reticulated shell, spatial truss beam, and arch truss combination—within the same parametric framework, treating the selection of structure type as a discrete optimization variable in the optimization process. This allows the optimization algorithm to automatically determine which structure type is more advantageous based on the building's morphological conditions and simultaneously optimize its geometric parameters after selecting the type. This overcomes the limitations of traditional design processes where the structure type is pre-selected by the designer and cannot be dynamically adjusted during optimization, achieving systematic comparison and collaborative optimization at the structural level.

[0058] As a preferred implementation, step S3, establishing a multi-objective optimization function, includes: Geometric analysis is used to calculate the building's exterior area and spatial volume of the proposed building scheme. Based on the stated spatial volume and the preset minimum spatial volume, establish volume constraints. The structural performance parameters of the building scheme to be optimized are calculated using finite element analysis; the structural performance parameters include the amount of steel used in the structure, the structural strain energy, and the maximum vertical displacement. Based on the strain energy of the structure, the stress ratio is calculated, and strength constraint conditions based on the stress ratio are established; Based on the maximum vertical displacement and the preset deflection limit, establish displacement constraint conditions; By combining the volume constraint, the strength constraint, and the displacement constraint, the constraint conditions are obtained; The building's external surface area, the amount of steel used in the structure, and the structural strain energy are each taken as optimization objects. With the goal of minimizing each of the optimization objects, and in combination with the constraints, a multi-objective optimization function is established for the building scheme to be optimized.

[0059] It should be noted that during the iterative optimization process, building schemes to be optimized will be continuously generated. For each building scheme to be optimized generated, its spatial performance parameters (including the building's external surface area and spatial volume) and structural performance parameters (including the amount of structural steel, structural strain energy, and maximum vertical displacement) will be calculated.

[0060] The building's external surface area refers to the total surface area of ​​the building's external envelope, including the roof and exterior wall areas. In a preferred embodiment of the invention, the building's external surface area is obtained through geometric calculation using a three-dimensional model. Spatial volume refers to the volume of space within the building that can be used for competitions and spectators; exemplarily, the space statistics of the competition hall are obtained through integral calculation using a volumetric model.

[0061] Volumetric constraints are used to ensure that the architectural design meets basic functional requirements. The preset minimum volume is a baseline value determined based on the dimensions of the playing field, seating capacity, and sightline quality requirements. For example, the volumetric constraint is expressed as: ; in, This is the preset minimum space volume. The spatial volume of the scheme to be optimized. These are the design parameters for the scheme to be optimized.

[0062] In some preferred embodiments, the structural performance parameters of the building scheme to be optimized are calculated using the Karamba3D finite element analysis plugin.

[0063] The amount of steel used in the structure is the sum of the weights of all components, obtained by multiplying the volume of each component by the density of steel and then summing the results, expressed as: ; in, Density of steel; To construct the cross-sectional area of ​​i; Let i be the length of the constructed value.

[0064] Structural strain energy reflects the overall stiffness efficiency of the structure. It is obtained by integrating the deformation energy of each component under axial force and bending moment, and is expressed as: ; in, Let be the axial force of the i-th component; The elastic modulus of the material; Let be the bending moment of the i-th component; Let be the moment of inertia of the section of the i-th component.

[0065] The maximum vertical displacement must meet the deflection limit requirement, expressed as: ; in, This refers to the span of the roof.

[0066] The stress ratio is the ratio of the actual stress in a structural member to the material's yield strength. Through structural strain energy analysis, the internal force distribution of each member can be obtained, and thus the stress ratio of each member can be calculated. Strength constraints ensure that the structure does not fail under load. For example, strength constraints are expressed as follows: ; in, The actual stress of component i; The yield strength of the material; The stress ratio is given.

[0067] Displacement constraints are used to ensure that the structure has sufficient stiffness. For example, displacement constraints are expressed as: ; in, This represents the maximum vertical displacement; This is the preset deflection limit.

[0068] In this embodiment of the invention, the building's external surface area, structural steel consumption, and structural strain energy are set as three independent optimization objects, each with the objective of minimizing these factors. However, these three objectives may conflict in actual execution. Therefore, the role of the multi-objective optimization function is to unify these three conflicting objectives under a single evaluation framework, enabling the optimization algorithm to weigh the trade-offs among them and find the optimal solution set.

[0069] For example, the multi-objective optimization function is expressed as: ; in, The external surface area of ​​the building in the proposed building scheme.

[0070] In the above scheme, the carbon emission control target and the structural performance target are unified in the same multi-objective optimization function. At the same time, three constraints of volume, strength and displacement are introduced to ensure the feasibility and safety of the scheme, so that the optimization algorithm can actively find the optimal solution that is low-carbon, safe and functional during the search process.

[0071] As a preferred implementation, step S4 involves performing global iterative optimization on the first and second variables according to the multi-objective optimization function to generate a building scheme, which is then executed through steps S41-S45. S41. Generate an initial population based on the first variable and the second variable; the building morphology parameters of each individual in the initial population are randomly generated based on the first variable, and the roof structure type of each individual is generated based on the building morphology parameters and the second variable. S42. Before the preset genetic iteration stopping condition is reached, calculate the fitness of each individual in the current population according to the multi-objective optimization function. S43. Based on the fitness, through non-dominated sorting, select mother individuals and father individuals from the current population, and cross the mother individuals and father individuals to generate offspring individuals; S44. Mutate the offspring individuals, and update the current population based on the mutated offspring individuals using an elite strategy; S45. When the preset genetic iteration stopping condition is met, the building scheme is obtained based on the optimal non-dominated solution set of the current population.

