Multi-objective form generation and optimization method for deepwater platform based on geometric proxy model

Abstract

The application discloses a kind of deep sea platform multi-objective form generation and optimization method based on geometric proxy model, belong to the cross technical field of ocean engineering and computer aided design.This application constructs the parametric voxel growth model based on the interference field of offshore wind and wave environment, aims to establish the digital generation logic of platform entity structure, simulates the modular assembly process;Based on the fast mapping relationship of three-dimensional voxel form set G to physical performance index set P, construct geometric proxy construction process and final total proxy function;Based on NSGA-II algorithm, construct multi-objective optimization problem, aims to find the Pareto optimal solution set in variable space;Finally, according to the specific environmental weight of target sea area, automatically output the final three-dimensional form model data and module assembly sequence.This application realizes the fast intelligent generation of floating platform form by constructing the mapping relationship of physical performance and geometric characteristics, using geometric calculation instead of expensive physical field simulation.

Description

TECHNICAL FIELD

[0001] The application belongs to the cross technical field of ocean engineering and computer-aided design, and specifically relates to a form automatic generation and multi-physical field performance rapid optimization method for a deep-sea floating farming complex in a conceptual design stage. BACKGROUND

[0002] Deep-sea industry has become an inevitable trend of marine industry development. The "multi-use floating platform" integrating deep-sea farming and renewable energy capture (such as offshore photovoltaic and wind energy) is a key equipment to solve the shortage of sea resources and realize energy self-sufficiency.

[0003] At present, the design of such large and complex floating structures mainly has the following technical bottlenecks:

[0004] 1. Low design efficiency: The traditional design process adopts a linear mode of "manual experience modeling-engineering verification". Each adjustment of the platform form requires high-fidelity computational fluid dynamics (CFD) and finite element analysis (FEA). A single calculation often takes several hours to several days, resulting in the inability to quickly explore a large number of form schemes in the early conceptual design stage.

[0005] 2. Difficulty in coupling multiple physical fields: Existing design methods often separate various specialties. Structural engineers focus on stability, energy engineers focus on power generation, and there is a lack of a unified generation mechanism that can simultaneously consider hydrodynamic stability, structural lightweighting, and energy acquisition efficiency. (Existing designs often focus on a single target (such as only considering hydrodynamic stability), ignoring the coupling relationship between energy acquisition efficiency and form, making it difficult to obtain a comprehensive optimal solution that considers "stability" and "energy efficiency").

[0006] 3. Lack of rapid evaluation means for discrete modules: With the development of modular construction technology, platform structures based on voxel assembly are increasingly attracting attention. However, for such stepped non-streamlined surfaces, there is a lack of a rapid performance proxy evaluation model based on geometric features, which cannot complete performance prediction for thousands of combination schemes in seconds. SUMMARY

[0007] The purpose of the present application is to provide a deep-sea platform multi-objective form generation and optimization method based on a geometric proxy model to overcome the shortcomings of the prior art.

[0008] To achieve the above purpose, the present application adopts the following technical solution:

[0009] A deep-sea platform multi-objective form generation and optimization method based on a geometric proxy model includes the following specific steps:

[0010] S1: Construct a parameterized voxel growth model based on the sea wave environment interference field, aiming to establish the digital generation logic of the platform entity structure, simulate the modular assembly process;

[0011] S2: Construct a fast performance proxy model based on geometric features (Geometric Surrogate Models); based on the fast mapping relationship between the three-dimensional voxel morphology set G and the physical performance index set P, construct the geometric proxy construction process and the final total proxy function;

[0012] S3: Multi-objective genetic algorithm iterative optimization: based on the NSGA-II algorithm to construct a multi-objective optimization problem, aiming to find the Pareto optimal solution set in the variable space;

[0013] S4: Optimal morphology matching and output; according to the specific environmental weight of the target sea area, automatically output the final three-dimensional morphology model data and module assembly sequence.

