Topological optimization algorithm-based lightweight design method for steel-concrete combined wind power tower drum

Abstract

The lightweight design method for the steel-concrete combined wind power tower drum based on the topological optimization algorithm comprises the steps that S1, a multi-physics field coupling model of the steel-concrete combined wind power tower drum is constructed, and coupling parameters are determined; s2, material layouts of the steel-concrete combined wind power tower drum on the macro scale and the micro scale are obtained, and multi-scale collaborative optimization is carried out by dynamically adjusting scale parameters; s3, constructing a material distribution strategy, determining the distribution form and orientation of a steel material in a tension area and a compression area of the tower drum, and reasonably configuring a concrete material; s4, predicting tower performance under different design variables through the deep neural network model; and S5, performing multi-target evaluation on the optimized steel-concrete combined wind power tower drum design scheme, and determining a lightweight design scheme by constructing a comprehensive evaluation function and weighing the relationship between targets. According to the method, on the premise that the mechanical property and the structural reliability of the wind power tower drum are guaranteed, accurate regulation and control and lightweight design of material distribution are achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of lightweight design of wind turbine towers, and in particular to a lightweight design method of a steel-concrete composite wind turbine tower based on a topology optimization algorithm. Background Art

[0002] As the global demand for clean energy continues to rise, the wind power industry, as an important renewable energy field, is ushering in a period of opportunity for vigorous development. As the key supporting structure of the wind turbine generator set, the performance and weight of the wind turbine tower directly affect the stability, safety and economy of the entire wind power system. In practical applications, the traditional wind turbine tower design has exposed many shortcomings in the face of the growing demand for wind power generation. On the one hand, the overweight tower not only increases the material cost, causing a significant increase in construction investment, but also increases the difficulty and cost of transportation and installation, limiting the development of wind power projects in some areas with inconvenient transportation; on the other hand, in complex natural environments, such as areas with strong winds and frequent earthquakes, the existing tower design is difficult to achieve lightweight while ensuring structural strength and stability. Therefore, the development of a technology that can effectively realize the lightweight design of steel-concrete composite wind turbine towers is of vital significance to enhancing the competitiveness of the wind power industry and promoting the sustainable development of the wind power industry.

[0003] Traditional steel-concrete composite wind turbine tower design usually adopts empirical design methods or simple mechanical calculations. In empirical design, engineers determine the structural dimensions and material configuration of the tower based on the experience accumulated from past projects and basic mechanical principles. Although this method can meet engineering requirements to a certain extent, it lacks accuracy and scientificity. Since the complex loads borne by the tower under actual working conditions, such as different wind conditions and earthquake effects, are not fully considered, the designed tower is often overly conservative, using a large amount of unnecessary materials, resulting in a waste of resources.

[0004] Although simple mechanical calculation methods are more accurate than empirical designs, they are limited to basic stress and strain analysis of the tower and are unable to fully consider the impact of multi-physical field coupling. In the face of complex actual environments, this design method cannot accurately evaluate the performance of the tower, resulting in safety hazards during the use of the tower. For example, in areas with strong winds, the tower vibrates or even gets damaged because it cannot withstand the complex wind loads; in earthquake-prone areas, traditional tower designs cannot effectively withstand the additional loads brought by earthquakes, threatening the safety of the entire wind power generation facility. In general, traditional technical solutions have obvious deficiencies in design accuracy, resource utilization efficiency, and the ability to cope with complex working conditions.

[0005] The existing technology has made certain progress in the design of wind turbine towers, adopting numerical simulation methods such as finite element analysis. Finite element analysis can simulate and analyze the mechanical properties of the tower in more detail, taking into account the geometric shape, material properties of the tower, and various load conditions, providing more reliable data support for the design. By establishing a finite element model of the tower, engineers can intuitively observe the stress and strain distribution of the tower under different loads, so as to optimize the design.

[0006] However, there are still some drawbacks in the existing technology. On the one hand, finite element analysis is usually carried out based on a single scale, unable to take into account the influence of material layout on the performance of the tower under both macro and micro scales at the same time. At the macro scale, it is impossible to accurately grasp the relationship between the force-bearing characteristics of the overall structure and the material distribution; at the micro scale, the material optimization design of complex structure areas and force-bearing parts is not fine enough. On the other hand, when dealing with multi-physical field coupling problems, although some factors can be considered, the coupling model is not perfect. The description of the coupling effects such as wind-structure and earthquake-structure is not accurate enough, resulting in a deviation between the calculation results and the actual situation. In addition, in terms of the material distribution strategy, the existing technology lacks a systematic method and it is difficult to achieve the optimal configuration of steel materials and concrete materials according to the force-bearing characteristics of different regions of the tower. These deficiencies limit the further development of wind turbine tower design in the direction of lightweight and high performance. Summary of the Invention

[0007] Based on the above content, this application proposes a lightweight design method for steel-concrete composite wind turbine towers based on the topology optimization algorithm, including:

[0008] S1. Construct a multi-physical field coupling model of the steel-concrete composite wind turbine tower and determine the coupling parameters;

[0009] S2. Process the multi-physical field coupling model through an adaptive multi-scale topology optimization algorithm to obtain the material layout at the macro scale and the micro scale. By dynamically adjusting the scale parameters, automatically switch the optimization regions at the macro and micro scales according to the structural response, and perform multi-scale collaborative optimization;

[0010] S3. Construct a material distribution strategy. In the tensile and compressive regions of the tower, determine the distribution form and orientation of steel materials and rationally allocate concrete materials;

[0011] S4. Through a deep neural network model, learn the complex mapping relationship between different design parameters and the performance of the tower, and use the neural network model as a predictor to predict the performance of the tower under different design variables;

[0012] S5. Conduct a multi-objective evaluation of the optimized design scheme of the steel-concrete composite wind turbine tower. By constructing a comprehensive evaluation function, weigh the relationship between various objectives and determine the lightweight design scheme.

[0013] Preferably, before constructing the multi-physical field coupling model of the steel-concrete composite wind power tower in S1, multi-source data of the wind power tower's multi-physical fields are obtained; the multi-source data includes wind condition data, geological structure, and the structural parameters of the steel-concrete composite wind power tower.