[0072] In this embodiment of the invention, the first variable is a continuous variable, and the second variable is a discrete variable. The genetic algorithm can handle optimization problems with mixed variable types using a unified encoding method. The genetic algorithm is a type of stochastic search optimization algorithm that simulates natural selection and genetic mechanisms. In the application scenario of this invention, the design variable space is huge, and the genetic algorithm can perform a good global search.

[0073] Of course, in addition to genetic algorithms, algorithms such as particle swarm optimization and simulated annealing can also perform global iterative optimization of variables. In practical applications, the specific optimization algorithm can be selected according to the equipment performance.

[0074] Furthermore, considering the accuracy of the scheme, the improved non-dominated sorting genetic algorithm (NSGA-II) is preferentially selected in the embodiments of the present invention.

[0075] The initial population is the starting point for the iterative optimization of the genetic algorithm. The population consists of several individuals, each representing a complete building scheme to be optimized. In this embodiment of the invention, each building scheme to be optimized must include the selection of building morphology parameters and roof structure type.

[0076] It should be noted that, in the embodiments of the present invention, the roof structure type of individuals in the initial population is not selected completely randomly, but is selected based on preliminary judgment of building morphology parameters. This can improve the quality of the initial population and avoid the overall fitness of the initial population being too low.

[0077] Preferably, in order to preserve the population's ability to explore various structural types, a structural type matching mode based on building morphology parameters is adopted for most individuals, while a random structural type matching mode is adopted for a small number of individuals.

[0078] In the Improved Non-Dominated Sorting Genetic Algorithm (NSGA-II), before reaching the preset stopping condition for genetic iteration, each iteration requires fitness evaluation for every individual in the current population. The fitness evaluation method involves calling a predefined multi-objective optimization function for each individual representing the proposed building scheme. Individuals that satisfy all constraints and perform well on each optimization objective have higher fitness. The fitness calculation results serve as the basis for subsequent selection operations.

[0079] In some preferred embodiments, the constraints in the multi-objective optimization function are handled by a penalty function: ; in, To optimize the function value of object i; The preset weighting coefficients for constraint j.

[0080] Non-dominated ranking is the core mechanism in NSGA-II used to evaluate the quality of individuals. Its basic logic is to stratify all individuals in the population according to Pareto dominance. Individual A is considered to dominate individual B if it is no worse than individual B on all optimization objectives and is better than individual B on at least one objective. The individuals not dominated by any other individual constitute the optimal non-dominated solution set (Pareto front solution set). Then, a new non-dominated layer is selected from the remaining individuals as the second layer, and so on. Within the same non-dominated layer, crowding (the density of individuals in the objective space) also needs to be calculated to maintain population diversity.

[0081] Based on the results of the non-dominated ordination, individuals with high fitness are selected from the current population as mother and father individuals. Then, a crossover operation is performed on the selected mother and father individuals, exchanging or combining some design variables to generate one or more offspring individuals. This crossover operation simulates gene recombination in biological inheritance, allowing offspring individuals to inherit the desirable traits of both parents.

[0082] In some preferred embodiments, an adaptive crossover operator is used to adaptively adjust the crossover probability as the iteration process progresses. The adaptive crossover operator is expressed as: ; in, The crossover probability; The maximum crossover probability is preset. The minimum crossover probability is preset. This represents the current iteration number; This is the preset maximum number of iterations.

[0083] For continuous variables, mutation adds a random perturbation to the vicinity of the current value; for discrete variables, mutation switches to another type with a certain probability. Mutation can introduce new genetic diversity and prevent the algorithm from getting trapped in local optima too early.

[0084] In some preferred embodiments, the variable asynchrony length is adjusted by an adaptive mutation operator, causing the variable asynchrony length to adaptively decrease with the degree of convergence. The adaptive mutation operator is expressed as: ; in, The variable-asynchronous length of the current iteration; The initial variable asynchronous length; This is the preset final variable length.

[0085] The elite strategy is a crucial mechanism for ensuring algorithm convergence. It ensures that optimal solutions from previous generations are not lost during iteration by directly retaining the highest-fitting individuals from the current population into the next generation. Specifically, mutated offspring individuals are merged with elite individuals from the current population to form a new generation. If the size of the new generation exceeds a preset population size, individuals with the highest fitness are retained as the new generation, sorted by fitness.

[0086] When the preset genetic iteration stopping condition is met, all non-dominated solutions in the current population constitute the optimal non-dominated solution set, also known as the Pareto front. Each individual in this solution set represents a building scheme that achieves equilibrium among the three optimization objectives and cannot be dominated by other individuals. Depending on the actual project requirements, one or more schemes can be selected from the Pareto front as the final output building scheme.

[0087] In some preferred embodiments, an objective weighted multi-attribute decision-making method based on information entropy is used to comprehensively rank the optimal non-dominated solution set of the current population. Objective weights are calculated based on the differences in information distribution of each target in the solution set, so that targets with stronger distinguishing ability receive higher weights. Then, the relative proximity of each non-dominated solution to the positive and negative ideal solutions is calculated to obtain at least one final building scheme.