[0014] Further, the S1 specifically includes:

[0015] S1-1: Set the design domain: discretize the three-dimensional virtual sea space into a plurality of N × N × M three-dimensional voxel grids (Voxel Grid), and the topological dimension of the grid is , the center of gravity of each grid is ; Set the discretized grid set as ; , the geometric center coordinates of the th potential voxel unit in the set are , where is the traversal index, ;

[0016] S1-2: Set the standard functional module: set the standardized floating module unit, which includes the upper energy component (photovoltaic support), the middle buoyancy component (buoyancy tank), and the lower functional component (net cage interface);

[0017] S1-3: Construct the interference field generation logic: set at least one morphology control source (Attractor) in the design domain, which is one or more virtual geometric entities (such as points, spheres, or curves), used to construct a scalar field to drive the birth and death of voxels; corresponding to the functional core area of the platform (such as the feeding operation center, the energy convergence center, or the traffic stop point). Set the morphology control source as , whose spatial position is defined by the coordinate vector , , and its range of action is defined by the interference radius ; based on the definition, execute the generation algorithm.

[0018] Furthermore, the generation algorithm includes:

[0019] (1) Distance calculation: Traverse the set of points and calculate each potential point using the Euclidean distance formula. To the morphological control source center distance : ;

[0020] (2) Topology screening: setting growth threshold (usually related to radius) (Related); Establish Boolean judgment rules:

[0021] like Then mark the position. The status is "Active", and an entity functional module is generated here.

[0022] like Then mark the position. The state is "Inactive", and this area remains an empty water area.

[0023] (3) Morphological evolution: By continuously changing the coordinates of the control source and threshold This can drive the platform's overall topology to undergo a continuous evolution from discrete to aggregated.

[0024] Furthermore, S2 includes:

[0025] S2-1: Constructing a Hydrodynamic Drag Proxy Model: Setting the Design Sea Level Height and dominant wave direction vector Extract all located in The following modular geometry is constructed using... Given the projection plane of the normal, calculate the total area of ​​the orthographic projection of the complex underwater geometry onto this plane. :

[0026] The ;

[0027] This sub-item's target function:

[0028] To minimize the underwater projected area in order to reduce shape drag;

[0029] S2-2: Constructing a geometric stability proxy model: Establish a global coordinate system with the positive Z-axis pointing upwards; design a fixed draft / still water surface. The voxel density is a constant ρ, therefore the voxel mass is proportional to the volume. In stability proxy calculations, volume weighting can be used instead of mass weighting; morphological sets From voxel units Composition, physical fitness coordinates are The volume is The geometric centroids of all generated modules are iterated in real time, and the volume-weighted geometric center of all voxel units in the morphological set G is calculated and defined as the centroid. Under the assumption of uniform density, the centroid height of the shape set G is defined as:

[0030] ;

[0031] The following voxel section defines the set of immersed voxels:

[0032] ;

[0033] The volume-weighted geometric center of the submerged portion is defined as the center of buoyancy. The buoyancy height of the shape set G is defined as:

[0034] ;

[0035] Constructor function: Let For a stable surrogate mapping function, then: ;

[0036] The optimization objective is to use the stability surrogate as a sub-objective function in a multi-objective optimization: ;

[0037] S2-3: Constructing an Energy Efficiency Proxy Model: Extract the top surface mesh of all generated modules and use a ray-tracing algorithm to calculate the cumulative solar radiation value of each mesh surface under preset meteorological data. ;set up For energy efficiency proxy mapping function, Let the unit area emissivity be 1, then: = (G) = ;

[0038] The optimization objective is to maximize Then the objective function of this sub-item is: ;

[0039] S2-4: Constructing a Structural Cost Proxy Model: Traversing the Parametrically Generated Set of Forms , the total number of activated (i.e. state is True) voxel units is counted, denoted as ; according to the preset standardized voxel unit size (edge length ), the volume of a single module is calculated ; the overall entity volume of the platform is calculated , as a direct proxy indicator of building material consumption and carbon emissions; a function is constructed: let be the cost proxy mapping function, then: ;

[0040] Optimization goal: minimize or , the sub-objective function: .