[0014] Preferably, when constructing the multi-physical field coupling model of the steel-concrete composite wind power tower in S1 by combining the obtained wind condition data, geological structure, and the structural parameters of the steel-concrete composite wind power tower, it specifically includes:

[0015] Obtain the wind field data vector W from the wind condition data to get the wind load vector F w , the formula is: F w =R w ·W, where R w is the wind field influence coefficient matrix; obtain the seismic response coefficient matrix R e from the geological structure to get the additional load vector F e under the earthquake action, the formula is: F e =R e ·E, where E is the seismic excitation vector; obtain the comprehensive characteristic coefficient ξ of the steel-concrete composite material from the structural parameters of the steel-concrete composite wind power tower, the formula is: ξ = K sc ·(P s +P c ), where P s is the vector of steel material characteristic parameters, P c is the vector of concrete characteristic parameters, and K sc is the steel-concrete composite coefficient matrix;

[0016] The multi-physical field coupling model of the steel-concrete composite wind power tower is formed by incorporating the wind-structure coupling parameter and the earthquake-structure coupling parameter into the mechanical equilibrium equation of the tower; the wind-structure coupling parameter γ w , through the formula: is determined; the earthquake-structure coupling parameter γ e , through the formula: is determined, and the coupling parameters are incorporated into the mechanical equilibrium equation of the tower KU = F w +F e in which K is the tower structure stiffness matrix, U is the nodal displacement vector, to form the multi-physical field coupling model of the steel-concrete composite wind power tower.

[0017] Preferably, in S2, an adaptive multi-scale topology optimization algorithm is used to process the above-mentioned model to obtain material layouts at macro and micro scales; at the macro scale, according to the overall stress characteristics and design requirements of the tower, larger-sized design units are divided to determine the macro layout of material distribution; at the micro scale, according to the stress parts and complex structural areas of the tower, the design units are refined and the material distribution is adjusted; by dynamically adjusting the scale parameters, the macro and micro scale material layout areas are automatically switched according to the structural response, and multi-scale collaborative optimization is performed.

[0018] Preferably, the macro scale is obtained by obtaining the height and diameter of the tower, and an adaptive hierarchical division strategy is adopted. Taking the height direction of the tower as an example, the height interval where the stress concentration area and the connection part are located is divided according to a smaller height interval Δh1. For the middle main part with relatively uniform stress distribution, a larger height interval Δh2 is used for division. In the circumferential direction of the tower, according to the circumferential stress change curve under the action of wind load, the area with a larger stress change gradient is divided more densely, and the division scale of other areas is appropriately relaxed.

[0019] Preferably, the macroscopic layout of the material distribution on the macroscopic scale is determined by a multi-objective optimization function O, and the formula is: O=w1·σ+w2·∈+w3·m, where σ is the stress level, ∈ is the strain energy, m is the material dosage, w1 is the stress level weight coefficient, w2 is the strain energy weight coefficient, and w3 is the material dosage weight coefficient; for areas subjected to greater tensile stress, steel is preferentially laid out, and the steel distribution ratio p s According to the tensile stress σ t according to Determine, where σ c is the regional compressive stress, is the preset maximum distribution ratio of steel; for the compression area, which is mainly concrete, according to the compressive stress σ c Adjust the concrete strength grade so that the strength grade and compressive stress meet C min is the minimum design strength grade, k is the strength adjustment coefficient, The minimum compressive stress threshold is used to achieve a reasonable layout of materials on a macro scale.

[0020] Preferably, at the microscopic scale of S2, detailed design units are performed on the stress-bearing parts and complex structural areas of the tower and material distribution is adjusted, specifically including:

[0021] Construct a microscopic digital model that maps 1:1 with the actual structure through digital twin technology. Refine the unit size to the micron level according to the direction and magnitude of the force. For the tensile stress concentration area, arrange steel fibers along the direction of the microscopic stress streamline, and control the angle between the direction of the steel fibers and the principal tensile stress direction within a very small range. Calculate the optimal spacing d of the steel fibers through an optimization algorithm s , the formula is: where k s is the steel characteristic coefficient, E s is the elastic modulus of the steel, σ max is the maximum tensile stress, G s is the shear modulus of the steel, ρ s is the density of the steel.

[0022] Preferably, the construction of the material distribution strategy in S3 is through biomechanical simulation. In the tensile and compressive regions of the tower barrel, determine the distribution form and orientation of the steel material, and rationally configure the concrete material. Specifically, it includes:

[0023] In the tensile region, through multi-scale finite element simulation of the stress distribution of the tower barrel under actual working conditions, according to the simulation results, use the genetic algorithm to optimize the distribution form and orientation of the steel material. For the parts with large axial tensile force, design the steel material into a continuous spiral structure, and the spiral angle θ is determined according to the tensile force F a and the tower barrel radius r through the formula where k θ is the material characteristic correlation coefficient, σ θ is the yield strength of the steel;

[0024] In the compressive region, use the topology optimization algorithm to determine the optimal porosity and aggregate distribution of the concrete. The formula is: where is the initial porosity, k c is the concrete characteristic correlation coefficient.

[0025] Preferably, the deep neural network model in S4 is constructed by fusing the attention mechanism and transfer learning. Specifically, it includes the following steps:

[0026] S41. Collect a large amount of historical data of wind turbine towers covering different design parameters and corresponding performance indicators, perform normalization processing, and divide it into a training set, a validation set, and a test set;

[0027] S42. Construct a multi-layer perceptron as the basic architecture, embed an attention module between key layers, and adaptively focus on the design parameters that have a greater impact on the tower performance by calculating the correlation weights between different input features;

[0028] S43. By using transfer learning technology, select a pre-trained model that performs excellently in the field of large mechanical structure design, transfer the weights of some of its layers to the newly built neural network, fine-tune the transferred model, and optimize the network parameters to adapt to the tower barrel design data;

[0029] S44. During the training process, adopt an adaptive learning rate adjustment strategy to converge the learning rate near the optimal solution; the learning rate gradually decays according to the cosine function as the training progresses, and finely adjusts the parameters when approaching the optimal solution;

[0030] S45. When using the trained deep neural network model as a predictor, input different design variables, perform weighted processing on the input features through the embedded attention module to highlight the key design parameters, and after the non-linear transformation of the multi-layer perceptron, map the fused feature information to the tower barrel performance space, and output the predicted value of the tower barrel performance under the corresponding design variables.

