A method and system for intelligently generating bridge section design diagrams

Abstract

The embodiment of the present invention discloses a method and system for intelligently generating a bridge section design diagram, and relates to the technical field of bridge structure design. The method can solve the problem of low efficiency caused by the reliance on manual adjustment of traditional section design methods. The method comprises: constructing a bridge numerical model based on a completed bridge, and obtaining a data set of design parameters and corresponding design conditions based on the bridge numerical model; generating an original bridge section design diagram based on the bridge section design parameters in the data set, and generating conditional characteristic data based on the design conditions; constructing a bridge section intelligent generation model based on the CF‑VAE‑GAN algorithm, and training the bridge section intelligent generation model based on the original bridge section design diagram and conditional characteristic data; inputting the target conditional characteristic data into the trained bridge section intelligent generation model to obtain a bridge section design diagram for the target design conditions. The present invention is applicable to beam structure design scenarios.

Description

Technical Field

[0001] The present invention relates to the technical field of bridge structure design, and in particular to a method and system for intelligently generating a bridge cross-section design diagram. Background Art

[0002] In the field of bridge engineering, the application of steel-concrete composite beam bridges is increasing. However, with the continuous improvement of span and load requirements in bridge design, the accuracy and efficiency of bridge section design have also been challenged. Traditional section design methods mainly rely on empirical formulas and complex finite element calculation software. This method is not only inefficient when manually adjusting design parameters, but also highly dependent on the professional knowledge and experience of engineers, which to some extent limits the innovation and progress of bridge design technology.

[0003] The rapid development of deep learning technology and the popularization of high-performance hardware have greatly promoted the overall innovation in the field of artificial intelligence. In particular, intelligent algorithms that can achieve rapid reasoning and design in engineering practice have injected new vitality into industrial construction fields such as civil engineering. However, in the field of intelligent design of bridge engineering, especially the cross-sectional design of steel-concrete composite beam bridges, there is still a significant research gap. Summary of the invention

[0004] In view of this, an embodiment of the present invention provides a method and system for intelligently generating a bridge section design diagram to solve the problem of low efficiency caused by the traditional section design method relying on manual adjustment.

[0005] In a first aspect, an embodiment of the present invention provides a method for intelligently generating a bridge section design diagram, comprising: constructing a bridge numerical model based on a completed bridge, and obtaining a data set of design parameters and corresponding design conditions based on the bridge numerical model; generating an original bridge section design diagram based on the bridge section design parameters in the data set, and generating conditional characteristic data based on the design conditions; constructing a bridge section intelligent generation model based on the CF-VAE-GAN algorithm, and training the bridge section intelligent generation model based on the original bridge section design diagram and the conditional characteristic data; inputting the target conditional characteristic data into the trained bridge section intelligent generation model to obtain a bridge section design diagram for the target design conditions.

[0006] Optionally, the bridge section design parameters include: mid-span bridge deck size, mid-span steel box girder size, support bridge deck size and support steel box girder size; the design conditions include: construction method, number of lanes, lateral distribution coefficient, temperature change, ambient humidity and vehicle load level; the bridge section design parameters match the design conditions.

[0007] Optionally, the method of constructing a bridge numerical model based on a completed bridge and obtaining a data set of design parameters and corresponding design conditions based on the bridge numerical model includes: obtaining bridge section design parameters of the completed bridge and design conditions corresponding to the section design parameters to construct a bridge numerical model; the bridge numerical model is used to simulate the structural performance of the bridge under different design conditions; using the bridge numerical model to perform simulations, and adjusting the bridge section design parameters so that the bridge structural performance can meet the requirements of different design conditions; generating a data set of design parameters and corresponding design conditions based on the simulation results; the data set is used to record the bridge section design parameters and bridge structural performance under different design conditions.

[0008] Optionally, generating an original bridge section design schematic based on the bridge section design parameters in the data set, and generating conditional characteristic data based on the design conditions, include: using a matplotlib library to generate an original bridge section design schematic according to the bridge section design parameters; wherein the original bridge section design schematic is a symmetrical half-section design schematic; using a data processing tool to convert the design conditions into conditional characteristic data of a predetermined dimension; wherein the data format of the conditional characteristic data is a data format that can be recognized and processed by a computer.