[0088] In the above scheme, the first and second variables are jointly iteratively optimized within the same genetic algorithm framework, achieving parameterized coordination of morphology and structure. Multiple optimization objectives are handled through non-dominated sorting, automatically outputting the Pareto front solution set, avoiding the inefficiency of subjective trade-offs in traditional methods. The combination of elitist strategy and mutation operation ensures both the convergence of the algorithm and the diversity of the population, enabling efficient searching of the globally optimal solution set within a vast design space.

[0089] Further, preferably, step S41, generating an initial population based on the first variable and the second variable, includes: Based on the first variable, several building form parameters are randomly generated; Based on the architectural morphology parameters and the second variable, select the roof structure type; Generate a roof geometric model based on the building morphology parameters and the corresponding roof structure type; Based on the roof geometry model, the maximum vertical displacement and the stress ratio of each component are obtained; Calculate the support position based on the maximum vertical displacement, and add supports in the roof geometry model based on the support position; Based on the stress ratio, the component's cross-section is optimized in the roof geometry model by cross-section scaling or material replacement. Based on the aforementioned architectural morphology parameters and the corresponding roof geometric structure model, several individuals are generated to form an initial population.

[0090] Once the building morphology parameters and roof structure type are determined, the geometric topology of the structure can be automatically generated. For example, when a double-layer reticulated shell structure is selected, an isoparametric surface mapping method is used to generate the roof geometric model, as shown below: ; in, The parametric equations for the roof surface; The thickness of the double-layered reticulated shell; It is the surface normal vector.

[0091] After the roof geometric model is generated, its structural performance can be calculated through finite element analysis to obtain the maximum vertical displacement of all nodes, i.e., the maximum vertical displacement, which is used to evaluate whether the rigidity of the structure meets the code requirements; and the stress ratio of the components is obtained to evaluate whether the strength of the components meets the code requirements.

[0092] In this embodiment of the invention, based on the maximum vertical displacement, areas with large spans can be automatically identified, and the optimal location for adding intermediate supports can be calculated. The calculation criterion for the support location is to select the location that minimizes the maximum vertical displacement of the structure as the support addition location. After adding supports, the overall stiffness of the structure increases, and the maximum vertical displacement decreases.

[0093] As a preferred embodiment, the support position is represented as follows: ; in, For the support position is Time Vertical displacement of each node.

[0094] In addition to structural intervention strategies involving the addition of supports, preferred embodiments of the present invention also provide structural intervention strategies for cross-section optimization to improve the rationality of material distribution.

[0095] For components with excessively high stress ratios (e.g., greater than 0.9), it indicates that the component's load-bearing capacity is insufficient, and the cross-section needs to be enlarged to reduce stress. For components with excessively low stress ratios (e.g., less than 0.3), it indicates that the material utilization rate of the component is too low, resulting in waste. The cross-section can be reduced or the material replaced to reduce the amount of steel used.

[0096] After cross-section optimization, the material distribution of the structure is more reasonable, and the amount of steel used is reduced while still meeting the requirements for strength and stiffness.

[0097] In the above scheme, by embedding structural intervention strategies in the selection process of roof structure type, it is possible to minimize steel consumption while ensuring structural safety, and improve the efficiency and quality of optimization solution.

[0098] Preferably, step S42, before the preset genetic iteration stopping condition is reached, calculates the fitness of each individual in the current population according to the multi-objective optimization function, including: Before the preset genetic iteration stopping condition is met and the number of generations is not greater than the preset generation threshold, the first function value is obtained by performing finite element analysis on the individuals in the current population and calculating the multi-objective optimization function of each individual. Add the individuals in the current population and their corresponding first function values ​​to the training sample database; A Kriging surrogate model is constructed for each optimization objective to predict the multi-objective optimization function for each objective; the hyperparameter updates of the Kriging surrogate model respond to the updates of the training sample database. Before reaching the preset genetic iteration stopping condition and after the number of generations is greater than the preset generation threshold, the multi-objective optimization function of each individual in the current population is predicted through the surrogate model to obtain the second function value and its uncertainty. Based on the second function value and the uncertainty, calculate the expected improvement of aggregation for each individual in the current population; If the expected improvement of the aggregate for the individual is not less than the expected adaptive improvement threshold, then the second function value of the individual is updated through finite element analysis, and the second function value is added to the training sample database; the expected adaptive improvement threshold is calculated based on the number of iterations. Calculate the relative error of the second function value before and after the update. If the relative error is greater than a preset accuracy threshold, perform finite element analysis on all individuals in the current population and update the surrogate model based on the finite element analysis results. The fitness of each individual in the current population is obtained based on the first function value or the second function value.

[0099] Traditional genetic algorithms, when performing multi-objective optimization, require finite element analysis of each individual in the population. Finite element analysis of large-span structures takes a long time per calculation. With a typical population size of 100 and 200 generations, approximately 20,000 finite element analyses are required, with a total computation time of tens of hours or even days, which severely restricts the efficiency of optimization methods in practical design processes.

[0100] During the optimization startup phase (pre-) (For the surrogate model, since it has not yet been established or the training samples are insufficient, in order to ensure the accuracy of the evaluation, Karamba3D is directly called to perform finite element analysis for each individual in the current population to obtain a true evaluation.)