[0031] Preferably, in S5, the multi-objective evaluation selects the degree of light weight, fatigue life, anti-wind vibration stability, and manufacturing cost as evaluation objectives, constructs a comprehensive evaluation function, weighs the relationship between each objective, and determines the light weight design scheme;

[0032] When constructing the comprehensive evaluation function, determine the objective weight w j by the analytic hierarchy process, and the formula is: where is the initial weight of the matrix, α is the balance coefficient, is the entropy weight of the jth objective, and obtain the comprehensive evaluation value E, and the formula is: where f j represents the jth evaluation objective, and select the scheme with the largest comprehensive evaluation value E through the comprehensive evaluation function as the light weight design scheme.

[0033] Compared with the prior art, the technical solution of the present application has the following technical effects:

[0034] The present invention constructs a multi - physical - field coupling model for a steel - concrete composite wind turbine tower, determines the coupling parameters, and solves the technical problem that the traditional design method cannot accurately consider the actual stress conditions of the wind turbine tower under the action of multiple complex physical fields. In practical applications, the wind turbine tower is affected by multiple physical fields such as wind loads and seismic actions. Due to the lack of comprehensive consideration of these complex factors in traditional design, the performance of the designed tower is difficult to meet the actual requirements. In this application, by obtaining wind condition data, geological structures, and tower structure parameters, the wind load vector, the additional load vector under seismic action, and the comprehensive characteristic coefficient of the steel - concrete material are determined respectively, and then a multi - physical - field coupling model is formed. This enables the accurate simulation of the mechanical behavior of the tower in the actual environment during the design stage, obtaining a design result that is more in line with the actual working conditions, effectively improving the reliability and stability of the tower in a complex environment, and providing a solid guarantee for the safe operation of the wind power generation system.

[0035] The present invention uses an adaptive multi - scale topology optimization algorithm to process the multi - physical - field coupling model, dynamically adjusts the scale parameters, and conducts a multi - scale collaborative optimization technical solution, solving the problem that traditional design cannot take into account both the macroscopic overall structural characteristics and the refined design of microscopic complex regions in material layout. When traditional design conducts material layout, it either only focuses on the overall macroscopic layout and ignores the optimization of microscopic complex structural regions; or only pays attention to microscopic details but it is difficult to ensure the rationality of the overall structure. In this application, at the macroscopic scale, according to the overall stress characteristics and design requirements of the tower, larger - sized design units are divided to determine the macroscopic material layout. For example, the height and diameter of the tower are adaptively graded and divided, and the division scale of the stress - concentration region and the connection part is optimized; at the microscopic scale, for the stressed parts and complex structural regions of the tower, the design units are refined, and the digital twin technology is used to construct a microscopic model to arrange steel fibers. Through multi - scale collaborative optimization, the reasonable distribution of materials is achieved. While ensuring the structural strength and stability of the tower, the material consumption is effectively reduced, the purpose of lightweight design is achieved, the material utilization efficiency is improved, and the production cost is reduced.

[0036] The present invention adopts a construction material distribution strategy, and determines the distribution form, orientation of steel materials and rationally allocates concrete materials in the tension area and compression area of the tower barrel, solving the problem of unreasonable configuration of steel-concrete materials in different stress areas of the tower barrel in traditional designs. Traditional designs often lack a scientific material distribution method, unable to fully utilize the tensile properties of steel in the tension area and unable to rationally utilize the compressive characteristics of concrete in the compression area. In this application, in the tension area, through multi-scale finite element simulation and genetic algorithm optimization of the steel material distribution form and orientation, for example, the part with a large axial tension is designed into a continuous spiral structure and the spiral angle is determined; in the compression area, the topology optimization algorithm is used to determine the optimal porosity and aggregate distribution of the concrete. Such a design can give full play to the advantages of both steel and concrete materials, making the material properties of each part of the tower barrel match the stress conditions, improving the bearing capacity and durability of the tower barrel, extending the service life of the tower barrel, reducing the maintenance cost, and enhancing the economic benefits of wind power generation facilities.

[0037] The present invention solves the problem that it is difficult to accurately establish the complex mapping relationship between design parameters and the performance of the tower barrel by constructing a deep neural network model integrating an attention mechanism and transfer learning and using it as a predictor to predict the performance of the tower barrel. Traditional prediction methods often rely on simple mathematical models or empirical formulas and cannot accurately reflect the non-linear and complex relationship between numerous design parameters and the performance of the tower barrel. This application collects a large amount of historical data of wind power tower barrels and normalizes them, constructs a multi-layer perceptron and embeds an attention module to make the model adaptively focus on key design parameters. At the same time, transfer learning is used to transfer the weights of the pre-trained model and fine-tune it to make the model more adaptable to the tower barrel design data. In the prediction process, the role of key design parameters is highlighted to achieve accurate prediction. This provides a more accurate design reference basis for designers, helps them quickly evaluate the performance of different design schemes, greatly shortens the design cycle, improves the design efficiency, and accelerates the research and development process of new wind power tower barrels.

[0038] The above description is only an overview of the technical solution of this application. In order to be able to understand the technical means of this application more clearly, so as to be implemented in accordance with the content of the specification, and in order to make the above and other purposes, features and advantages of this application more obvious and understandable, the following takes the preferred embodiments of this application and combines with the drawings to describe in detail as follows.

[0039] According to the following detailed description of the specific embodiments of this application in combination with the drawings, those skilled in the art will be more clear about the above and other purposes, advantages and features of this application. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] To more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the accompanying drawings required for the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other accompanying drawings can also be obtained based on these drawings. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts do not necessarily draw according to the actual scale.