[0041] Further, in the S2-1, the construction process includes:

[0042] (1) Geometric extraction: traverse all voxel units in the morphology set G, and filter out the underwater unit set with a centroid Z coordinate less than the sea level height ; ;

[0043] (2) Boolean fusion: perform Boolean union operation (Boolean Union) on the set to generate a single underwater entity model to eliminate internal overlapping surfaces.

[0044] (3) Projection calculation: set the dominant flow direction vector , and construct the projection plane perpendicular to . Calculate the orthographic projection contour of the entity on the plane π, and obtain the area of the contour.

[0045] (4) Function construction: let be the resistance proxy mapping function, then:

[0046] ;

[0047] The sub-objective function:

[0048] , to minimize the underwater projection area to reduce the form drag.

[0049] Further, in the S2-3, the function construction process is:

[0050] (1) Light-receiving surface extraction: traverse the morphology set G, extract the photovoltaic module surface at the top of each voxel unit, and form a light-receiving surface set ;

[0051] (2) Occlusion and radiation calculation: For each face in the set , calculate its effective light receiving area or radiation receiving amount based on ray tracing method (or simplified area method); finally, add up the performance values of all effective faces;

[0052] (3) Construction function: Let be the energy efficiency agent mapping function, be the radiation coefficient per unit area (or set to 1), then: = (G) = ;

[0053] (4) The optimization goal is to maximize , then the sub-objective function is: .

[0054] Further, in S2, based on the four sub-objective functions, a total vector function for performance evaluation of deep-sea aquaculture platform is constructed :

[0055] ;

[0056] This function maps any generated geometric shape to a point in a four-dimensional performance space , realizing the rapid transformation from "shape domain" to "high-dimensional performance domain".

[0057] Further, S3 includes:

[0058] S3-1: Define decision variables: Define gene coding vector :

[0059] ;

[0060] where is the center coordinate of the interference source; R is the action radius of the interference field; T is the growth screening threshold. The shape generation process is represented as mapping

[0061] S3-2: Objective function construction: Establish a multi-objective function group:

[0062] ;

[0063] Or written in the form of unified minimization (vector form):

[0064] ;

[0065] Where the constraint condition is: ​

[0066] ;

[0067] S3-3: Evolutionary computation: Non-dominated Sorting Genetic Algorithm II (NSGA-II) with elitist strategy is adopted to iteratively search the total proxy model. The population size and iteration number are set, and the non-dominated solution is screened in the solution space.

[0068] S3-4: Result output: The Pareto Front is output, that is, a set of optimal shapes achieving different trade-offs between resistance, stability and energy efficiency.

[0069] Compared with the prior art, the present application has at least the following beneficial effects:

[0070] 1. Extremely high design and evaluation efficiency: The present application innovatively proposes a "geometric proxy model" for voxelized floating structure, reducing the time of single fluid and stability evaluation from hours to milliseconds, making it possible to perform thousands of shape iterations in the conceptual design phase.

[0071] 2. Realize "shape-energy-stability" multi-field coupling: Overcome the problem of separation of structure, hydrodynamic and energy design in traditional design, through a unified algorithm framework, automatically generate optimal shapes that meet the requirements of floating stability and maximize the use of ocean space for energy capture.

[0072] 3. Solve the problem of overall design of discrete modules: For modular assembly offshore buildings, an intelligent generation logic is provided from "unordered blocks" to "ordered functional bodies", which has strong engineering guiding significance. BRIEF DESCRIPTION OF DRAWINGS

[0073] Figure 1 A schematic diagram is generated for the parametric voxel growth space of the present application.

