A box girder structure lightweight design method based on topology optimization

Abstract

The application relates to the technical field of structural optimization design, and specifically discloses a box beam structure lightweight design method based on topology optimization, which comprises the following steps: a three-dimensional parameterized model of multi-physical field coupling is constructed; cross-scale analysis of structural stiffness, thermal deformation, vibration mode and fatigue life is integrated; a variable density method and a level set method mixed topology optimization strategy is adopted; the relationship between density and stress and strain is modeled; and an initial material distribution scheme meeting the structural functional integrity is generated; an intelligent optimization engine of quantum genetic algorithm and deep reinforcement learning fusion is started; through the three-dimensional parameterized model of multi-physical field coupling, the variable density method and the level set method mixed topology optimization strategy are combined, the relationship between the structural lightweight target and the additive manufacturing process is effectively balanced, multi-physical field analysis such as thermal response, vibration and fatigue is introduced, and the box beam can meet the mechanical performance and other functional requirements in actual application.

Description

Technical Field

[0001] This invention belongs to the field of structural optimization design technology, specifically relating to a lightweight design method for box girder structures based on topology optimization. Background Technology

[0002] With the continuous development of engineering technology, especially the widespread application of additive manufacturing technology, traditional structural design methods are proving inadequate in dealing with complex mechanical properties and production process requirements. Box girders, as a common structural form, are widely used in construction, transportation and other fields. Their design requirements not only need to meet basic mechanical properties, such as strength and stiffness, but also need to consider the multi-physics field effects that the structure may encounter in actual use, including thermal response, vibration, fatigue and other issues.

[0003] In traditional design methods, designers often focus on optimizing a single physics field, such as considering only mechanical properties or thermal response, neglecting the interactions between multiple physics fields. This can lead to design results that fail to achieve the expected overall performance in practical applications. Furthermore, with the maturity of additive manufacturing technology, it offers unique advantages over traditional manufacturing processes, such as enabling free design of complex structures and reducing material waste. However, additive manufacturing also presents new challenges, especially in the structural design stage. Unlike traditional manufacturing processes, additive manufacturing requires consideration of specific process constraints such as material deposition sequence, temperature distribution, minimum feature size, and supporting structures. These constraints make it difficult for traditional design methods to meet the specific requirements of additive manufacturing, potentially preventing the successful transformation of design results into a practically producible model.

[0004] Therefore, it is necessary to propose a lightweight design method for box girder structures based on topology optimization to solve the problem of how to balance the relationship between the lightweight goal of the structure and the manufacturing process in the existing technology.

[0005] The information disclosed above in this background section is only for enhancing the understanding of the background section of this invention, and therefore may include prior art that is not known to those skilled in the art. Summary of the Invention

[0006] The purpose of this invention is to provide a lightweight design method for box girder structures based on topology optimization, so as to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution:

[0008] A lightweight design method for box girder structures based on topology optimization includes:

[0009] A three-dimensional parametric model with multi-physics coupling is constructed. Through cross-scale analysis of structural stiffness, thermal deformation, vibration modes and fatigue life, and a hybrid topology optimization strategy of variable density method and level set method is adopted, the relationship between density and stress-strain is modeled to generate an initial material distribution scheme that satisfies the structural functional integrity.

[0010]

[0011] In the formula, ρ is the material density distribution, J(ρ) is the objective function, σ(ρ) is the stress tensor of the material, ε(ρ) is the strain tensor of the material, and Ω is the design domain;

[0012] An intelligent optimization engine integrating quantum genetic algorithm and deep reinforcement learning is launched to dynamically optimize the initial material distribution scheme. By adjusting the optimization target weight coefficient and combining it with manufacturing process constraints, the optimal topology configuration that meets the requirements of additive manufacturing process is generated.