[0041] Figure 1 Flow chart of the lightweight design method of the steel-concrete composite wind turbine tower based on the topology optimization algorithm of the present invention;

[0042] Figure 2 Flow chart of the construction of the deep neural network model for the lightweight design of the steel-concrete composite wind turbine tower based on the topology optimization algorithm of the present invention. Detailed implementation manners

[0043] To make the objectives, technical solutions, and advantages of the embodiments of the present application clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are some, but not all, of the embodiments of the present application. In the following description, specific details such as specific configurations and components are provided only to assist in a comprehensive understanding of the embodiments of the present application. Therefore, those skilled in the art should clearly understand that various changes and modifications can be made to the embodiments described here without departing from the scope and spirit of the present application. In addition, for the sake of clarity and conciseness, the description of known functions and structures is omitted in the embodiments.

[0044] It should be understood that the "one embodiment" or "the present embodiment" mentioned throughout the specification means that the specific features, structures, or characteristics related to the embodiment are included in at least one embodiment of the present application. Therefore, the "one embodiment" or "the present embodiment" appearing throughout the specification does not necessarily refer to the same embodiment. In addition, these specific features, structures, or characteristics can be combined in any suitable manner in one or more embodiments.

[0045] In addition, the present application may repeat reference numerals and / or letters in different examples. This repetition is for the purpose of simplification and clarity, and does not itself indicate the relationship between the various embodiments and / or settings discussed.

[0046] In this text, the term "and / or" is merely a description of the relationship between associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, B exists alone, and both A and B exist simultaneously. The term " / and" in this text describes another relationship between associated objects, indicating that there can be two relationships. For example, A / and B can represent: A exists alone, and both A and B exist. Additionally, the character " / " in this text generally indicates that the associated objects before and after are in an "or" relationship.

[0047] The term "at least one" in this text is merely a description of the relationship between associated objects, indicating that there can be three relationships. For example, at least one of A and B can represent: A exists alone, both A and B exist simultaneously, and B exists alone.

[0048] It should also be noted that in this text, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variant thereof are intended to cover non-exclusive inclusion.

[0049] Embodiment 1

[0050] This embodiment details a lightweight design method for a steel-concrete composite wind turbine tower based on a topology optimization algorithm, as Figure 1 , including the following steps:

[0051] S1. Construct a multi-physics coupling model of the steel-concrete composite wind turbine tower and determine the coupling parameters;

[0052] S2. Process the multi-physics coupling model through an adaptive multi-scale topology optimization algorithm to obtain the material layout at the macro scale and the micro scale. By dynamically adjusting the scale parameters, automatically switch the optimization regions at the macro and micro scales according to the structural response for multi-scale collaborative optimization;

[0053] S3. Construct a material distribution strategy. In the tensile and compressive regions of the tower, determine the distribution form and orientation of the steel material and rationally allocate the concrete material;

[0054] S4. Through a deep neural network model, learn the complex mapping relationship between different design parameters and the tower performance. Use the neural network model as a predictor to predict the tower performance under different design variables;

[0055] S5. Conduct a multi-objective evaluation of the optimized design scheme of the steel-concrete composite wind turbine tower. By constructing a comprehensive evaluation function, weigh the relationship between various objectives to determine the lightweight design scheme.

[0056] Further, before constructing the multi-physical field coupling model of the steel-concrete composite wind power tower in S1, multi-source data of the wind power tower's multi-physical fields are obtained; the multi-source data includes wind condition data, geological structure, and structural parameters of the steel-concrete composite wind power tower.

[0057] Further, when constructing the multi-physical field coupling model of the steel-concrete composite wind power tower in S1 by combining the obtained wind condition data, geological structure, and structural parameters of the steel-concrete composite wind power tower, it specifically includes:

[0058] The wind field data vector W is obtained from the wind condition data to get the wind load vector F w , and the formula is: F w = R w · W, where R w is the wind field influence coefficient matrix; the seismic response coefficient matrix R e is obtained from the geological structure to get the additional load vector F e under seismic action, and the formula is: F e = R e · E, where E is the seismic excitation vector; the comprehensive characteristic coefficient ξ of the steel-concrete composite material is obtained from the structural parameters of the steel-concrete composite wind power tower, and the formula is: ξ = K sc · (P s + P c ), where P s is the vector of steel material characteristic parameters, P c is the vector of concrete characteristic parameters, and K sc is the steel-concrete composite coefficient matrix;

[0059] The multi-physical field coupling model of the steel-concrete composite wind power tower is formed by incorporating the wind-structure coupling parameter and the seismic-structure coupling parameter into the mechanical equilibrium equation of the tower; the wind-structure coupling parameter γ w , is determined by the formula: ; the seismic-structure coupling parameter γ e , is determined by the formula: , and the coupling parameters are incorporated into the mechanical equilibrium equation of the tower KU = F w + F e where K is the tower structure stiffness matrix, U is the nodal displacement vector, to form the multi-physical field coupling model of the steel-concrete composite wind power tower.

[0060] Furthermore, an adaptive multi-scale topology optimization algorithm is used in S2 to process the above model to obtain the material layout at the macro and micro scales. At the macro scale, the design units of larger sizes are divided according to the overall stress characteristics and design requirements of the tower, and the macro layout of the material distribution is determined. At the micro scale, the design units are refined and the material distribution is adjusted according to the stress parts and complex structural areas of the tower. By dynamically adjusting the scale parameters, the material layout areas at the macro and micro scales are automatically switched according to the structural response, and multi-scale collaborative optimization is performed.

[0061] Furthermore, at the macro scale, by obtaining the height and diameter of the tower, an adaptive hierarchical division strategy is adopted. Taking the height direction of the tower as an example, the height interval where the stress concentration area and the connection part are located is divided according to a smaller height interval Δh1. For the middle main part with relatively uniform stress distribution, a larger height interval Δh2 is used for division. In the circumferential direction of the tower, according to the circumferential stress change curve under the action of wind load, the area with a larger stress change gradient is divided more densely, and the division scale of other areas is appropriately relaxed.

[0062] Furthermore, the macroscopic layout of material distribution on a macroscopic scale is determined by a multi-objective optimization function O, and the formula is: O = w1·σ+w2·∈+w3·m, where σ is the stress level, ∈ is the strain energy, m is the material dosage, w1 is the stress level weight coefficient, w2 is the strain energy weight coefficient, and w3 is the material dosage weight coefficient; for areas subject to greater tensile stress, steel is given priority, and the steel distribution ratio p s According to the tensile stress σ t according to Determine, where σ c is the regional compressive stress, is the preset maximum distribution ratio of steel; for the compression area, which is mainly concrete, according to the compressive stress σ c Adjust the concrete strength grade so that the strength grade and compressive stress meet C min is the minimum design strength grade, k is the strength adjustment coefficient, The minimum compressive stress threshold is used to achieve a reasonable layout of materials on a macro scale.

