A tower section scale synchronous similar design method and system

Abstract

This invention belongs to the field of structural design and simulation testing technology for renewable energy equipment in the marine sector. It discloses a synchronous similarity design method and system for scaled-down towers. This method introduces an innovative composite penalty fitness function to drive a genetic algorithm for automatic optimization, thereby completing the segmented parameter design of the scaled-down tower in a single, high-precision manner. It synchronously ensures a high degree of similarity between the model and the actual model in key dynamic characteristics (first-order natural frequency) and accurate reproduction in three-dimensional geometry, while strictly complying with all engineering manufacturing constraints. This invention significantly improves optimization efficiency, achieves excellent design accuracy and completeness, and has strong engineering applicability.

Description

Technical Field

[0001] This invention belongs to, but is not limited to, the technical field of structural design and simulation testing of renewable energy equipment in the marine field, and particularly relates to a tower scaling synchronous similarity design method and system based on a composite penalty fitness genetic algorithm. Background Technology

[0002] The tower is the core supporting structure of a wind turbine generator, and its dynamic characteristics directly affect the operational stability and safety of the unit. Due to the large size of the actual tower, the high cost of testing, and the susceptibility of field testing to environmental factors, scaled-down model testing has become a key technical means to study the dynamic characteristics of the tower.

[0003] Currently, scaled-down tower design faces two major technical bottlenecks:

[0004] (1) The contradiction between dynamic similarity and design efficiency: Traditional methods often use enumeration to adjust the cross-sectional parameters to meet the scaling requirements of the first natural frequency. This process is inefficient and has poor optimization ability. It is also difficult to meet the frequency requirements while taking into account multiple constraints such as mass distribution and structural strength. This may lead to deviations between the model and the real model in terms of multi-order dynamic characteristics and structural stress distribution, making it difficult to fully reflect the actual working state of the real tower.

[0005] (2) Conflict between geometric similarity and dynamic similarity: Current technical approaches are often contradictory. One approach only optimizes the cross-section to satisfy first-order frequency similarity, but this leads to geometric distortion, affecting aerodynamic load simulation and higher-order modal characteristics; another approach strictly maintains geometric scaling, but due to stiffness and mass distribution distortion, the dynamic characteristics deviate significantly from the similarity criteria. Existing methods lack the ability to simultaneously achieve automated design for dynamic and geometric similarity within a unified framework.

[0006] Therefore, there is an urgent need for a design method that can automatically, efficiently, and accurately search for the parameters of a scaled-down tower model that simultaneously meets complex constraints such as dynamic performance (frequency), mass quota, geometric feasibility, and shape fidelity.

[0007] Based on the above analysis, the urgent technical problems that need to be solved in the existing technology are:

[0008] (1) Inefficient design and difficult to find global optimization: Traditional methods (such as enumeration) rely on manual trial and error to adjust the cross-sectional parameters. The process is time-consuming and prone to getting stuck in local optima. The optimization process is separated from the verification of multiple constraints (quality, geometry), resulting in a large number of invalid iterations, which seriously restricts the design cycle and model reliability.

[0009] (2) It is difficult to achieve both dynamic similarity and geometric similarity: There is a contradiction in the existing technical approach: focusing on frequency similarity will lead to distortion of geometric shape; while strict geometric scaling will cause distortion of mass and stiffness distribution, causing dynamic characteristics (especially frequency) to deviate seriously from the target. There is a lack of effective means to achieve both similarities in a unified model simultaneously.

[0010] (3) Weak ability to process multiple constraints: Traditional optimization processes are difficult to systematically process multiple and coupled engineering constraints such as frequency, quality, minimum wall thickness, diameter-to-thickness ratio, and processing feasibility. They often require repeated compromises and manual intervention, resulting in insufficient comprehensive feasibility and robustness of the design results. Summary of the Invention

[0011] To address the problems existing in the prior art, this invention provides a method and system for synchronous similar design of tower scaling based on a composite penalty fitness genetic algorithm.

[0012] This invention is implemented as follows: a method for synchronous similarity design of tower scaling based on a composite penalty fitness genetic algorithm, specifically including:

[0013] S1: Quantification of design goals and constraints;

[0014] S2: Design a composite penalty fitness function for individual evaluation in genetic algorithms, transforming a complex engineering optimization problem with multiple objectives and constraints into a single-objective optimization problem that can directly drive efficient search by genetic algorithms;

[0015] S3: Encoding and initialization: Encode the changes in the inner and outer diameters of each segment of the tower into chromosomes, and randomly generate an initial population within the feasible region;

[0016] S4: Evaluation, decoding each individual model, performing parametric modeling and finite element analysis, and calculating the first natural frequency of the scaled model. With the total mass of the model Substitute the values ​​into the established composite penalty fitness function to evaluate;

[0017] S5: Evolution, which uses operations such as elite selection, simulated binary crossover (SBX), and polynomial mutation to generate offspring;

[0018] S6: Convergence and Output. When the optimal fitness value is stable for multiple generations or reaches the maximum number of iterations, output the design parameters corresponding to the best individuals in each generation as the final scheme of the core load-bearing structure.