[0009] Optionally, the bridge section intelligent generation model based on the CF-VAE-GAN algorithm includes: an encoder module, a generator module, a discriminator module and the similarity discriminator module; the encoder module includes an image encoder and a distribution encoder; the image encoder is used to extract key features from the input original bridge section design diagram and output the feature tensor of the key features; the distribution encoder is used to compress the input feature tensor and conditional feature data into low-dimensional latent variables; the generator module is used to use multiple network modules to process the low-dimensional latent variables and the conditional feature data to generate a new bridge section design diagram; the discriminator module is used to generate respective discriminant feature tensors based on the new bridge section design diagram and the original bridge section design diagram in combination with the conditional feature data to distinguish the new bridge section design diagram from the original bridge section design diagram; the similarity discriminator module is used to discriminate based on the feature tensors obtained by the image encoder from the new bridge section design diagram and the original bridge section design diagram, and evaluate the similarity between the new bridge section design diagram and the original bridge section design diagram.

[0010] Optionally, the bridge section intelligent generation model based on the CF-VAE-GAN algorithm is constructed, and the bridge section intelligent generation model is trained based on the original bridge section design schematic and the conditional feature data, including: inputting the original bridge section design schematic and the conditional feature data into the bridge section intelligent generation model, processing them through the encoder module to obtain latent variables; inputting the latent variables and the conditional feature data into the generator module to obtain a new bridge section design schematic; optimizing the bridge section intelligent generation model based on the discriminator and similarity discriminator modules to obtain a trained bridge section intelligent generation model.

[0011] Optionally, the objective function of the CF-VAE-GAN algorithm is: ;

[0012] Among them, λ V and λ C is a hyperparameter that controls the impact of the loss term;

[0013] V(G min ,D max ) is the adversarial loss between the generator G and the discriminator D, expressed as V G min , D max = E x~ P data (x) log D(x|y) + E x~ P z(z) log [1-D(G(z|y))] ; x is the original bridge section design diagram, y is the conditional characteristic data, and z is the hidden variable;

[0014] L C is the loss of the similarity discriminator C, expressed as ;f x f is the feature tensor obtained by inputting the original bridge section design diagram into the image encoder, x 'The generated new bridge cross-section diagram is re-input into the image encoder to obtain a new feature tensor;

[0015] D KL represents the KL divergence loss, the purpose is to make the encoding distribution as close to the standard normal distribution as possible, and its expression is ;

[0016] N(μ,σ) is an independent Gaussian distribution, μ is the mean and σ is the variance.

[0017] In a second aspect, the present invention also provides an embodiment of a bridge section design schematic intelligent generation system, including: a data acquisition module, used to construct a bridge numerical model based on a completed bridge, and obtain a data set of design parameters and corresponding design conditions based on the bridge numerical model; a data processing module, used to generate an original bridge section design schematic based on the bridge section design parameters in the data set, and generate conditional characteristic data based on the design conditions; a model training module, used to construct a bridge section intelligent generation model based on the CF-VAE-GAN algorithm, and train the bridge section intelligent generation model based on the original bridge section design schematic and conditional characteristic data; a design schematic generation module, used to input the target conditional characteristic data into the trained bridge section intelligent generation model to obtain a bridge section design schematic for the target design conditions.

[0018] In a third aspect, the present invention also provides an electronic device in an embodiment, comprising: a housing, a processor, a memory, a circuit board and a power supply circuit, wherein the circuit board is arranged inside the space enclosed by the housing, and the processor and the memory are arranged on the circuit board; the power supply circuit is used to supply power to various circuits or devices of the above-mentioned electronic device; the memory is used to store executable program code; the processor runs a program corresponding to the executable program code by reading the executable program code stored in the memory, so as to execute any of the intelligent generation methods of bridge section design schematics described in the first aspect.

[0019] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the method for intelligently generating bridge section design schematics as described in any one of the first aspects.

[0020] The embodiment of the present invention provides a method and system for intelligently generating a bridge section design diagram, which constructs a bridge numerical model based on a completed bridge, obtains a data set of design parameters and corresponding design conditions based on the bridge numerical model; generates an original bridge section design diagram based on the bridge section design parameters, and generates conditional characteristic data based on the design conditions; and uses a bridge section intelligent generation model based on the CF-VAE-GAN algorithm to generate a bridge section design diagram for target design conditions according to the rules between the conditional characteristic data and the bridge section design diagram. In this way, a bridge section design diagram for target design conditions is intelligently generated by the bridge section intelligent generation model to solve the problem of low efficiency caused by the traditional section design method relying on manual adjustment. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0022] Figure 1 A schematic diagram of a process flow of a method for intelligently generating a bridge cross-section design diagram provided by an embodiment of the present invention;