[0074] Figure 2 A schematic diagram of the geometric performance proxy model principle of the present application.

[0075] Figure 3 A schematic diagram of the multi-objective optimization result based on NSGA-II algorithm of the present application.

[0076] Figure 4 A flowchart of the deep sea farming platform shape adaptive generation method of the present application. DETAILED DESCRIPTION

[0077] For the purposes of the present disclosure, the technical solutions and advantages are more clearly and obviously understood, the present disclosure is further described in detail below with reference to specific examples and with reference to the drawings. Obviously, the described examples are part of the embodiments of the present disclosure, rather than all the embodiments. Based on the embodiments in the present disclosure, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present disclosure.

[0078] Embodiment 1:

[0079] A deep sea platform multi-objective form generation and optimization method based on a geometric proxy model, as shown in Figure 4 , includes the following specific steps:

[0080] S1: Construct a parameterized voxel growth model based on the sea wave environment interference field, aiming to establish the digital generation logic of the platform entity structure, and simulate the modular assembly process;

[0081] S1-1: Construction of 3D Voxel Grid: as shown in Figure 1 , the figure shows a discrete three-dimensional lattice (voxel grid), and a cluster of functional modules generated by aggregation under the influence of a spherical interference source. Set a three-dimensional virtual sea area design space, which is discretized into a regular lattice set composed of potential unit center points. This lattice is mathematically represented as an ordered list of three-dimensional coordinates, representing all possible "potential positions" for placing entity functional modules. Set the topological dimension of the lattice to . Among them:

[0082] N (Horizontal Count): represents the number of voxel units (integer) divided in the horizontal plane X axis and Y axis direction of the design domain. For example, if the design domain length and width are both 100m, and the edge length of a single module is 10m, then .

[0083] M (Vertical Count): represents the number of voxel units (integer) divided in the vertical Z axis direction of the design domain. For example, if the total of the draft depth and the height above water is 50m, and the height of a single module is 10m, then .

[0084] Variable : Let the discretized lattice set be . , represents the geometric center coordinates of the th potential voxel unit in the set , where is the traversal index, .

[0085] S1-2: Set standard functional module: Set the standard floating module unit, which contains the upper energy component (photovoltaic support), the middle buoyancy component (buoyancy tank) and the lower functional component (net cage interface);

[0086] S1-3: Construct interference field generation logic:

[0087] At least one attractor is set in the design domain, which is one or more virtual geometric entities (such as points, spheres or curves) for constructing a scalar field to drive the birth and death of voxels; corresponding to the functional core area of the platform (such as the feeding operation center, the energy convergence center or the traffic stop point).

[0088] Let the attractor be , whose spatial position is defined by the coordinate vector , , and its action range is defined by the interference radius . Based on the above definition, the following generation algorithm is executed:

[0089] Distance calculation: traverse the point array set, and calculate the distance from each potential point to the center of the attractor : ;

[0090] Topology screening: set the growth threshold , which is usually related to the radius . Establish the Boolean judgment rule:

[0091] If , mark the position as "active (Active)", and generate an entity functional module at this position.

[0092] If , mark the position as "inactive (Inactive)", and keep it as a blank water area.

[0093] Morphological evolution: by continuously changing the control source coordinates and the threshold , the overall topology of the platform can be driven to continuously evolve from discrete to aggregated.

[0094] S2: Constructing geometric surrogate models for fast performance proxy; fast mapping relationship from the three-dimensional voxel morphology set G to the physical performance index set P, define three independent geometric proxy construction processes and the final total proxy function. As shown in Figure 2 , the principle diagram of performance proxy model based on geometric characteristics. A shows the principle of using underwater projection area to proxy hydrodynamic resistance; B shows the principle of using the height difference between the center of buoyancy and the center of gravity to proxy geometric stability.