[0019] S7: Geometric Shape Reproduction: Based on the optimized core structure, the external geometric contour of the actual tower is accurately reproduced by applying a custom low-density, low-stiffness foam ring.

[0020] Furthermore, in S1, the core optimization objective is to reduce the first-order natural frequency of the scaled model. Infinitely approaching the target frequency derived from the similarity law The key constraints are as follows:

[0021] Quality constraint: Total model mass Must not exceed target quality The allowable deviation range;

[0022] Geometric and engineering constraints: including minimum wall thickness at segmentation points The logical relationship that the inner diameter of each segment is smaller than the outer diameter, and the processing feasibility range of all dimensional parameters.

[0023] Furthermore, in S2, the fitness function is used for individual evaluation in the genetic algorithm:

[0024]

[0025] In the formula:

[0026] The individual fitness value (minimizing it is the objective);

[0027] The first term is the frequency deviation term, which drives the design to converge toward the target frequency;

[0028] The second item is the quality penalty item, which applies a linear penalty when the quality exceeds the standard. This is the weighting coefficient for quality penalty;

[0029] The third item is the geometric penalty item, which quantifies the penalty for each violation of geometric / engineering constraints;

[0030] and As an adjustable parameter, it enables differentiated weight configuration for different constraints of varying importance.

[0031] Another objective of this invention is to provide a tower scaling synchronous similarity design system based on a composite penalty fitness genetic algorithm, the system specifically comprising:

[0032] The composite penalty fitness function construction module is used to construct a fitness function that combines goal-driven and constraint-penalized approaches.

[0033] The core load-bearing structure optimization module is used to optimize the design of the core tower structure that meets the target frequency and mass constraints using a genetic algorithm based on a composite penalty fitness function and main materials such as aluminum alloy.

[0034] The external geometry coating replication module is used to accurately replicate the external contour of the original tower with the same geometric scale by applying a custom low-density, low-stiffness foam ring.

[0035] Based on the above technical solutions and the technical problems solved, the advantages and positive effects of the technical solution to be protected by this invention are as follows:

[0036] This invention maps multi-objective, multi-constraint problems into a single optimization driver by using a composite penalty fitness function. Innovatively, this invention introduces first-order intrinsic frequency deviation, quality exceedance degree, and geometric and manufacturing constraint violation degree into the same fitness function. By replacing the traditional hard selection rules with a continuous penalty mechanism, the genetic algorithm can automatically weigh the importance of various constraints during the search process.

[0037] This invention embeds dynamic similarity as the dominant optimization direction into the search process itself: by directly using the relative deviation between the first-order natural frequency calculated by the scaled model and the target frequency of the similarity law as the core driving term of the fitness function, the algorithm always revolves around dynamic similarity in the entire evolution process, rather than relying on later frequency tuning or empirical correction.

[0038] This invention incorporates engineering manufacturability constraints into the optimization process from the outset: engineering constraints such as minimum wall thickness, machinable dimensional range, and logical relationship between inner and outer diameters are no longer used as post-verification conditions, but are embedded in fitness evaluation in the form of penalty terms, so that unmanufacturable solutions are naturally eliminated in the early stages of evolution, fundamentally avoiding ineffective design cycles.

[0039] This invention achieves synergistic satisfaction of geometric similarity and dynamic similarity through the functional decoupling design of the core load-bearing structure and the external geometric cladding: For the first time, this invention clearly distinguishes between the dynamic load-bearing function and the shape reproduction function in the scaled tower design, so that the core structure can focus on dynamics and mass optimization, while the external geometric contour is achieved through a low-density, low-stiffness cladding, thus eliminating the mutual interference between the two types of similarity from a mechanism perspective.

[0040] This invention significantly improves the overall efficiency of complex scaled model design: by constructing a composite penalty fitness function, the multi-constraint engineering design process, which originally required repeated manual calculations and verifications, is transformed into a single-objective search problem that can be directly and efficiently handled by a genetic algorithm. Compared with traditional enumeration methods or manual parameter tuning methods, the design cycle is shortened from several weeks to several hours, the optimization efficiency is improved by more than 90%, and the problem of suboptimal solutions caused by local adjustments is avoided.

[0041] This invention achieves high-precision simultaneous fulfillment of dynamic similarity, mass equivalence, and geometric feasibility:

[0042] The results of the examples show that, without the need for subsequent counterweights or frequency tuning corrections, the first-order natural frequency error of the scaled model can be stably controlled within 1.5%, the total mass error is less than 1%, and the structural parameters of each segment meet the engineering processing constraints. This ensures the reliability and repeatability of the dynamic characteristics of the scaled model in terms of both principle and results.