[0063] Furthermore, at the microscopic scale of S2, the stress-bearing parts and complex structural areas of the tower are refined and the material distribution is adjusted, including:

[0064] Through digital twin technology, a microscopic digital model with a 1:1 mapping to the actual structure is constructed. According to the direction and magnitude of the force, the unit size is refined to the micron level. For the tensile stress concentration area, the steel fibers are arranged according to the direction of the microscopic stress streamlines, so that the angle between the steel fiber direction and the main tensile stress direction is controlled within a very small range. The optimal spacing d of the steel fibers is calculated through an optimization algorithm.s , the formula is: where k s is the characteristic coefficient of the steel, E s is the elastic modulus of the steel, σ max is the maximum tensile stress, G s is the shear modulus of the steel, ρ s is the density of the steel.

[0065] Furthermore, the construction of the material distribution strategy in S3 is carried out through biomechanical simulation. In the tensile and compressive regions of the tower barrel, the distribution form and orientation of the steel material are determined, and the concrete material is reasonably configured, specifically including:

[0066] In the tensile region, through multi-scale finite element simulation of the stress distribution of the tower barrel under actual working conditions, according to the simulation results, the genetic algorithm is used to optimize the distribution form and orientation of the steel material. For the parts with large axial tensile force, the steel material is designed into a continuous spiral structure, and the spiral angle θ is determined according to the tensile force F a and the tower barrel radius r through the formula where k θ is the material characteristic correlation coefficient, σ θ is the yield strength of the steel;

[0067] In the compressive region, using the topology optimization algorithm, the optimal porosity and aggregate distribution of the concrete are determined, and the formula is: where is the initial porosity, k c is the concrete characteristic correlation coefficient.

[0068] Furthermore, the deep neural network model in S4 is constructed by fusing the attention mechanism and transfer learning, as Figure 2 shown, specifically including the following steps:

[0069] S41. Collect a large amount of historical data of wind turbine towers covering different design parameters and corresponding performance indicators, perform normalization processing, and divide it into a training set, a validation set, and a test set;

[0070] S42. Construct a multi-layer perceptron as the basic architecture, embed an attention module between key layers, and adaptively focus on the design parameters that have a greater impact on the tower performance by calculating the correlation weights between different input features;

[0071] S43. Through transfer learning technology, select a pre-trained model that performs well in the field of large mechanical structure design, transfer the weights of some of its layers to the newly built neural network, and fine-tune the transferred model to optimize the network parameters to adapt to the tower design data;

[0072] S44. During the training process, an adaptive learning rate adjustment strategy is adopted, and the learning rate converges to near the optimal solution; as the training progresses, the learning rate gradually decays according to the cosine function, and the parameters are finely adjusted when approaching the optimal solution;

[0073] S45. When the trained deep neural network model is used as a predictor, different design variables are input, and the input features are weighted by the embedded attention module to highlight the key design parameters. After the non-linear transformation of the multi-layer perceptron, the fused feature information is mapped to the tower performance space, and the tower performance prediction value corresponding to the design variables is output.

[0074] Furthermore, in S5, the multi-objective evaluation selects the degree of light weight, fatigue life, anti-wind vibration stability, and manufacturing cost as the evaluation objectives, constructs a comprehensive evaluation function, weighs the relationship between each objective, and determines the light weight design scheme;

[0075] When constructing the comprehensive evaluation function, the objective weight w is determined by the analytic hierarchy process j , and the formula is: where is the initial weight of the matrix, α is the balance coefficient, is the entropy weight of the j-th objective, and the comprehensive evaluation value E is obtained. The formula is: where f j represents the j-th evaluation objective, and the scheme with the largest comprehensive evaluation value E selected by the comprehensive evaluation function is used as the light weight design scheme.

[0076] This embodiment details the realization of macro-micro multi-scale collaborative optimization and accurate determination of material layout by constructing a multi-physical field coupling model of a steel-concrete composite wind turbine tower and combining an adaptive multi-scale topology optimization algorithm; constructing a material distribution strategy by biomechanical simulation to rationally allocate steel-concrete materials; predicting the tower performance with a deep neural network model integrating the attention mechanism and transfer learning; and determining the light weight design scheme through multi-objective evaluation. It effectively solves the deficiencies of traditional designs in considering complex working conditions, material layout, performance prediction, and scheme evaluation, etc., and achieves the core technical effects of significantly reducing weight, lowering costs, and improving design efficiency while ensuring the strength, stability, and service life of the tower.

[0077] Based on Embodiment 1, this embodiment describes that the multi-source data in S1 includes wind condition data, geological structure, and the structural parameters of the steel-concrete composite wind turbine tower, specifically:

[0078] Among the multi-source data obtained before constructing the multi-physical field coupling model of the steel-concrete composite wind turbine tower, in terms of wind condition data, a detection drone is used to conduct real-time monitoring of the wind conditions in all directions and at multiple height levels in the surrounding area of the tower, collecting conventional wind speed and wind direction data, and accurately measuring the wind shear rate and gust factor at different heights. At the same time, high-resolution images of meteorological satellites and numerical weather prediction models are used to obtain long-term wind condition change trend data, including seasonal and annual wind speed fluctuation laws and wind direction change characteristic information;

[0079] For obtaining geological structure data, three-dimensional seismic exploration technology and microtremor detection method are adopted; three-dimensional seismic exploration can accurately draw the geological structure image within hundreds of meters deep underground in the area where the tower is located, clearly distinguishing the lithology, thickness of different strata, and geological structure characteristics such as faults and folds; the microtremor detection method collects microtremor signals from underground media by arranging multiple microtremor sensors on the ground, analyzes their spectral characteristics, and then inversely calculates the shallow geological structure to obtain more detailed near-surface geological information, such as the layered structure of the soil, shear wave velocity and other data, providing a basis for accurately evaluating the impact of earthquakes on the tower.