[0013] Deploy a digital twin-driven closed-loop verification system to trigger an iterative correction mechanism for optimization algorithms by synchronizing stress-strain data between the physical prototype and the virtual model in real time, thereby enabling bidirectional adaptive compensation for design parameters and manufacturing errors.

[0014] The blockchain evidence storage module is invoked to perform tamper-proof distributed storage of key parameters, simulation results, and version iteration records during the optimization process.

[0015] Preferably, the multiphysics coupling model is constructed in the following manner:

[0016] Use isogeometric analysis techniques to perform bidirectional mapping of geometric parameters between CAD and CAE models;

[0017] A thermo-mechanical coupled finite element scheme is introduced, and the nonlinearity of the contact interface is handled by the substructure condensation method.

[0018] Thermo-mechanical coupling finite element formula:

[0019] [K(χ)+K thermal (χ)]{u}={F}

[0020] In the formula, K(χ) is the structural stiffness matrix, K thermal (χ) is the thermo-mechanical coupling stiffness matrix, u is the nodal displacement vector, and F is the external load;

[0021] A fatigue life prediction network based on transfer learning was constructed, and small sample experimental data was used to correct the deviation of simulation results.

[0022] Preferably, the intelligent optimization engine achieves dynamic collaboration through the following mechanism:

[0023] The quantum genetic algorithm is initialized by encoding topological variables with qubits and performing a global search through quantum rotation gate operations.

[0024] Quantum Revolving Door Operation Formula:

[0025] |Ψ new >=U(θ)|Ψ old >

[0026] In the formula, |Ψ new >and|Ψ old > is the quantum state vector, U(θ) is the quantum rotating door material matrix, and θ is the angle controlling the rotation;

[0027] Configure a deep reinforcement learning agent based on the Actor-Critic framework, using structural performance metrics as the reward function, and dynamically adjust the optimization path;

[0028] R(s,a)=ω1·P(s,a)-ω2·MC(s,a)

[0029] In the formula, P(s,a) is the structural performance index, MC(s,a) is the manufacturing cost, and ω1 and ω2 are weighting coefficients;

[0030] Embedded manufacturing process constraints are used to correct the fitness evaluation through a penalty function mechanism, including control of minimum feature size, draft angle, and overhang angle.

[0031] Preferably, the operation process of the closed-loop verification system includes:

[0032] Deploy a physical prototype with an embedded fiber Bragg grating sensor network to collect strain-temperature data at key measurement points in real time;

[0033] Construct a virtual sensor network based on digital twins and predict the response status of unmonitored areas using LSTM neural networks;

[0034] The improved Kalman filter algorithm is used to fuse virtual and real data, and the parameters of the optimization engine are readjusted when the error threshold is triggered.

[0035] x′ k|k =x′ k|k-1 +K k (z k -H k x′ k|k-1 )

[0036] In the formula, x′ k|k This is the updated state estimate, K k It is the Kalman gain, z k It is a measured value, H k It is a measurement matrix.

[0037] Preferably, the blockchain evidence storage module is implemented through smart contracts:

[0038] When a design version changes, a hash value calculation is automatically triggered, and the result is stored in the blockchain;

[0039] Access to key design parameters is controlled through asymmetric encryption mechanisms;

[0040] The evidence data complies with the ISO 19770-1 standard, enabling the tracing of design responsibility and the proof of intellectual property rights.

[0041] Preferably, the topology generation process includes:

[0042] A topology innovation design module based on generative adversarial networks is constructed to generate unconventional topology configurations through style transfer technology;

[0043] Loss function for generative adversarial networks:

[0044] L GAN =E[log(D(r))]+E[log(1-D(G(z)))]

[0045] In the formula, D(r) is the discriminator, G(z) is the generator, z is random noise, and L... GAN It is the loss of generative adversarial networks;

[0046] A multi-fidelity model fusion technique is adopted, combining coarse-grained proxy models with fine finite element analysis, to balance computational efficiency and solution accuracy.