[0043] This invention achieves 100% reproduction of the external geometric contour while maintaining a high degree of consistency with the dynamic core: by replicating the shape of the actual tower through a low-density, low-stiffness cladding structure, the scaled model maintains consistency with the actual model in terms of aerodynamic load distribution, fluid flow characteristics and appearance proportions, while avoiding significant impact on the dynamic characteristics of the core structure, and significantly improving the applicability of the model in various scenarios such as wind tunnel tests and water tank tests.

[0044] This invention presents a plug-and-play scaled-down model design scheme that is engineeringable and scalable: the parameters output by this method can be directly used for manufacturing and have been verified by finite element simulation. It provides a standardized and systematic design path for high-confidence physical tests of wind power equipment and similar tall structures, and has outstanding engineering application value.

[0045] (1) The expected benefits and commercial value of the technical solution of this invention after transformation are as follows:

[0046] The tower scaling synchronous similarity design method proposed in this invention can directly serve the model design needs of wind turbine manufacturers, research institutes, and testing organizations in scenarios such as wind tunnel testing, water tank testing, and modal testing. Compared with traditional methods that rely on experience and manual correction, this invention significantly reduces the design cycle and trial-and-error costs of scaled models, and improves the reliability and repeatability of test results.

[0047] At the commercial level, this method can be industrialized and promoted as a scaled model design software system or engineering service module, with good economic benefits and industrial transformation prospects.

[0048] (2) The technical solution of this invention fills a technical gap in the industry both domestically and internationally:

[0049] Based on the search and analysis, existing domestic and foreign public literature and patents mainly focus on the following two technical paths: one is to adjust local parameters only for dynamic frequency and ignore geometric similarity; the other is to strictly maintain the geometric scaling ratio, but cannot effectively control the deviation of dynamic characteristics.

[0050] No existing technical solution has been found that can automatically achieve simultaneous satisfaction of dynamic similarity, mass constraints, geometric feasibility, and shape reproduction within a unified optimization framework through a composite penalty fitness mechanism. The systematic design method proposed in this invention, combining composite penalty fitness with core structure and shape decoupling, fills the technological gap in automated, high-precision scaled model design in this field.

[0051] (3) The technical solution of the present invention solves a technical problem that people have long wanted to solve but have never been able to solve successfully:

[0052] The design of scaled-down models of towers and similar tall structures has long relied on engineering experience and repeated manual corrections. Particularly in maintaining geometric consistency while satisfying first-order natural frequencies and mass constraints, a repeatable and scalable methodology has been consistently lacking. This invention, through an algorithmic and systematic design mechanism, fundamentally solves this long-standing technical problem in the engineering field, achieving a paradigm shift from experience-driven to algorithm-driven approaches.

[0053] (4) The technical solution of the present invention overcomes technical bias:

[0054] A common technical bias in existing technologies is the belief that dynamic similarity and geometric similarity cannot be simultaneously and precisely satisfied in scaled model design, and can only be achieved through compromise or post-hoc correction. This invention, by structurally separating load-bearing and shape-related functions and employing a composite penalty fitness optimization mechanism, demonstrates that, within a reasonable design framework, both can be achieved synchronously and with high precision during the design phase, thus overcoming long-standing limitations in this field's understanding. Attached Figure Description

[0055] Figure 1 This is a flowchart of the tower scaling synchronous similarity design method using a composite penalty fitness genetic algorithm provided in this embodiment of the invention;

[0056] Figure 2 This is a block diagram of a tower scaling synchronous similarity design system based on a composite penalty fitness genetic algorithm provided in an embodiment of the present invention.

[0057] Figure 3 This is the optimization and iterative process provided in the embodiments of the present invention;

[0058] Figure 4 This is the tower design optimization result provided by the embodiments of the present invention;

[0059] Figure 5 This is a physical model of a stiffness-similar tower manufactured according to the design method of this invention, provided in an embodiment of the invention.

[0060] Figure 6 This is a geometrically similar tower physical model manufactured according to the design method of this invention, provided in an embodiment of the invention. Detailed Implementation

[0061] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0062] In scaled-down physical tests of large wind turbine towers, offshore tension leg structures, and tall cylindrical shell structures, the industry has long faced a structural contradiction:

[0063] On the one hand, scaled-down models need to strictly adhere to the principles of geometric similarity and mass similarity in terms of mass, geometric dimensions, and external contours to ensure the authenticity of external load distribution and boundary conditions in wind tunnel tests, modal tests, or hydrodynamic tests. On the other hand, under engineering constraints such as material mechanical properties, processing wall thickness, and connection methods, the overall stiffness distribution and dynamic characteristics of scaled-down structures often cannot be consistent with the target modal parameters derived by the similarity law, especially the first-order natural frequency is prone to systematic deviation.