[0080] Regarding the structural parameters of the steel-concrete composite wind turbine tower, high-precision three-dimensional laser scanning technology is used to scan the tower comprehensively to obtain the accurate external dimensions of the tower, including the tower height, diameter of each section, and wall thickness data, with an accuracy reaching the millimeter level; through non-destructive testing technologies such as ultrasonic flaw detection and magnetic particle flaw detection, the quality status of the steel and concrete inside the tower is detected to obtain information such as the internal defect distribution of the steel and the compactness of the concrete. In addition, a material testing machine is used to conduct mechanical property tests on the steel-concrete material samples collected from the tower to obtain detailed material parameters such as the elastic modulus, yield strength, fatigue limit of the steel, and compressive strength, tensile strength, and elastic modulus of the concrete, providing comprehensive and accurate structural parameter data for subsequent model construction.

[0081] This embodiment details the technical means for obtaining multi-source data, which can provide comprehensive and accurate data support for the construction of the multi-physical field coupling model of the steel-concrete composite wind turbine tower; the comprehensive collection of wind condition data enables the model to accurately simulate the action of wind on the tower; the precise detection of geological structure data can more accurately evaluate the earthquake impact; the accurate acquisition of the tower structural parameters ensures that the model's simulation of the mechanical properties of the tower conforms to the actual situation. The comprehensive application of these data greatly improves the accuracy and reliability of the model, lays a solid data foundation for the lightweight design of the tower, and effectively promotes the development of wind turbine tower design technology.

[0082] Based on Embodiment 1, this embodiment describes that the deep neural network model in S4 is constructed by integrating the attention mechanism and transfer learning, specifically as follows:

[0083] S41. Collect a large amount of historical data of wind turbine towers covering different design parameters and corresponding performance indicators, perform normalization processing, and divide it into a training set, a validation set, and a test set;

[0084] The design parameters include the geometric dimensions, material properties, and connection methods of the tower; the performance indicators involve strength, stiffness, stability, and fatigue life; to ensure data quality, the collected data needs to be cleaned to remove outliers and missing values; use the Z-score normalization method to normalize the data, scale the data to a range with a mean of 0 and a standard deviation of 1, and divide the normalized data into a training set, a validation set, and a test set according to a ratio of 7:2:1 to ensure the performance of the model on different data sets;

[0085] S42. Construct a multi-layer perceptron as the basic architecture, embed an attention module between key layers, and adaptively focus on the design parameters that have a greater impact on the tower performance by calculating the correlation weights between different input features;

[0086] Construct a multi-layer perceptron as the basic architecture, which includes an input layer, multiple hidden layers, and an output layer. The number of neurons in the input layer is the same as the dimension of the design parameters, and the number of neurons in the output layer is the same as the number of performance indicators; embed an attention module between key layers. The attention module consists of three parts: query, key, and value. The input features are respectively linearly transformed to obtain query, key, and value vectors, then calculate the similarity score between the query vector and the key vector, use dot product operation and scaling, and use the Softmax function to convert the score into a probability distribution to obtain the attention weight. Multiply the attention weight by the value vector and sum to obtain the weighted feature representation, which can adaptively focus on the design parameters that have a greater impact on the tower performance.

[0087] S43. Through transfer learning technology, select a pre-trained model that performs well in the field of large mechanical structure design, transfer the weights of some of its layers to the newly built neural network, and fine-tune the transferred model to optimize the network parameters to adapt to the tower design data;

[0088] Using transfer learning technology, pre-trained models with excellent performance in the design field are selected. These models have been trained on a large amount of mechanical structure design data and have learned some general feature representations. The weights of some layers of the pre-trained model (such as convolutional layers and early fully connected layers) are transferred to the newly built neural network to initialize the corresponding layers of the newly built model. During the transfer process, the weights of some transferred layers are frozen, and only the weights of subsequent layers are allowed to be updated to avoid overfitting. Then, the transferred model is fine-tuned using the training set data of the wind turbine tower. During the fine-tuning process, the Stochastic Gradient Descent (SGD) optimization algorithm is adopted, and the gradients are calculated through backpropagation to update the weights of the trainable layers, enabling the model to gradually adapt to the characteristics of the tower design data;

[0089] S44. During the training process, an adaptive learning rate adjustment strategy is adopted to converge the learning rate to near the optimal solution; the learning rate gradually decays according to the cosine function as the training progresses, and the parameters are finely adjusted when approaching the optimal solution;

[0090] During the training process, an adaptive learning rate adjustment strategy is adopted. Initially, a relatively large learning rate is set so that the model can quickly converge to near the optimal solution; as the training progresses, the learning rate gradually decays according to the cosine annealing learning rate scheduling method, and the cosine annealing learning rate scheduling formula is where η t is the learning rate of the current training epoch, η min is the minimum learning rate, η max is the maximum learning rate, T c is the current training epoch, T max is the total number of training epochs. In this way, when approaching the optimal solution, the learning rate becomes smaller, which can finely adjust the model parameters and improve the training accuracy of the model;

[0091] S45. When the trained deep neural network model is used as a predictor, different design variables are input, and the input features are weighted through the embedded attention module to highlight the key design parameters. After the non-linear transformation of the multi-layer perceptron, the fused feature information is mapped to the tower performance space, and the tower performance prediction value corresponding to the input design variables is output;

[0092] Taking the trained deep neural network model as a predictor, when different design variables are input, the input features are weighted by the embedded attention module. The attention module calculates weights based on the relevance of the input features to highlight the key design parameters. After the non-linear transformation of the multi-layer perceptron, the fused feature information is mapped through a series of neurons and activation functions. The activation function uses the ReLU (Rectified Linear Unit) function to introduce non-linear characteristics, and maps the processed feature information to the tower performance space, outputting the predicted values of the tower performance under the corresponding design variables, including strength, stiffness, stability, and fatigue life indicators.

[0093] This embodiment details the deep neural network model constructed by integrating the attention mechanism and transfer learning, which significantly improves the modeling accuracy and efficiency of the mapping relationship between the design parameters and performance of wind turbine towers. The attention module dynamically assigns feature weights, enabling the model to automatically focus on key design parameters such as tower height and material ratio, avoiding interference from secondary information. Combining with the cosine annealing learning rate strategy, it breaks through the limitations of traditional empirical formulas and single physical field models, providing an efficient and accurate digital tool for predicting the performance of towers under multi-field coupling effects, and supporting the rapid decision-making of multi-objective optimization in lightweight design.