[0047] We introduce manufacturing defect sensitivity analysis and use Monte Carlo simulation to quantify the impact of porosity and interlayer bonding strength on structural performance.

[0048]

[0049] In the formula, Sy is the sensitivity, and P i This is the i-th simulation result, α i These are manufacturing defect parameters.

[0050] Preferably, the method further includes:

[0051] We constructed a materials-structure co-optimization module driven by a materials gene library, and screened composite material systems that meet multiple objective constraints through high-throughput computation.

[0052] Develop a WebGL-based 3D visualization interface for designers to perform real-time editing and performance pre-evaluation of topology configurations;

[0053] Deploy edge computing nodes to perform distributed parallel processing of the design-analysis-optimization process, shortening the design cycle.

[0054] Compared with the prior art, the beneficial effects of the present invention are:

[0055] This invention effectively balances the relationship between structural lightweighting and additive manufacturing processes by employing a multi-physics coupled three-dimensional parametric model and a hybrid topology optimization strategy combining the variable density method and the level set method. The design process incorporates multi-physics analysis of thermal response, vibration, and fatigue to ensure that the box girder meets mechanical performance and other functional requirements in practical applications. Furthermore, an intelligent optimization engine combining quantum genetic algorithms and deep reinforcement learning dynamically adjusts weight coefficients and incorporates manufacturing process constraints during optimization, ensuring that the optimization results meet the specific needs of additive manufacturing and thus avoiding the limitations of traditional design methods. In addition, a digital twin-driven closed-loop verification system synchronizes stress-strain data between the physical prototype and the virtual model in real time, achieving bidirectional adaptive compensation for design parameters and manufacturing errors, thereby improving the accuracy and reliability of the design. Attached Figure Description

[0056] Figure 1 This is a flowchart of the lightweight design method for box girder structures based on topology optimization according to the present invention. Detailed Implementation

[0057] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0058] Example 1:

[0059] Please see Figure 1 As shown, a lightweight design method for box girder structures based on topology optimization includes:

[0060] A three-dimensional parametric model with multi-physics coupling is constructed. Through cross-scale analysis of structural stiffness, thermal deformation, vibration modes and fatigue life, and a hybrid topology optimization strategy of variable density method and level set method is adopted, the relationship between density and stress-strain is modeled to generate an initial material distribution scheme that satisfies the structural functional integrity.

[0061] The multiphysics coupling model is constructed in the following way:

[0062] Use isogeometric analysis techniques to perform bidirectional mapping of geometric parameters between CAD and CAE models;

[0063] A thermo-mechanical coupled finite element scheme is introduced, and the nonlinearity of the contact interface is handled by the substructure condensation method.

[0064] A fatigue life prediction network based on transfer learning was constructed, and small sample experimental data was used to correct the deviation of simulation results.

[0065] Furthermore, through multiphysics coupling and cross-scale analysis, the overall performance of the structure is more comprehensively evaluated and optimized, improving the accuracy and reliability of the design. A hybrid topology optimization strategy effectively allocates materials, improves the functional integrity of the structure, and reduces redundant materials. By introducing thermo-mechanical coupled finite element analysis and fatigue life prediction through transfer learning, complex nonlinear problems are solved, thereby achieving a more efficient and stable design scheme.

[0066] An intelligent optimization engine integrating quantum genetic algorithm and deep reinforcement learning is launched to dynamically optimize the initial material distribution scheme. By adjusting the optimization target weight coefficient and combining it with manufacturing process constraints, the optimal topology configuration that meets the requirements of additive manufacturing process is generated.

[0067] The intelligent optimization engine achieves dynamic collaboration through the following mechanisms:

[0068] The quantum genetic algorithm is initialized by encoding topological variables with qubits and performing a global search through quantum rotation gate operations.