[0101] In this embodiment of the invention, a Kriging surrogate model based on Gaussian process regression is introduced to assist the optimization strategy by constructing a high-precision surrogate mapping from design variables to the objective function. Specifically, an independent Kriging surrogate model is constructed for each optimization objective, with the design variables as input and the predicted value of the optimization objective as output.

[0102] The Kriging surrogate model assumes that the objective function can be decomposed into the sum of a global trend term and a local random bias term. In some preferred embodiments, the optimization objective... The Kriging agent model is represented as: ; in, This is the global trend function (this invention uses a constant trend, i.e., ordinary Kriging). For a Gaussian random process with zero mean, its covariance function is defined as: ; in, For process variance, This is the spatial correlation function.

[0103] In this embodiment of the invention, the spatial correlation function is a Gaussian correlation function, expressed as: in, To design variable dimensions, hyperparameters Solve by maximizing the log-likelihood function: in, The correlation matrix is ​​based on the training samples. This is the maximum likelihood estimate of the process variance.

[0104] Therefore, in order for the Kriging proxy model to learn the mapping relationship between design variables and optimization objectives, and thus achieve rapid function value prediction, the results obtained through finite element analysis will be used to... Design variables for individuals With the corresponding first function value Stored in the training sample database The Kriging proxy model was trained using a training sample database.

[0105] Preferably, the training sample database Represented as: in, The preset algebraic threshold; Population size.

[0106] Once the number of iterations exceeds a preset threshold, the training sample database has reached a certain size, and the surrogate model has achieved good prediction accuracy. At this point, the accelerated optimization phase begins. During the optimization iteration process, Kriging surrogate model predictions replace most finite element analysis calls, and only high-value candidate solutions are verified using real finite element methods. This significantly reduces the total computational load while maintaining optimization accuracy.

[0107] exist In each iteration of the genetic algorithm after the first generation, for all newly generated individuals in the population, the function value of each optimization objective is first quickly predicted using the Kriging surrogate model to obtain the second function value and its uncertainty.

[0108] In this embodiment of the invention, the second function value and its uncertainty are obtained by using the sum and predicted standard deviation of the individual values ​​through a Kriging surrogate model. The predicted mean provides the best estimate of the objective function, while the predicted standard deviation quantifies the degree of uncertainty in the prediction.

[0109] In some preferred embodiments, after training, the Kriging surrogate model applies design variables. The predicted mean With prediction variance They are respectively: ; ; in, To design the correlation vector between variables and training samples; The target value vector for the training samples.

[0110] Expected Improvement (EI) is a surrogate-based collection function used to assess the improvement potential of an individual.

[0111] For a single objective, the expected improvement is expressed as: ; in, For the currently known number The optimal value for each objective. and These are the cumulative distribution function and probability density function of the standard normal distribution, respectively.

[0112] Furthermore, in this embodiment of the invention, for multi-objective scenarios, the expected improvement based on Pareto dominance is adopted, expressed as: ; in, The normalized weighting coefficients for each objective are... This refers to the current population.

[0113] It should be noted that, in this embodiment of the invention, the evaluation criterion for high-value candidate solutions is based on an adaptively improved expectation threshold. For example, the adaptively improved expectation threshold is expressed as: ; in, The decay exponent is set to 1.0-2.0. As iterations proceed, the adaptive improvement expectation threshold gradually decreases, allowing more individuals to receive realistic evaluations in later stages, thus ensuring the accuracy of the final Pareto front.

[0114] like If the individual is considered to have high optimization potential or large prediction uncertainty, it is sent to Karamba3D to perform real finite element analysis to obtain accurate function values, and the second function value is updated based on these function values.

[0115] like If the individual is deemed to have low potential for improvement, the predicted second function value from the Kriging surrogate model is directly used as the target function value for subsequent genetic operations.

[0116] In an embodiment of the present invention, whenever a new real finite element analysis result is generated, it is added to the training database, and the hyperparameters of the Kriging surrogate model are retrained.

[0117] Preferably, to balance training accuracy and computational efficiency, when the number of training samples exceeds the upper limit... At that time, a distance-based sample elimination strategy is adopted to remove samples that are incompatible with new samples. The nearest old sample in the design space is represented as: ; This is to ensure that the training database maintains a good uniformity of distribution within the design space.

[0118] Simultaneously, in each iteration, for the individual that underwent a true finite element analysis, the relative error between the surrogate model's predicted value (the second function value before the update) and the actual value (the second function value after the update) is calculated. For example, the relative error is expressed as: in, For iteration The number of individuals performing real finite element analysis in the data; For an individual that has undergone a real finite element analysis, the single objective function value for the optimization objective k is given.

[0119] If the average relative error Greater than the preset accuracy threshold (Usually taken as 0.05-0.10), then the true evaluation of the entire population is triggered, the proxy model is forced to be updated, and the optimization direction is not misled by the proxy error.

[0120] By employing the aforementioned proxy model-assisted strategy, in a typical 100 population × 200 generations optimization scenario, the number of actual finite element analysis calls can be reduced from approximately 20,000 to approximately 3,000-5,000, improving computational efficiency by approximately 4-6 times, while keeping the Pareto front mass loss (measured by the hypervolume index HV) below 5%. This invention significantly reduces computational load while maintaining optimization accuracy.