[0095] S2-2: Hydrodynamic drag proxy model:

[0096] Technical principle: For non-streamlined voxel stacking structure, its shape resistance is positively correlated with the effective projection area of the flow surface. The projection area calculation is particularly suitable for discrete stepped surface (Discrete Stepped Surface), which extracts the edge contour through Boolean operation, avoiding the non-convergence problem of traditional grid method for non-streamlined objects.

[0097] Implementation method: Set the design sea level height and the dominant wave direction vector . Extract all modules of geometric bodies located below. Construct the projection plane with as the normal vector, calculate the total projection area of the underwater complex geometric body on the plane .

[0098] Optimization goal: Minimize (corresponding to minimizing fluid resistance).

[0099] Construction process:

[0100] (1) Geometric extraction: traverse all voxel units in the morphology set G, and select the underwater unit set with the center of mass Z coordinate less than the sea level height .

[0101] (2) Boolean fusion: perform Boolean union operation (Boolean Union) on the set to generate a single underwater entity model to eliminate internal overlapping surfaces.

[0102] (3) Projection calculation: set the dominant flow direction vector , construct the projection plane perpendicular to . Calculate the normal projection contour of the entity on the plane π, and get the area of the contour.

[0103] (4) Construction function: Let be the stability proxy mapping function, then:

[0104] ;

[0105] Sub-objective function: , to minimize the underwater projection area to reduce the form drag.

[0106] S2-2: Geometric stability proxy model (Stability Proxy):

[0107] Technical principle: Based on the principle of hydrostatics of floating bodies, the relative position of the center of buoyancy (Center of Buoyancy, ) and the center of gravity (Center of Gravity, ) determines the initial stability.

[0108] Implementation method: Real-time traversal of all generated geometric centroids of the module, based on the volume-weighted method to calculate the overall center of gravity ( ); extract the water surface line The following modules, based on the volume-weighted method to calculate the center of buoyancy ( ).

[0109] Optimization objective: Maximize the vertical distance = - (pursue the center of buoyancy higher than the center of gravity, or the center of gravity as low as possible).

[0110] Construction process:

[0111] (1) Whole floating body calculation: Calculate the volume-weighted geometric center of all voxel elements in the shape set G, defined as the center of gravity , whose vertical height is . The following voxel part, calculate its volume-weighted geometric center, defined as the center of buoyancy , whose vertical height is .

[0112] (2) Height difference calculation: Calculate the algebraic difference of the center of buoyancy and the center of gravity in the Z-axis direction.

[0113] (3) Construction function: Let be the stability proxy mapping function, then: ;

[0114] (4) Sub-objective function: Maximize the initial stability height, pursue "bottom heavy head light" or deep draft state.

[0115] S2-3: Energy capture efficiency proxy model (Energy Efficiency Proxy):

[0116] Technical principle: Based on solar geometry, calculate the effective radiation received by the top surface of the module.

[0117] Implementation method: Extract the top surface grid of all generated modules, and use the ray-tracing algorithm to calculate the cumulative solar radiation value of each grid surface under the preset meteorological data .

[0118] Optimization goal: Maximize .

[0119] Construction process:

[0120] (1) Light receiving surface extraction: traverse the shape set G, extract the photovoltaic component surface on the top of each voxel unit, and form the light receiving surface set .

[0121] (2) Occlusion removal and radiation calculation: for each face in the set , calculate its effective light receiving area or radiation received based on the ray-tracing method (or simplified area method).

[0122] (3) Total sum: add up the efficiency values of all effective faces.

[0123] (4) Construction function: let be the energy efficiency proxy mapping function, be the radiation coefficient per unit area (or set to 1), then: = (G) = ;

[0124] Maximize the effective photovoltaic laying area or total radiation received, and the project objective function is: ;

[0125] S2-4: Structural lightweighting and cost proxy model (Structural Cost Proxy):

[0126] Technical principle: Under the premise of meeting functional requirements, the total amount of modules of the platform directly corresponds to the construction cost and carbon emissions.