[0069] Configure a deep reinforcement learning agent based on the Actor-Critic framework, using structural performance metrics as the reward function, and dynamically adjust the optimization path;

[0070] Embedded manufacturing process constraints are used to correct the fitness evaluation through a penalty function mechanism, including control of minimum feature size, draft angle, and overhang angle.

[0071] Furthermore, by integrating quantum genetic algorithms with deep reinforcement learning, dynamic optimization of the initial material distribution scheme is achieved, improving the accuracy and efficiency of topology design. The combination of the global search capability of quantum genetic algorithms and the adaptive optimization capability of deep reinforcement learning effectively improves structural performance while adjusting the optimization path according to manufacturing process constraints, ensuring that the design meets the requirements of additive manufacturing. By embedding manufacturing process constraints, the fitness evaluation is corrected in real time during the optimization process, avoiding designs that do not meet actual production requirements, ultimately generating a more efficient and manufacturing-compliant optimal topology.

[0072] The topology generation process includes:

[0073] A topology innovation design module based on generative adversarial networks is constructed to generate unconventional topology configurations through style transfer technology;

[0074] A multi-fidelity model fusion technique is adopted, combining coarse-grained proxy models with fine finite element analysis, to balance computational efficiency and solution accuracy.

[0075] We introduce manufacturing defect sensitivity analysis and use Monte Carlo simulation to quantify the impact of porosity and interlayer bonding strength on structural performance.

[0076] Furthermore, by employing generative adversarial networks and style transfer techniques, unconventional topologies were innovatively designed, enhancing design diversity and performance optimization potential. Multi-fidelity model fusion technology optimized the balance between computational efficiency and accuracy, ensuring rapid and accurate design analysis. Simultaneously, manufacturing defect sensitivity analysis combined with Monte Carlo simulation quantified the impact of porosity and interlayer bonding strength on structural performance, thereby enhancing the reliability and stability of the design scheme in actual manufacturing.

[0077] Deploy a digital twin-driven closed-loop verification system to trigger an iterative correction mechanism for optimization algorithms by synchronizing stress-strain data between the physical prototype and the virtual model in real time, thereby enabling bidirectional adaptive compensation for design parameters and manufacturing errors.

[0078] The operation process of the closed-loop verification system includes:

[0079] Deploy a physical prototype with an embedded fiber Bragg grating sensor network to collect strain-temperature data at key measurement points in real time;

[0080] Construct a virtual sensor network based on digital twins and predict the response status of unmonitored areas using LSTM neural networks;

[0081] The improved Kalman filter algorithm is used to fuse virtual and real data, and the parameters of the optimization engine are readjusted when the error threshold is triggered.

[0082] Furthermore, the digital twin-driven closed-loop verification system precisely monitors design parameters and manufacturing errors by synchronizing data from the physical prototype and virtual model in real time, dynamically adjusting optimization algorithms to achieve adaptive compensation in design and manufacturing. The combination of embedded fiber optic sensor networks and virtual sensor networks enhances real-time monitoring capabilities of key measurement points, while the application of LSTM neural networks and improved Kalman filtering algorithms enhances the accuracy of data fusion, enabling rapid response and optimization adjustments when errors exceed thresholds, ensuring the accuracy and reliability of product design.

[0083] The blockchain evidence storage module is invoked to perform tamper-proof distributed storage of key parameters, simulation results, and version iteration records during the optimization process.

[0084] The blockchain-based evidence storage module is implemented through smart contracts:

[0085] When a design version changes, a hash value calculation is automatically triggered, and the result is stored in the blockchain;

[0086] Access to key design parameters is controlled through asymmetric encryption mechanisms;

[0087] The evidence data complies with the ISO 19770-1 standard, enabling the tracing of design responsibility and the proof of intellectual property rights.

[0088] We constructed a materials-structure co-optimization module driven by a materials gene library, and screened composite material systems that meet multiple objective constraints through high-throughput computation.