[0121] Preferably, step S44 involves mutating the offspring individuals and updating the current population using an elite strategy based on the mutated offspring individuals, including: Extract the morphological features of each individual in the current population, and divide the current population into several morphological sub-regions based on the morphological features; the morphological features include the roof sag-to-span ratio, the roof plan aspect ratio, and the roof curvature variation coefficient; Every preset probability update cycle, the number of selected roof structure types in each morphological sub-region of the current non-dominated solution set is counted to obtain the Pareto contribution statistics. Based on the Pareto contribution statistics, calculate the adaptive selection probability of each roof structure type in each of the morphological sub-regions; Based on the adaptive selection probability, the roof structure type of the offspring individuals is mutated; The mutated offspring, mother, and father individuals are merged, and an elite strategy is adopted to update the current population.

[0122] Traditional genetic algorithms employ a random selection strategy with equal probability for discrete optimization variables. However, under specific building morphology conditions (such as roofs with high span-to-rise ratios), certain structural types (such as arch trusses) have inherent structural efficiency advantages, while others (such as planar trusses) are clearly unsuitable. The equal-probability search strategy leads to a significant waste of computational resources on unsuitable structural types, significantly reducing the search convergence speed.

[0123] In this embodiment of the invention, an adaptive structure type selection probability adjustment mechanism based on morphology-structure fit evaluation is introduced into the mutation process of the genetic algorithm. According to the performance of each structure type in the current non-dominated solution set (Pareto front) within different morphological parameter ranges, the selection probability of the structure type is dynamically adjusted so that the search resources are automatically tilted towards more competitive structure types, thereby improving the convergence efficiency of cross-structure type optimization.

[0124] In some preferred embodiments, an equal-frequency binning strategy is used to divide the current population into J morphological sub-regions. Ensure that each morphological sub-region contains approximately the same number of individuals: ; in, These are the quantile boundary values ​​for the corresponding parameters.

[0125] Every Generation (preset probability update period, usually 10-20 generations), statistically analyze each morphological sub-region in the current non-dominated solution set. The number of selected roof structure types was used to obtain the Pareto statistical contribution value: in, For the first The structure type number of each non-dominated solution. It corresponds to three types of roof structure.

[0126] Within a certain morphological sub-region, the higher the Pareto selection percentage of a certain roof structure type, the greater its probability of being selected in subsequent searches should be. Furthermore, in this embodiment of the invention, based on Pareto contribution statistics, the adaptive selection probability of each roof structure type within each morphological sub-region is calculated, while Laplace smoothing is introduced to prevent the probability from degenerating to zero. ; in, For the smoothing parameter (taken in the embodiment of the invention) This ensures that each structure type always retains the lowest exploration probability, avoiding search omissions caused by premature convergence.

[0127] In some preferred embodiments, to prevent the adaptive probability from converging too quickly, the present invention introduces a temperature decay mechanism to smoothly adjust the distribution of the adaptive selection probability. In the initial stage, the temperature is higher and the probability distribution tends to be more uniform, which encourages exploration; in the later stage, the temperature decreases and the probability distribution tends to be more concentrated, which enhances utilization.

[0128] The adaptive selection probability distribution after temperature decay is expressed as: ; Among them, temperature parameter The reduction gradually occurs as the iteration progresses: ; The initial temperature (taken as 2.0-5.0). The termination temperature is set at 0.5-1.0. In the initial stage, the temperature is higher, and the probability distribution tends to be more uniform, encouraging exploration; in the later stage, the temperature decreases, and the probability distribution tends to be more concentrated, emphasizing utilization.

[0129] Preferably, when the number of non-dominated solutions in a certain morphological subregion is insufficient ( The present invention takes When this occurs, the morphological sub-region is maintained with equal probability allocation. This avoids probability estimation bias caused by small samples.

[0130] In this embodiment of the invention, the mutation operation of the genetic algorithm no longer uses equal-probability random selection for the mutation of discrete variables, but instead uses the morphological sub-region where the current individual is located. According to adaptive probability Perform probability sampling: ; This strategy increases the probability of selecting structural types that have demonstrated a competitive advantage under specific morphological conditions, guiding search resources toward more promising design spaces.

[0131] After adaptive mutation, a new generation of offspring individuals is obtained. To ensure that the optimal solutions from previous generations are not lost, an elitist strategy is used to update the population. The mutated offspring individuals, the selected mother individuals from the previous generation, and the parent individuals are merged to form a temporary population. Then, all individuals in this temporary population are sorted by fitness and their crowding is calculated. Finally, based on the preset population size, individuals are selected to form a new generation of the population, and the remaining individuals are eliminated.

[0132] In the above scheme, the adaptive probability adjustment mechanism based on morphology-structure adaptability enables the genetic algorithm to dynamically adjust the selection probability of each structure type in different morphological subregions according to the feedback of the current Pareto front during the search process, thereby reducing the waste of resources caused by searching on unsuitable structure types.

[0133] The architectural scheme generation method based on morphological and structural collaboration provided by the present invention can support automatic comparison of multiple structural types and simultaneously optimize architectural morphological parameters and roof structure types within the same framework. This breaks the traditional design process that separates architecture and structure, and realizes parametric collaboration between morphology and structure.