[0064] In traditional engineering practice, frequency adjustment is often achieved through empirical correction or local addition or subtraction of counterweights. This approach has the following drawbacks:

[0065] First, counterweight frequency tuning is a post-hoc correction, which makes it difficult to ensure accurate frequency convergence while meeting quality constraints. Second, frequency adjustment and geometric shape reproduction are coupled, and relying solely on material or counterweight correction can easily destroy shape similarity. Third, engineering constraints (minimum wall thickness, processing limits, inter-segment logical relationships) are usually introduced through manual verification, which severely compresses the design space and makes it difficult to guarantee global optimality.

[0066] Therefore, in industrial applications, there is an urgent need for a systematic approach that can simultaneously coordinate dynamic similarity, mass constraints, and geometric realizability in the early stages of design, so that scaled-down models can be repeatable, manufacturable, and physically consistent when used as test subjects.

[0067] This method constructs a collaborative design mechanism linked by a composite penalty fitness, coupling similarity law constraints, engineering constraints, and the genetic algorithm search process within a unified evaluation framework. The collaborative relationship of this mechanism is reflected in the following three levels:

[0068] (1) Similarity law target-driven layer: By directly introducing the relative deviation between the first-order natural frequency of the scaled model and the similarity law target frequency into the fitness function, it is ensured that the optimization process is always dominated by dynamic similarity, rather than post-correction.

[0069] (2) Engineering constraint embedded layer: Quality constraints, geometric logic constraints and processing feasibility constraints are no longer used as independent screening conditions, but are embedded in the fitness function in the form of penalty terms, so that the genetic algorithm can automatically avoid unmanufacturable solutions during the search process.

[0070] (3) Structure-shape decoupling layer: By functionally separating the core load-bearing structure from the external geometric contour, structural optimization is carried out only for dynamic and mass characteristics, while shape similarity is subsequently reproduced through low-density, low-stiffness cladding to avoid mutual interference between the two.

[0071] The three mechanisms mentioned above work together within a unified optimization framework, fundamentally changing the design paradigm of scaled-down models that rely on repeated manual corrections.

[0072] In the initial design phase, the first-order target natural frequency corresponding to the scaled model was derived through similarity theory and used as the dominant optimization objective. Simultaneously, engineering factors such as the allowable deviation of the model's total mass, minimum wall thickness, the relationship between the inner and outer diameters of different segments, and the range of machining dimensions were transformed into quantifiable constraints. The key to this process is avoiding simply viewing constraints as binary judgments of whether they are satisfied; instead, it constructs continuously differentiable or piecewise continuous constraint measures for the subsequent penalty function, providing an effective gradient direction for the genetic algorithm.

[0073] The composite penalty fitness function plays a central role in this method; it is not merely a mathematical construct but a mapping of engineering design intent. The frequency deviation term directly reflects the degree of deviation between the structural dynamic response and similar targets, serving as the main driving force for the search direction. The quality penalty term is activated only when the quality exceeds the allowable range, enabling the algorithm to prioritize finding solutions with reasonable quality while satisfying dynamic similarity. The geometric penalty term cumulatively penalizes each violation of engineering constraints, naturally eliminating unmanufacturable or unreasonable structures at the fitness level.

[0074] By configuring the weighting coefficients, the importance of different constraints can be adjusted according to the specific engineering background, thereby achieving adaptation to actual manufacturing and testing needs.

[0075] The variations in the inner and outer diameters of each tower segment are encoded into the chromosome as continuous design variables. After each individual segment is decoded during the evaluation phase, a parametric structural model is automatically generated and enters the finite element solution process.

[0076] The natural frequencies and quality indices obtained from finite element analysis are directly fed back to the fitness function, achieving closed-loop coupling between the structural response and the optimization algorithm. This process avoids the inefficient cycle of traditional manual modeling, calculation, and correction.

[0077] The elite retention strategy ensures that the current optimal structure is not destroyed, simulated binary crossover is used to maintain the diversity of solutions in the continuous parameter space, and polynomial mutation is used to avoid premature convergence.

[0078] When the fitness value remains stable over multiple generations, it means that under given engineering constraints and similar objectives, the structural parameters have reached the optimal state of coordination between dynamics and mass. At this point, the output design scheme has clear engineering feasibility.

[0079] After obtaining the core load-bearing structure, the external geometry of the actual tower is reproduced by applying a custom-designed low-density, low-stiffness foam ring. This cladding material contributes minimally to the structural dynamics and its influence on modal characteristics can be ignored, thus achieving effective decoupling between shape similarity and dynamic similarity.