[0089] Develop a WebGL-based 3D visualization interface for designers to perform real-time editing and performance pre-evaluation of topology configurations;

[0090] Deploy edge computing nodes to perform distributed parallel processing of the design-analysis-optimization process, shortening the design cycle.

[0091] Furthermore, by constructing a material-structure co-optimization module driven by a material gene library, composite material systems that meet multiple objective constraints can be efficiently screened, thereby improving the accuracy of material selection and design efficiency. A WebGL-based 3D visualization interface allows designers to edit topology configurations and perform performance pre-evaluation in real time, enhancing the flexibility and operability of the design process. Deploying edge computing nodes for distributed parallel processing significantly shortens the design cycle, improves the computational efficiency of the design-analysis-optimization process, and accelerates the implementation of innovative results.

[0092] Example 2:

[0093] Application Example: Lightweighting Research of Box Girder Structures Based on Topology Optimization

[0094] In the manufacturing of high-end equipment such as aerospace and large bridges, box girders, as key load-bearing structures, face the dual challenge of achieving structural lightweighting to reduce overall weight and improve energy efficiency, while ensuring the feasibility and economy of manufacturing processes. Traditional design methods often focus on optimizing single performance indicators, making it difficult to find the optimal balance between lightweighting and manufacturing processes. Therefore, this project aims to effectively solve this technical challenge through a lightweight design method for box girder structures based on topology optimization.

[0095] I. Implementation of the Technical Solution

[0096] (1) Construction of multiphysics coupling model, laying the foundation for lightweight design.

[0097] By utilizing isogeometric analysis techniques, a high-precision mapping between the CAD model and CAE model of the box girder is achieved, ensuring that the structural form can be fully considered from the initial design stage.

[0098] By introducing thermo-mechanical coupled finite element analysis and combining it with the substructure condensation method to handle complex contact interface problems, a multi-physics coupled model is constructed to provide a comprehensive performance evaluation basis for lightweight design.

[0099] A hybrid topology optimization strategy combining the variable density method and the level set method was adopted to integrate structural stiffness, thermal deformation, vibration mode and fatigue life analysis, and generate an initial material distribution scheme that meets the basic functional requirements, laying the foundation for subsequent lightweight design.

[0100] (2) Intelligent optimization engine drives precise balance between lightweighting and manufacturing processes

[0101] We initialize a quantum genetic algorithm, use qubits to encode topological variables, and combine quantum rotation gate operations to perform efficient global search, rapidly exploring the possible space for lightweight design.

[0102] A deep reinforcement learning agent is configured based on the Actor-Critic framework. The reward function is a comprehensive indicator of structural performance and manufacturing cost. The optimization path is dynamically adjusted to ensure that the feasibility and economy of the manufacturing process are not ignored while pursuing lightweight design.

[0103] By embedding manufacturing process constraints, such as minimum feature size, draft angle, and overhang angle, and correcting the fitness evaluation through a penalty function mechanism, the optimization results are ensured to meet the requirements of advanced manufacturing processes such as additive manufacturing, thus achieving a precise balance between lightweighting and manufacturing processes.

[0104] (3) Closed-loop verification system to ensure the practical feasibility of lightweight design.

[0105] Deploy a physical prototype with an embedded fiber Bragg grating sensor network to collect strain-temperature data at key measurement points in real time, providing direct feedback on the actual effect of lightweight design.

[0106] A virtual sensor network based on digital twins is constructed, and an LSTM neural network is used to predict the response status of unmonitored areas, thereby achieving deep fusion of virtual and real data.

[0107] By using an improved Kalman filter algorithm, the difference between virtual and real data is analyzed in real time. When the error exceeds a preset threshold, the parameter readjustment mechanism of the optimization engine is automatically triggered to ensure the feasibility and accuracy of lightweight design in the actual manufacturing process.

[0108] (4) Blockchain-based evidence storage to protect the intellectual property rights of lightweight designs

[0109] When a design version changes, a hash value calculation is automatically triggered, and the result is stored in the blockchain to ensure the transparency and traceability of the lightweight design process.