[0134] This invention provides a building scheme generation system based on morphological and structural collaboration. Please refer to [link / reference]. Figure 2 The architectural scheme generation system based on morphological structure collaboration includes a first variable construction module 11, a second variable construction module 12, an optimization function definition module 13, and an architectural scheme generation module 14, wherein: The first variable construction module 11 is used to calculate the building form parameters and their value range based on the building functional requirements and site constraints, and to construct the first variable. The second variable construction module 12 is used to construct a second variable based on the preset roof structure type and its corresponding geometric parameters; the roof structure type includes at least a double-layer reticulated shell structure, a space truss beam structure and an arch truss combination structure. The optimization function definition module 13 is used to establish a multi-objective optimization function; the optimization objectives of the multi-objective optimization function include minimizing the building's external surface area, minimizing the amount of steel used in the structure, and minimizing the structural strain energy. The building scheme generation module 14 is used to perform global iterative optimization on the first variable and the second variable according to the multi-objective optimization function to generate a building scheme.

[0135] In a preferred embodiment, the first variable construction module 11 is used for: Generate the grandstand cross-sectional geometry and roof outline based on the building's functional requirements and site constraints. Based on the grandstand cross-sectional geometry and the roof outline, the building form parameters and their value ranges are obtained; the building form parameters include roof height control parameters and roof outline boundary control parameters; The architectural morphology parameters and their value ranges are encapsulated as the first variable.

[0136] Further, preferably, generating the grandstand cross-sectional geometry and roof profile based on building functional requirements and site constraints includes: Obtain building seating parameters and building site core parameters to obtain building functional requirements; the building seating parameters include the number of seats, seat type, seat row spacing and line of sight slope; Obtain the building height limit, roof covering method, site boundary, setback requirements and the location of surrounding road interfaces to obtain the site constraints; Based on the building seating parameters and the site constraints, calculate the elevation of each row of seats and generate the grandstand profile geometry. Based on the grandstand profile geometry, the building core parameters, and the site constraints, a roof profile covering the core area and the seating area is generated.

[0137] In a preferred embodiment, the second variable construction module 12 is specifically used for: When the roof structure type is selected as a double-layer reticulated shell structure, the grid density, the distance between the upper and lower layers, and the node connection method are obtained as geometric structural parameters. When the roof structure type is selected as a spatial truss beam structure, the main truss spacing, main truss height, and secondary truss spacing are obtained as geometric structural parameters. When the roof structure type is selected as an arch truss combination structure, the main arch rise, main arch spacing, main arch landing height, and stabilizing truss type are obtained as geometric structural parameters; the stabilizing truss type includes horizontal support and diagonal support. Each roof structure type and its corresponding geometric parameters are encapsulated as a second variable.

[0138] In a preferred embodiment, the optimization function definition module 13 is specifically used for: Geometric analysis is used to calculate the building's exterior area and spatial volume of the proposed building scheme. Based on the stated spatial volume and the preset minimum spatial volume, establish volume constraints. The structural performance parameters of the building scheme to be optimized are calculated using finite element analysis; the structural performance parameters include the amount of steel used in the structure, the structural strain energy, and the maximum vertical displacement. Based on the strain energy of the structure, the stress ratio is calculated, and strength constraint conditions based on the stress ratio are established; Based on the maximum vertical displacement and the preset deflection limit, establish displacement constraint conditions; By combining the volume constraint, the strength constraint, and the displacement constraint, the constraint conditions are obtained; The building's external surface area, the amount of steel used in the structure, and the structural strain energy are each taken as optimization objects. With the goal of minimizing each of the optimization objects, and in combination with the constraints, a multi-objective optimization function is established for the building scheme to be optimized.

[0139] In a preferred embodiment, the building scheme generation module 14 includes: An initial population generation unit is used to generate an initial population based on the first variable and the second variable; the building morphology parameters of each individual in the initial population are randomly generated based on the first variable, and the roof structure type of each individual is generated based on the building morphology parameters and the second variable; The fitness calculation unit is used to calculate the fitness of each individual in the current population according to the multi-objective optimization function before the preset genetic iteration stopping condition is reached. The selection and crossover operation unit is used to select mother individuals and father individuals from the current population according to the fitness and through non-dominated sorting, and crossover the mother individuals and the father individuals to generate offspring individuals; The mutation operation unit is used to mutate the offspring individuals and update the current population based on the mutated offspring individuals using an elite strategy. The building scheme generation unit is used to obtain a building scheme based on the optimal non-dominated solution set of the current population when the preset genetic iteration stopping condition is met.

[0140] Further, preferably, the initial population generation unit is specifically used for: Based on the first variable, several building form parameters are randomly generated; Based on the architectural morphology parameters and the second variable, select the roof structure type; Generate a roof geometric model based on the building morphology parameters and the corresponding roof structure type; Based on the roof geometry model, the maximum vertical displacement and the stress ratio of each component are obtained; Calculate the support position based on the maximum vertical displacement, and add supports in the roof geometry model based on the support position; Based on the stress ratio, the component's cross-section is optimized in the roof geometry model by cross-section scaling or material replacement. Based on the aforementioned architectural morphology parameters and the corresponding roof geometric structure model, several individuals are generated to form an initial population.