[0094] Based on Embodiment 1, this embodiment details that in S5, when conducting multi-objective evaluation on the optimized steel-concrete composite wind turbine tower design scheme, the degree of lightweight, fatigue life, anti-wind vibration stability, and manufacturing cost are selected as the main evaluation objectives, specifically including:

[0095] The degree of lightweight is measured by the weight reduction ratio R w and the formula is: where W0 is the weight of the tower before optimization and W is the weight of the tower after optimization; the fatigue life is calculated by the Miner linear cumulative damage theory combined with the stress spectrum under the actual operating conditions of the tower, and the formula is: where N p is the fatigue life, is the number of stress cycles, is the material fatigue life at the corresponding stress level, and m p is the number of stress level types; the anti-wind vibration stability is evaluated by the wind vibration coefficient β, and the formula is: where ξ β is the pulsation amplification coefficient, ν is the pulsation influence coefficient, is the mode shape coefficient, μ s is the shape coefficient, and μ z is the wind pressure height change coefficient; the manufacturing cost C comprehensively considers factors such as material cost, processing cost, and transportation cost, and is calculated according to the actual market price and process parameters;

[0096] Construct a comprehensive evaluation function and determine the target weight w through the analytic hierarchy process j , and the formula is: where is the initial weight of the matrix, α is the balance coefficient, is the entropy weight of the j-th target, and the comprehensive evaluation value E is obtained. The formula is: where f j represents the j-th evaluation target, f1 = R w , ( is the minimum fatigue life that meets the design requirements), (C max , C min are the maximum and minimum manufacturing costs among all design schemes respectively). By comparing the comprehensive evaluation function values of different design schemes, the scheme with the largest value is selected as the final lightweight design scheme to balance the relationships of various targets and achieve multi-objective optimization.

[0097] This embodiment details the technology of multi-objective evaluation for the design scheme of a steel-concrete composite wind power tower. By accurately calculating indicators such as the degree of lightweight, fatigue life, anti-wind vibration stability, and manufacturing cost, the comprehensive evaluation function reasonably balances the relationships of various targets. The improved AHP method combined with the entropy weight method is used to determine the weight, taking into account both subjective and objective factors. Finally, the scheme with the largest comprehensive evaluation function value is selected, which can effectively achieve lightweight design, ensure the balance of the tower in terms of performance, cost, etc., and improve the practicability and economy of the design scheme.

[0098] The above are only the preferred embodiments of the present invention, and it does not limit the protection scope of the present invention. For those skilled in the art, the present invention can have various changes and modifications; all changes, modifications, substitutions, integrations, and parameter changes made to these embodiments through conventional substitutions or capable of achieving the same functions without departing from the principles and spirit of the present invention fall within the protection scope of the present invention.

Claims

1. A lightweight design method for steel-concrete composite wind turbine towers based on a topology optimization algorithm, characterized in that, Including: S1. Construct a multi - physical - field coupling model for the steel - concrete composite wind turbine tower and determine the coupling parameters; S2. Process the multi - physical - field coupling model through an adaptive multi - scale topology optimization algorithm to obtain the material layouts at the macro - scale and micro - scale. By dynamically adjusting the scale parameters, automatically switch the optimization regions between the macro - scale and micro - scale according to the structural response for multi - scale collaborative optimization; S3. Construct a material distribution strategy. In the tensile and compressive regions of the tower, determine the distribution form and orientation of steel materials and rationally allocate concrete materials; S4. Through a deep neural network model, learn the complex mapping relationship between different design parameters and the tower performance. Use the neural network model as a predictor to predict the tower performance under different design variables; S5. Conduct a multi - objective evaluation of the optimized steel - concrete composite wind turbine tower design scheme. By constructing a comprehensive evaluation function, weigh the relationships between various objectives to determine the lightweight design scheme.

2. The lightweight design method of the steel-concrete composite wind turbine tower based on the topology optimization algorithm according to claim 1, wherein Before constructing the multi - physical - field coupling model of the steel - concrete composite wind turbine tower in S1, obtain multi - source data of the multi - physical fields of the wind turbine tower; the multi - source data includes wind condition data, geological structure, and structural parameters of the steel - concrete composite wind turbine tower.

3. The lightweight design method of the steel-concrete composite wind turbine tower based on the topology optimization algorithm according to claim 2, characterized in that When constructing the multi - physical - field coupling model of the steel - concrete composite wind turbine tower in S1 by combining the obtained wind condition data, geological structure, and structural parameters of the steel - concrete composite wind turbine tower, it specifically includes: Obtain the wind farm data vector W from the wind condition data to get the wind load vector F w , and the formula is: F w = R w · W, where R w is the wind farm influence coefficient matrix; obtain the seismic response coefficient matrix R e from the geological structure to get the additional load vector F e under the seismic action, and the formula is: F e = R e · E, where E is the seismic excitation vector; obtain the comprehensive characteristic coefficient ξ of the steel-concrete composite wind power tower structure through the structural parameters of the steel-concrete composite wind power tower, and the formula is: ξ = K sc · (P s + P c ), where P s is the steel material characteristic parameter vector, P c is the concrete characteristic parameter vector, and K sc is the steel-concrete material combination coefficient matrix; The multi-physical field coupling model of the steel-concrete composite wind power tower is formed by incorporating the wind-structure coupling parameters and the earthquake-structure coupling parameters into the mechanical equilibrium equation of the tower; the wind-structure coupling parameter γ w , is determined by the formula: ; the earthquake-structure coupling parameter γ e , is determined by the formula: . Incorporate the coupling parameters into the mechanical equilibrium equation of the tower \(KU = F\) w + \(F\) e . Among them, \(K\) is the structural stiffness matrix of the tower, \(U\) is the nodal displacement vector, and the multi-physical field coupling model of the steel-concrete composite wind power tower is formed.