[0110] By employing asymmetric encryption mechanisms, access permissions to key design parameters are strictly controlled, thus protecting the core intellectual property rights of lightweight designs.

[0111] II. Topology generation and evaluation to optimize lightweighting effects

[0112] The topology innovation design module of generative adversarial networks is launched, which generates diverse and unconventional topology configurations through style transfer technology, providing more possibilities for lightweight design.

[0113] By employing multi-fidelity model fusion technology, combining coarse-grained proxy models with fine finite element analysis, we can efficiently evaluate the lightweighting effect and manufacturing feasibility of different topological configurations, ensuring that the final selected topological configuration achieves the best balance between lightweighting and manufacturing process.

[0114] III. Material-structure synergistic optimization to improve the overall performance of lightweight design

[0115] A material-structure co-optimization module driven by a material gene library is constructed. Through high-throughput computing, composite material systems that meet both lightweight requirements and are suitable for manufacturing processes are screened out to achieve the best match between materials and structures.

[0116] Develop a WebGL-based 3D visualization interface to support designers in real-time editing and performance pre-evaluation of topology configurations and material selection, thereby improving the flexibility and accuracy of lightweight design.

[0117] IV. Shortened design cycle and parallel processing accelerate the implementation of lightweight design.

[0118] Deploying edge computing nodes enables distributed parallel processing of the design-analysis-optimization process, significantly shortening the lightweight design cycle and accelerating the transformation of innovative achievements into actual products.

[0119] V. Conclusion

[0120] By implementing the aforementioned lightweight design method for box girder structures based on topology optimization, this project successfully found the optimal balance between lightweight objectives and manufacturing processes, designing a box girder structure that meets both multiphysics performance requirements and advanced manufacturing processes. This method not only improves design efficiency and accuracy but also ensures the practical feasibility of lightweight design and intellectual property protection, providing strong technical support for the high-end equipment manufacturing field.

[0121] Example 3:

[0122] This invention also provides a computer-readable storage medium storing a program for a lightweight design method for box girder structures based on topology optimization, as described in any of the above embodiments. When executed by a processor, this program implements the various processes of the lightweight design method embodiments and achieves the same technical effects. To avoid repetition, further details are omitted here. The computer-readable storage medium may include read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk.

[0123] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Furthermore, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0124] The accompanying drawings of the embodiments disclosed in this invention only involve the structures involved in the embodiments disclosed in this invention. Other structures can refer to general designs. In the absence of conflict, the same embodiment and different embodiments of this invention can be combined with each other.

[0125] The flowchart shown in the attached diagram is for illustrative purposes only and does not necessarily include all content and operations / steps, nor does it necessarily have to be performed in the order described. For example, some operations / steps can be broken down, combined, or partially merged, so the actual execution order may change depending on the actual situation.

[0126] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A lightweight design method for box girder structures based on topology optimization, characterized in that, include: A three-dimensional parametric model with multi-physics coupling is constructed. Through cross-scale analysis of structural stiffness, thermal deformation, vibration modes and fatigue life, and a hybrid topology optimization strategy of variable density method and level set method is adopted, the relationship between density and stress-strain is modeled to generate an initial material distribution scheme that satisfies the structural functional integrity. ； In the formula, It is the material density distribution. It is the objective function. It is the stress tensor of the material. It is the strain tensor of the material. It is a design domain; An intelligent optimization engine integrating quantum genetic algorithm and deep reinforcement learning is launched to dynamically optimize the initial material distribution scheme. By adjusting the optimization target weight coefficient and combining it with manufacturing process constraints, the optimal topology configuration that meets the requirements of additive manufacturing process is generated. Deploy a digital twin-driven closed-loop verification system to trigger an iterative correction mechanism for optimization algorithms by synchronizing stress-strain data between the physical prototype and the virtual model in real time, thereby enabling bidirectional adaptive compensation for design parameters and manufacturing errors. The blockchain evidence storage module is invoked to perform tamper-proof distributed storage of key parameters, simulation results, and version iteration records during the optimization process; The operation process of the closed-loop verification system includes: Deploy a physical prototype with an embedded fiber Bragg grating sensor network to collect strain-temperature data at key measurement points in real time; Construct a virtual sensor network based on digital twins and predict the response status of unmonitored areas using LSTM neural networks; The improved Kalman filter algorithm is used to fuse virtual and real data, and the parameters of the optimization engine are readjusted when the error threshold is triggered. ； In the formula, This is the updated state estimate. It is Kalman gain. These are measured values. It is a measurement matrix.