[0141] Preferably, the fitness calculation unit is specifically used for: Before the preset genetic iteration stopping condition is met and the number of generations is not greater than the preset generation threshold, the first function value is obtained by performing finite element analysis on the individuals in the current population and calculating the multi-objective optimization function of each individual. Add the individuals in the current population and their corresponding first function values ​​to the training sample database; A Kriging surrogate model is constructed for each optimization objective to predict the multi-objective optimization function for each objective; the hyperparameter updates of the Kriging surrogate model respond to the updates of the training sample database. Before reaching the preset genetic iteration stopping condition and after the number of generations is greater than the preset generation threshold, the multi-objective optimization function of each individual in the current population is predicted through the surrogate model to obtain the second function value and its uncertainty. Based on the second function value and the uncertainty, calculate the expected improvement of aggregation for each individual in the current population; If the expected improvement of the aggregate for the individual is not less than the expected adaptive improvement threshold, then the second function value of the individual is updated through finite element analysis, and the second function value is added to the training sample database; the expected adaptive improvement threshold is calculated based on the number of iterations. Calculate the relative error of the second function value before and after the update. If the relative error is greater than a preset accuracy threshold, perform finite element analysis on all individuals in the current population and update the surrogate model based on the finite element analysis results. The fitness of each individual in the current population is obtained based on the first function value or the second function value.

[0142] Preferably, the mutation operation unit is specifically used for: Extract the morphological features of each individual in the current population, and divide the current population into several morphological sub-regions based on the morphological features; the morphological features include the roof sag-to-span ratio, the roof plan aspect ratio, and the roof curvature variation coefficient; Every preset probability update cycle, the number of selected roof structure types in each morphological sub-region of the current non-dominated solution set is counted to obtain the Pareto contribution statistics. Based on the Pareto contribution statistics, calculate the adaptive selection probability of each roof structure type in each of the morphological sub-regions; Based on the adaptive selection probability, the roof structure type of the offspring individuals is mutated; The mutated offspring, mother, and father individuals are merged, and an elite strategy is adopted to update the current population.

[0143] The architectural scheme generation system based on morphological and structural collaboration provided by the present invention can support automatic comparison of multiple structural types and simultaneously optimize architectural morphological parameters and roof structure types within the same framework. This breaks the traditional design process that separates architecture and structure, and realizes parametric collaboration between morphology and structure.

[0144] This invention also provides a building scheme generation device based on morphological structure collaboration, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements a building scheme generation method based on morphological structure collaboration as described above. The working principles and beneficial effects of the two are one-to-one, so they will not be described in detail here.

[0145] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0146] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A method for generating architectural schemes based on morphological and structural collaboration, characterized in that, include: Based on the building's functional requirements and site constraints, the building's morphological parameters and their value ranges are calculated to construct the first variable; A second variable is constructed based on the preset roof structure type and its corresponding geometric parameters; The roof structure types include at least double-layer reticulated shell structure, space truss beam structure and arch truss combination structure; A multi-objective optimization function is established; the optimization objectives of the multi-objective optimization function include minimizing the building's external surface area, minimizing the amount of steel used in the structure, and minimizing the structural strain energy. Based on the multi-objective optimization function, a global iterative optimization is performed on the first variable and the second variable to generate a building scheme.

2. The architectural scheme generation method based on morphological structure collaboration as described in claim 1, characterized in that, The process of calculating building form parameters and their value ranges based on building functional requirements and site constraints to construct the first variable includes: Generate the grandstand cross-sectional geometry and roof outline based on the building's functional requirements and site constraints. Based on the grandstand cross-sectional geometry and the roof outline, the building form parameters and their value ranges are obtained; the building form parameters include roof height control parameters and roof outline boundary control parameters; The architectural morphology parameters and their value ranges are encapsulated as the first variable.

3. The architectural scheme generation method based on morphological structure collaboration as described in claim 2, characterized in that, The process of generating the grandstand cross-sectional geometry and roof outline based on building functional requirements and site constraints includes: Obtain building seating parameters and building site core parameters to obtain building functional requirements; the building seating parameters include the number of seats, seat type, seat row spacing and line of sight slope; Obtain the building height limit, roof covering method, site boundary, setback requirements and the location of surrounding road interfaces to obtain the site constraints; Based on the building seating parameters and the site constraints, calculate the elevation of each row of seats and generate the grandstand profile geometry. Based on the grandstand profile geometry, the building core parameters, and the site constraints, a roof profile covering the core area and the seating area is generated.

4. The architectural scheme generation method based on morphological structure collaboration as described in claim 1, characterized in that, The step of constructing a second variable based on a preset roof structure type and its corresponding geometric parameters includes: When the roof structure type is selected as a double-layer reticulated shell structure, the grid density, the distance between the upper and lower layers, and the node connection method are obtained as geometric structural parameters. When the roof structure type is selected as a spatial truss beam structure, the main truss spacing, main truss height, and secondary truss spacing are obtained as geometric structural parameters. When the roof structure type is selected as an arch truss combination structure, the main arch rise, main arch spacing, main arch landing height, and stabilizing truss type are obtained as geometric structural parameters; the stabilizing truss type includes horizontal support and diagonal support. Each roof structure type and its corresponding geometric parameters are encapsulated as a second variable.