[0127] Implementation method: Count the total number of entity modules generated in the current morphological scheme or the total volume .

[0128] Optimization objective: Minimize or .

[0129] Construction process:

[0130] Entity statistics: Traverse the parameterized generated morphology set , and count the total number of activated (i.e., state is True) voxel units, denoted as .

[0131] Volume calculation: According to the preset standardized voxel unit size (edge length ), calculate the volume of a single module .

[0132] Total sum: Calculate the overall entity volume of the platform , as a direct proxy indicator of construction material consumption and carbon emissions.

[0133] Construction function: Let be the cost proxy mapping function, then: ;

[0134] Under the premise of meeting other functional constraints, minimize material usage to reduce construction cost, the objective function is:

[0135] .

[0136] This objective is to form a Pareto Front (Pareto Front) constraint relationship. Because usually, you want to "generate more electricity" (large area) will lead to "high resistance" or "high cost", there is a trade-off relationship between the four objectives, which is the necessity of using multi-objective genetic algorithm (NSGA-II).

[0137] Total Surrogate Function (Total Surrogate Function), based on the above four sub-items, constructs a multi-physical field performance evaluation total vector function of deep sea aquaculture platform :

[0138] ;

[0139] This function maps any generated geometric morphology to a point in a four-dimensional performance space , achieving a rapid transformation from "morphology domain" to "high-dimensional performance domain".

[0140] S3: Multi-objective genetic algorithm iterative optimization; based on NSGA-II algorithm to construct multi-objective optimization problem, aiming to find the Pareto optimal solution set in the variable space.

[0141] S3-1: Define decision variables: Define the gene encoding vector :

[0142] ;

[0143] wherein is the interference source center coordinate; R is the interference field action radius; T is the growth screening threshold. The morphological generation process is represented as a mapping

[0144] S3-2: Objective function construction: Establish a multi-objective function group:

[0145] ;

[0146] Or written in a unified minimization form (vector form):

[0147] ;

[0148] wherein the constraint condition is:

[0149] ;

[0150] S3-3: Evolutionary calculation: Use the non-dominated sorting genetic algorithm (NSGA-II) with an elite strategy to iteratively search the total agent model. Set the population size and iteration number, and select the non-dominated solution in the solution space.

[0151] S3-4: Result output: Output the Pareto optimal front (Pareto Front), that is, a set of optimal morphologies that achieve different trade-offs between resistance, stability, and energy efficiency.

[0152] S4: Optimal morphology matching and output; According to the specific environmental weight of the target sea area (for example, high sea state sea area preferentially matches the minimum resistance scheme, low sea state sea area preferentially matches the highest capacity scheme), automatically output the final three-dimensional morphological model data and module assembly sequence.

[0153] Example 2

[0154] This example is based on Example 1, and a comprehensive platform for intensive aquaculture and photovoltaic power generation is designed for a deep sea area (water depth > 50 m).

[0155] 1. Initialization setting:

[0156] Set the design domain to be a space of 100m × 100m × 50m, and the size of a single voxel module to be 10m × 10m × 10m. Set two mobile control sources to simulate the layout of dual-core functions.

[0157] 2. Proxy indicator calculation process:

[0158] (1) Resistance calculation: Set the dominant wave direction as south. The program automatically extracts the modules below the waterline and projects them vertically onto the east-west plane. The area of the resulting projected profile is taken as the resistance proxy value.

[0159] (2) Stability calculation: The program iterates through the centroid coordinates of all generated modules in real time, and calculates the overall barycenter by weighted calculation. Similarly, the geometric center of the underwater displacement volume is calculated as the center of buoyancy. The difference between the Z-axis coordinates of the two is calculated.

[0160] (3) Energy efficiency calculation: Import the EPW meteorological file of the target sea area, and calculate the annual effective radiation energy of the top surface of all modules.