4. The lightweight design method of the steel-concrete composite wind turbine tower based on the topology optimization algorithm according to claim 1, characterized in that In S2, use the adaptive multi - scale topology optimization algorithm to process the above - mentioned model to obtain the material layouts at the macro - scale and micro - scale; at the macro - scale, according to the overall force characteristics and design requirements of the tower, divide larger - sized design units to determine the macro - layout of material distribution; At the micro - scale, according to the force - bearing parts and complex - structure regions of the tower, refine the design units and adjust the material distribution; by dynamically adjusting the scale parameters, automatically switch the material layout regions between the macro - scale and micro - scale according to the structural response for multi - scale collaborative optimization.

5. The lightweight design method of the steel-concrete composite wind turbine tower based on the topology optimization algorithm according to claim 4, characterized in that, For the macro - scale, by obtaining the height and diameter of the tower and adopting an adaptive hierarchical division strategy, taking the height direction of the tower as an example, divide the height intervals where stress concentration regions and connection parts are located according to a smaller height interval Δh1, and for the middle main body part with relatively uniform stress distribution, divide it according to a larger height interval Δh2. In the circumferential direction of the tower, according to the circumferential stress change curve under wind load, densely divide the regions with a large stress change gradient, and appropriately relax the division scale for other regions.

6. The lightweight design method of the steel-concrete composite wind turbine tower based on the topology optimization algorithm according to claim 4, characterized in that The macroscopic layout of the material distribution at the macroscopic scale is determined by a multi-objective optimization function O, and the formula is: O = w1·σ + w2·∈ + w3·m, where σ is the stress level, ∈ is the strain energy, m is the material usage, w1 is the stress level weight coefficient, w2 is the strain energy weight coefficient, and w3 is the material usage weight coefficient; for regions subjected to large tensile stresses, steel is preferentially laid out, and the steel distribution ratio p s is determined according to the magnitude of the tensile stress σ t in accordance with , where σ c is the compressive stress of the region, and p smax is the preset maximum steel distribution ratio; for the compression region, concrete is mainly used, and the concrete strength grade is adjusted according to the magnitude of the compressive stress σ c , and the strength grade and the compressive stress satisfy C = C min + k·(σ c - σ cmin ), C min is the lowest design strength grade, k is the strength adjustment coefficient, and σ cmin is the minimum compressive stress threshold, so as to achieve a reasonable layout of materials at the macroscopic scale.

7. The lightweight design method of the steel-concrete composite wind power tower barrel based on the topology optimization algorithm according to claim 4, characterized in that On the micro - scale of S2, for the force - bearing parts and complex - structure regions of the tower, refine the design units and adjust the material distribution, which specifically includes: Construct a microscopic digital model that maps 1:1 with the actual structure through digital twin technology. According to the direction and magnitude of the force, refine the element size to the micron level. For the tensile stress concentration area, arrange steel fibers along the direction of the microscopic stress streamline, and control the angle between the steel fiber direction and the principal tensile stress direction within a very small range. Calculate the optimal spacing d of the steel fibers through an optimization algorithm s , the formula is: where k s is the steel characteristic coefficient, E s is the elastic modulus of the steel, σ max is the maximum tensile stress, G s is the shear modulus of the steel, ρ s is the density of the steel.

8. The lightweight design method of the steel-concrete composite wind power tower barrel based on the topology optimization algorithm according to claim 1, characterized in that When constructing the material distribution strategy in S3, through biomechanical simulation, in the tensile and compressive regions of the tower, determine the distribution form and orientation of steel materials and rationally allocate concrete materials, which specifically includes: In the tensile region, the stress distribution of the tower barrel under actual working conditions is simulated by multi-scale finite element method. According to the simulation results, the distribution form and orientation of steel materials are optimized by genetic algorithm. For the parts with large axial tensile force, the steel materials are designed into a continuous spiral structure, and the spiral angle θ is determined according to the tensile force F a and the tower barrel radius r through the formula where k θ is the material property correlation coefficient, and σ θ is the yield strength of steel; In the compression area, the optimal porosity of concrete and the aggregate distribution are determined using a topology optimization algorithm, with the formula: where is the initial porosity and k is the correlation coefficient related to the properties of concrete. c ​ 9. The lightweight design method of the steel-concrete composite wind power tower barrel based on the topology optimization algorithm according to claim 1, characterized in that, The deep neural network model in S4 is constructed by fusing the attention mechanism and transfer learning, which specifically includes the following steps: S41. Collect a large amount of historical data of wind turbine towers covering different design parameters and corresponding performance indicators, conduct normalization processing, and divide it into a training set, a validation set, and a test set; S42. Construct a multi-layer perceptron as the basic architecture, embed an attention module between key layers, and adaptively focus on the design parameters that have a greater impact on the performance of the tower barrel by calculating the correlation weights between different input features; S43. Through transfer learning technology, select a pre-trained model that performs excellently in the field of large mechanical structure design, transfer the weights of some of its layers to the newly built neural network, and fine-tune the transferred model to optimize the network parameters to adapt to the tower barrel design data; S44. During the training process, adopt an adaptive learning rate adjustment strategy to converge the learning rate to near the optimal solution; the learning rate gradually decays according to the cosine function as the training progresses, and finely adjusts the parameters when approaching the optimal solution; S45. When using the trained deep neural network model as a predictor, input different design variables, perform weighted processing on the input features through the embedded attention module to highlight the key design parameters, and after the non-linear transformation of the multi-layer perceptron, map the fused feature information to the tower barrel performance space to output the predicted value of the tower barrel performance under the corresponding design variables.

10. The lightweight design method of the steel-concrete composite wind power tower barrel based on the topology optimization algorithm according to claim 1, wherein In step S5, the multi-objective evaluation selects the degree of light weight, fatigue life, anti-wind vibration stability, and manufacturing cost as the evaluation objectives, constructs a comprehensive evaluation function, weighs the relationship between each objective, and determines the light weight design scheme; When constructing the comprehensive evaluation function, the target weight w is determined by the analytic hierarchy process j , and the formula is: where is the initial weight of the matrix, α is the balance coefficient, is the entropy weight of the j-th target, and the comprehensive evaluation value E is obtained. The formula is: where f j represents the j-th evaluation target, and the solution with the maximum comprehensive evaluation value E selected by the comprehensive evaluation function is used as the lightweight design solution.

Images