2. The lightweight design method for box girder structures based on topology optimization according to claim 1, characterized in that, The multiphysics coupling model is constructed in the following way: Use isogeometric analysis techniques to perform bidirectional mapping of geometric parameters between CAD and CAE models; A thermo-mechanical coupled finite element scheme is introduced, and the nonlinearity of the contact interface is handled by the substructure condensation method. Thermo-mechanical coupling finite element formula: ； In the formula, It is the structural stiffness matrix. It is the thermo-mechanical coupling stiffness matrix. It is a nodal displacement vector. It is an external load; A fatigue life prediction network based on transfer learning was constructed, and small sample experimental data was used to correct the deviation of simulation results.

3. The lightweight design method for box girder structures based on topology optimization according to claim 2, characterized in that, The intelligent optimization engine achieves dynamic collaboration through the following mechanism: The quantum genetic algorithm is initialized by encoding topological variables with qubits and performing a global search through quantum rotation gate operations. Quantum Revolving Door Operation Formula: ； In the formula, and It is a quantum state vector. It is a quantum revolving door material matrix. It controls the angle of rotation; Configure a deep reinforcement learning agent based on the Actor-Critic framework, using structural performance metrics as the reward function, and dynamically adjust the optimization path; ； In the formula, It is a structural performance indicator. It is the manufacturing cost. and These are weighting coefficients; Embedded manufacturing process constraints are used to correct the fitness evaluation through a penalty function mechanism, including control of minimum feature size, draft angle, and overhang angle.

4. The lightweight design method for box girder structures based on topology optimization according to claim 1, characterized in that, The blockchain-based evidence storage module is implemented through smart contracts: When a design version changes, a hash value calculation is automatically triggered, and the result is stored in the blockchain; Access to key design parameters is controlled through asymmetric encryption mechanisms; The evidence data complies with the ISO 19770-1 standard, enabling the tracing of design responsibility and the proof of intellectual property rights.

5. The lightweight design method for box girder structures based on topology optimization according to claim 4, characterized in that, The topology generation process includes: A topology innovation design module based on generative adversarial networks is constructed to generate unconventional topology configurations through style transfer technology; Loss function for generative adversarial networks: ； In the formula, It is a discriminator. It is a generator. It is random noise. It is the loss of generative adversarial networks; A multi-fidelity model fusion technique is adopted, combining coarse-grained proxy models with fine finite element analysis, to balance computational efficiency and solution accuracy. We introduce manufacturing defect sensitivity analysis and use Monte Carlo simulation to quantify the impact of porosity and interlayer bonding strength on structural performance. ； In the formula, It's about sensitivity. It is the first One simulation result, These are manufacturing defect parameters.

6. The lightweight design method for box girder structures based on topology optimization according to claim 5, characterized in that, The method further includes: We constructed a materials-structure co-optimization module driven by a materials gene library, and screened composite material systems that meet multiple objective constraints through high-throughput computation. Develop a WebGL-based 3D visualization interface for designers to perform real-time editing and performance pre-evaluation of topology configurations; Deploy edge computing nodes to perform distributed parallel processing of the design-analysis-optimization process, shortening the design cycle.

Images