[0080] The design strategy ensures that the scaled-down model is highly consistent in terms of wind field effects, fluid flow around it, or appearance scale, while avoiding the reverse interference of shape modification on the performance of the core structure.

[0081] Through the synergy of the above mechanisms, this method has achieved the following technical effects in industrial applications:

[0082] The dynamic similarity of the scaled-down model is inherently guaranteed during the design phase, significantly reducing the workload of frequency tuning and counterweight adjustment in the later stages.

[0083] Structural parameters are globally optimized while meeting manufacturing constraints, improving model consistency and repeatability;

[0084] Decoupling the shape from the load-bearing function improves the adaptability of the scaled-down test model in various test scenarios.

[0085] This method provides an engineering-friendly, scalable, and reproducible system design path for scaled-down testing of tall structures, and has clear engineering application value and industrial promotion prospects.

[0086] like Figure 1 As shown, this embodiment of the invention provides a method for synchronous similarity design of tower scaling based on a composite penalty fitness genetic algorithm. The method specifically includes:

[0087] S1: Quantification of design goals and constraints;

[0088] S2: Design a composite penalty fitness function for individual evaluation in genetic algorithms, transforming a complex engineering optimization problem with multiple objectives and constraints into a single-objective optimization problem that can directly drive efficient search by genetic algorithms;

[0089] S3: Encoding and initialization: Encode the changes in the inner and outer diameters of each segment of the tower into chromosomes, and randomly generate an initial population within the feasible region;

[0090] S4: Evaluation, decoding each individual model, performing parametric modeling and finite element analysis, and calculating the first natural frequency of the scaled model. With the total mass of the model Substitute the values ​​into the established composite penalty fitness function to evaluate;

[0091] S5: Evolution, which uses operations such as elite selection, simulated binary crossover (SBX), and polynomial mutation to generate offspring;

[0092] S6: Convergence and Output. When the optimal fitness value is stable for multiple generations or reaches the maximum number of iterations, output the design parameters corresponding to the best individuals in each generation as the final scheme for the core load-bearing structure (such as an aluminum alloy tower).

[0093] S7: Geometric Shape Reproduction: Based on the optimized core structure, the external geometric contour of the actual tower is accurately reproduced by applying a custom low-density, low-stiffness foam ring, thereby achieving a "decoupling-integration" design between the dynamic core and the geometric shell.

[0094] The core optimization objective of S1 is to reduce the first-order natural frequency of the scaled model. Infinitely approaching the target frequency derived from the similarity law The key constraints are as follows:

[0095] Quality constraint: Total model mass Must not exceed target quality The allowable deviation range.

[0096] Geometric and engineering constraints: including minimum wall thickness at segmentation points The logical relationship that the inner diameter of each segment is smaller than the outer diameter, and the processing feasibility range of all dimensional parameters.

[0097] S2, the fitness function, is used for individual evaluation in the genetic algorithm:

[0098]

[0099] In the formula:

[0100] The individual fitness value (minimizing it is the objective).

[0101] The first term is the frequency deviation term, which drives the design to converge toward the target frequency.

[0102] The second item is the quality penalty item, which applies a linear penalty when the quality exceeds the standard. This is the quality penalty weighting coefficient.

[0103] The third item is the geometric penalty item, which quantifies the penalty for each violation of geometric / engineering constraints, such as insufficient wall thickness. , The geometric penalty weight coefficients are represented by Penalty(·). Penalty(·) denotes the penalty function, g j This represents the j-th constraint term. This indicates the minimum wall thickness at the segmentation point. This indicates the actual wall thickness at the segment.

[0104] and As an adjustable parameter, it enables differentiated weight configuration for different constraints of importance.

[0105] The innovative mechanism of this function lies in its transformation of hard constraints into soft penalties in fitness, enabling the algorithm to explore a wider solution space in the early stages of evolution. Simultaneously, as evolution progresses, it automatically and continuously guides the population towards feasible regions that fully satisfy all constraints. This avoids the ineffective iterations caused by optimizing first and then verifying in traditional methods, achieving a deep integration and dynamic balance between constraint satisfaction and objective optimization during the search process.

[0106] This invention provides a method for synchronous similarity design of tower scaling based on a composite penalty fitness genetic algorithm, comprising the following steps:

[0107] (1) Determine the core elements of the tower scale design. The core elements include the input variables, optimization objectives and constraints of the tower scale design. The input variables are the change rate of the tower inner diameter and outer diameter. The optimization objective is the first natural frequency of the tower. The constraints are the requirements for mass, wall thickness and inner diameter.