[0161] 3. Optimization process: Set the population size to 50, the iteration number to 50, and generate 2500 potential morphological schemes.

[0162] 4. Result analysis:

[0163] As shown in Figure 3 , the red point set in the figure forms the Pareto optimal frontier, and the system selects three schemes for the designer to decide, and the optimization results show a significant Pareto distribution characteristic:

[0164] Scheme A (stability dominant type): The morphological characteristics are wide at the bottom and deep in the water, showing an inverted pyramid shape, with the largest distance between the center of buoyancy and the center of gravity, suitable for extreme sea conditions.

[0165] Scheme B (energy dominant type): The morphological characteristics are flat and extend to all directions, maximizing the light receiving area, but the projected resistance is slightly larger, suitable for calm sea areas.

[0166] Scheme C (comprehensive compromise type): This shape ensures 80% of the maximum energy output while reducing the projected resistance by 30% through local streamline processing, making it the recommended scheme for this embodiment.

[0167] The above specific embodiments further illustrate the purpose, technical solutions and beneficial effects of the present disclosure. It should be understood that the above description is only a specific embodiment of the present disclosure and is not intended to limit the present disclosure. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present disclosure should be included in the protection scope of the present disclosure.

Claims

1. A method for generating and optimizing multi-objective configurations of a deep offshore platform based on a geometric proxy model, characterized in that, Comprise the following steps: S1: Construct a parameterized voxel growth model based on the sea wave environment interference field, aiming to establish the digital generation logic of the platform entity structure, simulate the modular assembly process; S2: Construct a fast performance proxy model based on geometric features; based on the fast mapping relationship between the three-dimensional voxel shape set G and the physical performance index set P, construct a geometric proxy model and the final total proxy function; the S2 specifically comprises: S2-1: Constructing the hydrodynamic resistance proxy model: setting the design sea level height and the dominant wave direction vector , extracting the module geometry located below, constructing the projection plane with as the normal line, calculating the total area of the normal projection of the underwater complex geometry on the plane : The sub-target function: to minimize underwater projected area to reduce form drag; S2-2: Construct a geometric stability proxy model: Establish a global coordinate system, with the positive direction of the Z axis upward; design the waterline / still water surface to be fixed as ; the voxel density is consistent as a constant p; the shape set is composed of voxel units , the voxel centroid coordinates are , and the volume is ; real-time traverse the geometric centroid of all generated modules, calculate the volume-weighted geometric center of all voxel units in the shape set G, defined as the barycenter , under the condition of consistent density, the barycenter height of the shape set G is defined as: The following voxel section defines the set of immersed voxels: The volume-weighted geometric center of the submerged portion is defined as the center of buoyancy The center of buoyancy height of the shape set G is defined as: ； Constructing function: Let For the stability of the proxy mapping function, then: The optimization objective is to maximize the vertical separation = - The sub-objective function is: ; S2-3: Build the energy capture efficiency proxy model: Extract the top surface grid of all generating modules, and calculate the cumulative solar radiation value of each grid surface under the preset meteorological data using the ray tracing algorithm ; Set as the energy efficiency proxy mapping function, as the radiation coefficient per unit area, then: = (G) = The optimization objective is to maximize The sub-objective function is then: S2-4: Constructing structure lightweight and cost agent model: traversing the morphological set generated by parameterization , the total number of activated voxel units is counted, denoted as ; according to the preset standardized voxel unit size, the volume of a single module is calculated; the overall entity volume of the platform is calculated as a direct proxy indicator of building material consumption and carbon emissions; construct the function: let be the cost agent mapping function, then: Optimization goal: Minimize or The sub-goal function: S3: Multi-objective genetic algorithm iterative optimization: based on the NSGA-II algorithm to construct a multi-objective optimization problem, aiming to find the Pareto optimal solution set in the variable space; S4: Optimal shape matching and output; according to the specific environmental weight of the target sea area, automatically output the final three-dimensional shape model data and the module assembly sequence.