5. The architectural scheme generation method based on morphological structure collaboration as described in claim 1, characterized in that, The establishment of the multi-objective optimization function includes: Geometric analysis is used to calculate the building's exterior area and spatial volume of the proposed building scheme. Based on the stated spatial volume and the preset minimum spatial volume, establish volume constraints. The structural performance parameters of the building scheme to be optimized are calculated using finite element analysis; the structural performance parameters include the amount of steel used in the structure, the structural strain energy, and the maximum vertical displacement. Based on the strain energy of the structure, the stress ratio is calculated, and strength constraint conditions based on the stress ratio are established; Based on the maximum vertical displacement and the preset deflection limit, establish displacement constraint conditions; By combining the volume constraint, the strength constraint, and the displacement constraint, the constraint conditions are obtained; The building's external surface area, the amount of steel used in the structure, and the structural strain energy are each taken as optimization objects. With the goal of minimizing each of the optimization objects, and in combination with the constraints, a multi-objective optimization function is established for the building scheme to be optimized.

6. The architectural scheme generation method based on morphological structure collaboration as described in claim 1, characterized in that, The step of generating a building scheme by performing global iterative optimization on the first and second variables according to the multi-objective optimization function includes: An initial population is generated based on the first variable and the second variable; the building morphology parameters of each individual in the initial population are randomly generated based on the first variable, and the roof structure type of each individual is generated based on the building morphology parameters and the second variable. Before the preset genetic iteration stopping condition is reached, the fitness of each individual in the current population is calculated according to the multi-objective optimization function. Based on the fitness, through non-dominated sorting, female and male individuals are selected from the current population, and the female and male individuals are cross-crossed to generate offspring individuals; The offspring individuals are mutated, and based on the mutated offspring individuals, an elite strategy is adopted to update the current population; When the preset genetic iteration stopping condition is met, the building scheme is obtained based on the optimal non-dominated solution set of the current population.

7. The architectural scheme generation method based on morphological structure collaboration as described in claim 6, characterized in that, The step of generating an initial population based on the first variable and the second variable includes: Based on the first variable, several building form parameters are randomly generated; Based on the architectural morphology parameters and the second variable, select the roof structure type; Generate a roof geometric model based on the building morphology parameters and the corresponding roof structure type; Based on the roof geometry model, the maximum vertical displacement and the stress ratio of each component are obtained; Calculate the support position based on the maximum vertical displacement, and add supports in the roof geometry model based on the support position; Based on the stress ratio, the component's cross-section is optimized in the roof geometry model by cross-section scaling or material replacement. Based on the aforementioned architectural morphology parameters and the corresponding roof geometric structure model, several individuals are generated to form an initial population.

8. The architectural scheme generation method based on morphological structure collaboration as described in claim 6, characterized in that, Before the preset genetic iteration stopping condition is reached, the fitness of each individual in the current population is calculated according to the multi-objective optimization function, including: Before the preset genetic iteration stopping condition is met and the number of generations is not greater than the preset generation threshold, the first function value is obtained by performing finite element analysis on the individuals in the current population and calculating the multi-objective optimization function of each individual. Add the individuals in the current population and their corresponding first function values ​​to the training sample database; A Kriging surrogate model is constructed for each optimization objective to predict the multi-objective optimization function for each objective; the hyperparameter updates of the Kriging surrogate model respond to the updates of the training sample database. Before reaching the preset genetic iteration stopping condition and after the number of generations is greater than the preset generation threshold, the multi-objective optimization function of each individual in the current population is predicted through the surrogate model to obtain the second function value and its uncertainty. Based on the second function value and the uncertainty, calculate the expected improvement of aggregation for each individual in the current population; If the expected improvement of the aggregate for the individual is not less than the expected adaptive improvement threshold, then the second function value of the individual is updated through finite element analysis, and the second function value is added to the training sample database; the expected adaptive improvement threshold is calculated based on the number of iterations. Calculate the relative error of the second function value before and after the update. If the relative error is greater than a preset accuracy threshold, perform finite element analysis on all individuals in the current population and update the surrogate model based on the finite element analysis results. The fitness of each individual in the current population is obtained based on the first function value or the second function value.

9. The architectural scheme generation method based on morphological structure collaboration as described in claim 6, characterized in that, The process of mutating the offspring individuals and updating the current population using an elite strategy based on the mutated offspring individuals includes: Extract the morphological features of each individual in the current population, and divide the current population into several morphological sub-regions based on the morphological features; the morphological features include the roof sag-to-span ratio, the roof plan aspect ratio, and the roof curvature variation coefficient; Every preset probability update cycle, the number of selected roof structure types in each morphological sub-region of the current non-dominated solution set is counted to obtain the Pareto contribution statistics. Based on the Pareto contribution statistics, calculate the adaptive selection probability of each roof structure type in each of the morphological sub-regions; Based on the adaptive selection probability, the roof structure type of the offspring individuals is mutated; The mutated offspring, mother, and father individuals are merged, and an elite strategy is adopted to update the current population.

10. A method for generating architectural schemes based on morphological and structural collaboration, characterized in that, include: The first variable construction module is used to calculate the building form parameters and their value range based on the building's functional requirements and site constraints, and to construct the first variable. The second variable construction module is used to construct a second variable based on the preset roof structure type and its corresponding geometric parameters; the roof structure type includes at least a double-layer reticulated shell structure, a space truss beam structure and an arch truss combination structure. The optimization function definition module is used to establish a multi-objective optimization function; the optimization objectives of the multi-objective optimization function include minimizing the building's external surface area, minimizing the amount of structural steel used, and minimizing the structural strain energy. The building scheme generation module is used to perform global iterative optimization on the first variable and the second variable according to the multi-objective optimization function to generate building schemes.

Images