[0108] (2) A composite penalty fitness genetic algorithm is used as the optimization method for the tower scaling parameters; the composite penalty fitness genetic algorithm is used to perform the optimization operation of the tower parameters to obtain the optimal scaling tower segment parameters; the optimization operation includes the following steps:

[0109] 1) Assign initial values ​​to the segment parameters of the tower to complete the tower parameter initialization;

[0110] 2) Based on the composite penalty rule, evaluate the error fit between the current parameters and the optimization objective to complete the parameter error fitness evaluation;

[0111] 3) Filter out the segment parameters that meet the preset requirements and complete the selection of the optimal parameters for each segment;

[0112] 4) Perform cross-operation on the selected segment parameters to generate new parameter combinations, completing the cross-operation of parameters for each segment of the tower;

[0113] 5) Perform mutation operations on the parameters after crossover to expand the parameter search range and complete the mutation of parameters for each section of the tower;

[0114] 6) Check whether the fitness of the current parameter meets the iteration termination condition. If it does not meet the condition, return to the sub-step of selecting the optimal parameter for each segment and repeat the iteration. If it meets the condition, proceed to the next step.

[0115] (3) Based on the Froude similarity criterion, a lightweight foam tower structure is designed. The optimal scaled tower segment parameters obtained through the optimization operation are wrapped in the lightweight foam tower structure to meet the requirement of similar structural shape. A scaled tower design scheme that takes into account both the similarity of the first-order natural frequency of the tower and the similarity of the structural shape is obtained.

[0116] like Figure 2 As shown in the figure, the tower scaling synchronous similarity design system based on a composite penalty fitness genetic algorithm provided in this embodiment of the invention specifically includes:

[0117] The composite penalty fitness function construction module is used to construct a fitness function that combines goal-driven and constraint-penalized approaches.

[0118] The core load-bearing structure optimization module is used to optimize the design of the core tower structure that meets the target frequency and mass constraints using a genetic algorithm based on a composite penalty fitness function and main materials such as aluminum alloy.

[0119] The external geometry coating replication module is used to accurately replicate the external contour of the original tower with the same geometric scale by applying a custom low-density, low-stiffness foam ring.

[0120] The following example uses the steel tower of an offshore wind turbine. Considering the available materials in the laboratory and their properties, Al 6063 material was selected. The parameters corresponding to Al 6063 material are a density of 2690. The Young's modulus is 68.3 GPa, and the Poisson's ratio is 0.3. For ease of fabrication, the model tower adopts a five-segment variable cross-section structure. The genetic algorithm is used below to determine the information for each cross-section; the specific operation procedure is detailed below. Figure 1 . Figure 3 This demonstrates the average and optimal fitness of each generation in the genetic algorithm's iterative process. After the 10th generation, the optimal fitness becomes fixed. When the optimal fitness remains unchanged for 10 consecutive generations, the final designed parameters are output. The fitness function contains... , The quality is allowed to increase by 5%, with a minimum wall thickness of 2.5mm. Running the algorithm of this invention, as follows... Figure 3 As shown, the optimization converged after approximately 10 generations. The fitness curves indicate that the algorithm rapidly reduced the penalty term (satisfying constraints) in the early stages, and then finely optimized the frequency deviation term in the later stages. The optimal inner and outer diameter parameters of the five-segment aluminum alloy tower were finally obtained. Finite element analysis was performed to verify the optimal design, and the results are shown in Table 1. Finally, based on the core tower's shape, EPS foam rings were designed and assembled to ensure that the model's outer contour perfectly matches the 1:70 geometric scale. Figure 4 The diagram shows a tower structure designed based on an algorithm. The black dashed line represents the structural shape constructed from foam rings, and the black solid line represents the model tower structure made of aluminum alloy.

[0121] Thus, a scaled-down tower model that simultaneously meets the requirements of dynamic characteristics, mass, structural strength, and geometric shape is designed and can be directly used for subsequent experiments. This invention systematically solves the contradictions in traditional design, achieving high-quality, high-efficiency automated design of scaled-down models.

[0122] Table 1 Comparison of Design Optimization Values ​​and Target Values

[0123]

[0124] Note: In the model test, due to testing requirements, six-component force sensors need to be installed at the top and bottom of the tower. The installation height of these sensors must be considered when designing the model tower. The tower height here is 1.5m.

[0125] Example 1: A Synchronous Similarity Design Method for Tower Scale Based on Dual Constraints of First-Order Natural Frequency and Mass

[0126] This embodiment provides a method for synchronous similarity design of scaled-down towers. First, the scaling ratio is determined based on similarity theory, and the first-order target natural frequency of the scaled-down model is derived as the dynamic similarity target. The tower is divided into multiple structural segments along its height, and the inner and outer diameters of each segment are used as design variables to construct a feasible design space that satisfies basic geometric logic relationships. Based on this, a composite penalty fitness evaluation mechanism is constructed. The deviation from the dynamic target is quantified by the relative deviation between the first-order natural frequency calculated from the scaled-down model and the target frequency. Simultaneously, a total model mass constraint is introduced; when the total mass exceeds the target allowable range, the fitness value of the corresponding design scheme is reduced through a penalty mechanism.