2. The deep offshore platform multi-objective form generation and optimization method of claim 1, wherein, The S1 specifically comprises: S1-1: Set the design domain: Set a three-dimensional virtual sea space, which is discretized into a plurality of N × N × M three-dimensional voxel point arrays, and the topological dimension of the point array is , the center of gravity of each point array is , and the set of discretized point arrays is ; , the geometric center coordinates of the th potential voxel unit in the set are , wherein is a traversal index, ; S1-2: Set the standard functional module: set the standardized floating module unit; S1-3: Constructing interference field generation logic: setting at least one morphological control source in the design domain, which is one or more virtual geometric entities, corresponding to the functional core area of the platform; setting the morphological control source as , the spatial position of which is defined by the coordinate vector , , the action range of which is defined by the interference radius ; and executing the generation algorithm.

3. The deep offshore platform multi-objective configuration generation and optimization method of claim 2, wherein, The generation algorithm comprises: (1) Distance calculation: traverse the point set collection, and calculate the distance of each potential point to the morphological control source center by using the Euclidean distance formula to the morphological control source center : : ; (2) Topology screening: Set a growth threshold , and establish a Boolean judgment rule: If , the state of the marker position is "Active", and an entity function module is generated at this point. If , the status of the marker position is "Inhibit Inactive", which is left blank here; (3) Morphological evolution: By continuously changing the control source coordinates and threshold values the overall topology of the platform can be driven to continuously evolve from discrete to aggregated.

4. The deep offshore platform multi-objective form generation and optimization method of claim 1, wherein, In the S2-1, the construction of the hydrodynamic resistance proxy model comprises: (1) Geometry extraction: traverse all the voxel units in the shape set G, and screen out the underwater unit set with the centroid Z coordinate less than the sea level height ;​ (2) Boolean fusion: union of sets performing a Boolean union operation to generate a single underwater solid model to eliminate internal overlapping faces; (3) Projection calculation: Set the dominant flow direction vector , construct the projection plane perpendicular to ; Calculate the positive projection profile of the entity on the plane π, and obtain the area of the profile; (4) Construct the function: Set For the resistance proxy mapping function, then: The sub-objective function: is to minimize the underwater projected area to reduce the form drag.

5. The deep offshore platform multi-objective form generation and optimization method of claim 1, wherein, The function construction process in the S2-3 is: (1) Light-receiving surface extraction: traverse the shape set G, extract the photovoltaic module surface on the top of each voxel unit, form a light-receiving surface set ; (2) Occlusion removal and radiation calculation: For the set Each face in Calculate its effective light-receiving area or radiation received amount; finally, sum the efficiency values ​​of all effective surfaces; construct the function: Let For energy efficiency proxy mapping function, Let be the radiation coefficient per unit area, then: = (G) = The optimization objective is to maximize The sub-objective function is then: .

6. The deep offshore platform multi-objective form generation and optimization method of claim 1, wherein, In the S2, the multi-physical field performance evaluation total agent function of the deep sea culture platform is constructed : This function maps any generated geometry into a point in the four-dimensional performance space .

7. The deep offshore platform multi-objective configuration generation and optimization method of claim 6, wherein, The S3 comprises: S3-1 : Defining decision variables: Defining the gene encoding vector : wherein is the interference source center coordinate; R is the interference field action radius; T is the growth selection threshold; the morphogenesis process is represented as a mapping S3-2: Objective function construction: establish a multi-objective function group: ； Or is a unified minimization form: Wherein, the constraint condition is: S3-3: Evolutionary computation: using non-dominated sorting genetic algorithm with elitism to solve the total agent function iterative search; set the population size and iteration number, and screen the non-dominated solution in the solution space; S3-4: Result output: output the Pareto optimal front, that is, a series of optimal shape sets that achieve different trade-offs between resistance, stability and energy efficiency.

Images