[0127] Driven by a composite penalty fitness evaluation mechanism, a swarm search optimization algorithm is used to iteratively update the design variables. In each generation of the search, the dynamic characteristics and mass parameters of the current design scheme are calculated through finite element analysis, and the analysis results are fed back to the fitness evaluation mechanism for sorting and filtering. As the search process progresses, the swarm gradually converges towards the optimal direction of dynamic similarity and engineering feasibility. When the optimal fitness value remains stable over multiple generations, the search is terminated and the corresponding core load-bearing structural parameters are output as the load-bearing structural design scheme for the scaled-down tower model.

[0128] Example 2: A composite penalty fitness construction method incorporating geometric and process constraints

[0129] This embodiment provides a composite penalty fitness construction method for synchronous similar design of scaled-down tower sections. In this method, dynamic similarity deviation, mass exceedance, and violation of geometric and manufacturing constraints are uniformly mapped to a single fitness evaluation metric. The geometric and manufacturing constraints include minimum wall thickness constraints for each structural segment, logical constraints that the inner diameter is smaller than the outer diameter, and constraints that the structural dimensional parameters are within the feasible manufacturing range. When any constraint is violated, a corresponding penalty is introduced to reduce the overall fitness value of the design scheme.

[0130] By employing the above method, the structural design problem, which originally had multiple objectives and various constraints, is transformed into a single fitness-driven optimization problem. During the swarm search process, each candidate design scheme is ranked and selected solely based on this single fitness evaluation metric. This avoids the difficulty in setting weights during multi-objective trade-offs, enabling the search process to naturally avoid inmanufacturable or unrealizable structural schemes while ensuring dynamic similarity.

[0131] Example 3: Scaled-down model design method for separating the core load-bearing structure from the external cladding structure

[0132] This embodiment provides a structural design method for a scaled-down tower model. The scaled-down model is functionally divided into a core load-bearing structure and an external cladding structure. The core load-bearing structure is used to support the main stiffness and mass, and its structural parameters are optimized based on dynamic similarity and engineering constraints. The external cladding structure is used to replicate the external geometric contour of the actual tower and is not considered a primary dynamic load-bearing component in the design optimization.

[0133] In practice, the external cladding structure is formed by low-density, low-stiffness components attached to the outside of the core load-bearing structure. This allows it to achieve a scaled-down shape consistent with the actual tower without significantly affecting the overall dynamic characteristics of the scaled-down model. This structural separation method avoids interference from geometric similarity requirements on the dynamic similarity optimization process, improving the degree of freedom and design stability of the scaled-down model's dynamic characteristic control.

[0134] Example 4: Simulation-based Optimization Closed-Loop Collaborative Method Based on Finite Element Analysis

[0135] This embodiment provides a simulation optimization method for synchronous similar design of scaled-down tower structures. First, the design variables are parameterized to automatically generate the corresponding tower structure model. Then, finite element analysis is performed based on this model to obtain the first-order natural frequency, total mass, and other dynamic and engineering parameters of the scaled-down model. The analysis results are input as physical responses into a composite penalty fitness evaluation mechanism to evaluate the merits of the current design scheme.

[0136] During the optimization process, a closed-loop collaborative mechanism of "parameter generation—physical response analysis—fitness evaluation" is used to continuously update design variables. As iterations proceed, design schemes that do not meet dynamic similarity or engineering constraints are gradually eliminated, while the proportion of schemes that meet the requirements in the population continuously increases. When the search process reaches the preset iteration limit or the optimal fitness value no longer improves significantly, the optimization terminates and the final structural design parameters are output.

[0137] Example 5: Implementation of Segmented Tower Structure Optimization Method with Continuously Valued Design Variables

[0138] This embodiment provides a specific method for setting design variables. In this embodiment, the design variables include the changes in the inner and outer diameters of each segment of the tower structure. These changes are continuously measured within a preset feasible range to ensure sufficient adjustment accuracy of the structural parameters during the optimization process. By using continuous values, the discontinuity of the solution space caused by discrete variables is avoided, thus improving the convergence efficiency of the swarm search.

[0139] In the actual optimization process, each set of design variables corresponds to a complete set of tower structure parameter configurations, and its dynamic characteristics and mass parameters are calculated through finite element analysis. Combined with a composite penalty fitness evaluation mechanism, the search process can stably approximate the synergistic optimal solution between dynamic similarity and engineering feasibility within a continuous design space, thereby obtaining a structural design scheme that can be directly used for scaled-down model manufacturing.

[0140] It should be noted that embodiments of the present invention can be implemented in hardware, software, or a combination of both. The hardware portion can be implemented using dedicated logic; the software portion can be stored in memory and executed by a suitable instruction execution system, such as a microprocessor or dedicated-design hardware. Those skilled in the art will understand that the above-described devices and methods can be implemented using computer-executable instructions and / or included in processor control code, for example, such code provided on a carrier medium such as a disk, CD, or DVD-ROM, a programmable memory such as read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. The devices and modules of the present invention can be implemented by hardware circuitry such as very large-scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, or programmable hardware devices such as field-programmable gate arrays, programmable logic devices, etc., or by software executed by various types of processors, or by a combination of the above-described hardware circuitry and software, such as firmware.

[0141] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications, equivalent substitutions, and improvements made by those skilled in the art within the scope of the technology disclosed in the present invention, and within the spirit and principles of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for synchronous similar design of scaled-down tower sections, characterized in that, Includes the following steps: Obtain the target dynamic characteristics of the scaled model based on similarity theory; A feasible design space is constructed using the structural parameters of each segment of the tower as design variables. Construct a composite penalty fitness evaluation mechanism that simultaneously characterizes the degree of deviation from dynamic objectives and the violation of engineering constraints; Fitness functions are used for individual evaluation in genetic algorithms: ； In the formula: For each individual fitness value, the first term is the frequency deviation term, which drives the design to converge to the target frequency, and the second term is the quality penalty term, which applies a linear penalty when the quality exceeds the limit. The first term is the quality penalty weighting coefficient, and the second term is the geometric penalty term. Each violation of geometric / engineering constraints is quantitatively penalized, especially when the wall thickness is insufficient. , G represents the geometric penalty weighting coefficient; Penalty(·) denotes the penalty function, g j This represents the j-th constraint term. This indicates the minimum wall thickness at the segmentation point. This indicates the actual wall thickness at the segmentation point. Indicates target quality; and As an adjustable parameter, it enables differentiated weight configuration for different constraints of varying importance; the first-order natural frequency of the scaled model is calculated. With the total mass of the model The first natural frequency of the scaled model Infinitely approaching the target frequency derived from the similarity law ; Driven by the composite penalty fitness evaluation mechanism, the design variables are optimized through a group search. Output the core load-bearing structural parameters that achieve the optimal balance between dynamic similarity and engineering feasibility, and use them as the load-bearing structural design scheme for the scaled-down model; The structural design methods for scaled-down tower models include: The scaled-down model is divided into a core load-bearing structure for supporting stiffness and mass, and an external cladding structure for replicating the external geometry of the actual tower. The core load-bearing structure is optimized based on dynamic similarity and engineering constraints, and the external cladding structure is used to achieve geometric similarity without affecting the dynamic characteristics. The external cladding structure is formed by low-density, low-stiffness components attached to the outside of the core load-bearing structure, so that the external geometric profile is consistent with the scaled-down shape of the solid tower.

2. The method as described in claim 1, characterized in that, The degree of deviation from the dynamic target is characterized by the relative deviation between the first-order natural frequency of the scaled model and the target frequency derived from similarity theory, so that the optimization process takes dynamic similarity as the dominant convergence direction.

3. The method as described in claim 1, characterized in that, The engineering constraints include the total mass constraint of the scaled model. When the total mass of the model exceeds the allowable range of the target mass, the fitness value of the corresponding design scheme is reduced by a penalty method, thereby suppressing the search probability of over-mass schemes.

4. The method as described in claim 1, characterized in that, The composite penalty fitness construction method for tower scale-down synchronous similarity design includes: mapping the dynamic similarity deviation, mass exceedance degree, and geometric and manufacturing constraint violation degree into a single fitness evaluation quantity; By ranking and filtering different design schemes using the single fitness evaluation metric, the multi-objective, multi-constraint design problem is transformed into a single-objective optimization problem that can directly drive swarm search.

5. The method as described in claim 4, characterized in that, The geometric and manufacturing constraints include minimum wall thickness constraints for segmented structures, logical constraints that the inner diameter is smaller than the outer diameter, and constraints that the dimensional parameters are within the feasible range for manufacturing. Each constraint is introduced into the fitness evaluation process through a corresponding penalty amount.

6. The method as described in claim 1, characterized in that, Simulation optimization methods for synchronous similar design of scaled towers include: parameterizing design variables to generate structural models; Finite element analysis is performed based on the structural model to obtain dynamic characteristics and mass parameters; The analysis results are fed back to the fitness evaluation mechanism, forming a closed-loop collaborative process between parameter generation, physical response analysis and optimization evaluation.

7. The method as described in claim 6, characterized in that, The design variables include the changes in the inner diameter and outer diameter of each segment of the tower structure, and the changes are continuously taken within a preset feasible range.

8. The method as described in any one of claims 1, 4, or 6, characterized in that, When the optimal fitness value remains stable over multiple generations of search, or when the preset iteration limit is reached, the search is terminated and the corresponding design parameters are output as the final structural scheme.

Images