[0023] Figure 2 This is a schematic diagram of a method flow of step S110 in an embodiment of the present invention;

[0024] Figure 3 This is a schematic diagram of a method flow of step S120 in an embodiment of the present invention;

[0025] Figure 4 This is a schematic diagram of a method flow of step S130 in an embodiment of the present invention;

[0026] Figure 5 A schematic diagram of a method for intelligently generating a bridge cross-section design diagram provided by an embodiment of the present invention;

[0027] Figure 6 A schematic diagram of an image encoder provided by an embodiment of the present invention;

[0028] Figure 7 A schematic diagram of the intelligent generation system architecture of a bridge section design diagram provided by an embodiment;

[0029] Figure 8 The present invention is a schematic block diagram of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0030] The embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0031] It should be clear that the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0032] Embodiment 1

[0033] See also Figure 1 As shown, the method for intelligently generating a bridge section design diagram provided by an embodiment of the present invention comprises the steps of:

[0034] S110, constructing a bridge numerical model based on the completed bridge, and acquiring a data set of design parameters and corresponding design conditions based on the bridge numerical model.

[0035] The embodiment of the present invention first obtains bridge data of the completed bridges based on the actual design and construction experience of the existing completed bridges, that is, the bridge section design parameters and the design conditions corresponding to the section design parameters to construct a data set. However, due to the small number of existing completed bridges and the limited data provided, the embodiment of the present invention also establishes a bridge numerical model based on these data, and continuously adjusts the parameters in the bridge numerical model to ensure that the structural performance of the bridge can meet different design conditions, and obtains a sufficient number of data sets of bridge section design parameters and corresponding design conditions based on the simulation results.

[0036] Optionally, in some embodiments, the bridge section design parameters include: mid-span bridge deck size, mid-span steel box girder size, support bridge deck size and support steel box girder size; the design conditions include: construction method, number of lanes, lateral distribution coefficient, temperature change, ambient humidity and vehicle load level; the bridge section design parameters match the design conditions.

[0037] Optionally, in some embodiments, see Figure 2 As shown, step S110 builds a bridge numerical model based on the completed bridge, and obtains a data set of design parameters and corresponding design conditions based on the bridge numerical model. The specific steps are as follows:

[0038] S111, obtaining the bridge section design parameters of the completed bridge and the design conditions corresponding to the section design parameters to construct a bridge numerical model; the bridge numerical model is used to simulate the structural performance of the bridge under different design conditions;

[0039] S113, using the bridge numerical model to perform simulation, and adjusting the bridge cross-section design parameters so that the bridge structure performance can meet the requirements of different design conditions;

[0040] S115. Generate a data set of design parameters and corresponding design conditions based on the simulation results; the data set is used to record the bridge section design parameters and bridge structure performance under different design conditions.

[0041] In this embodiment, bridge section design parameters and design conditions corresponding to the section design parameters are obtained based on existing completed bridges, and a bridge numerical model for simulating the structural performance of bridges under different design conditions is constructed based on these bridge section design parameters and design conditions. By using the bridge numerical model, the bridge section design parameters are continuously adjusted so that the structural performance of the bridge can meet the design requirements under various design conditions, wherein the bridge section design parameters include key information such as the size of the mid-span deck bridge, the size of the mid-span steel box girder, the size of the bearing bridge deck, and the size of the bearing steel box girder; the design conditions include the construction method (such as a method with temporary support or a method without temporary support), the number of lanes, the lateral distribution coefficient, the temperature change (including overall heating and cooling and gradient heating and cooling), the ambient humidity and the vehicle load level, etc., while ensuring that the design parameters match the design conditions. By setting different design conditions, the operation of the bridge under different working conditions can be simulated. In the simulation process, by continuously adjusting the bridge section design parameters, it is ensured that the bridge structure performance can meet the design requirements under various design conditions. Through the iteration and optimization process, the best combination of bridge section design parameters and design conditions is found. At the same time, according to the simulation results, a data set containing bridge section design parameters and corresponding design conditions is obtained. The data set records the bridge section design parameters and bridge structure performance under different design conditions. In this way, based on the bridge numerical model, a sufficient number of data sets of bridge section design parameters and corresponding design conditions can be obtained, which provides rich data support for subsequent analysis, evaluation and model training.

[0042] S120, generating an original bridge section design schematic based on the bridge section design parameters in the data set, and generating conditional characteristic data based on the design conditions.

[0043] In this step, in order to further use the data in the data set for model training, the bridge section design parameters and design conditions are converted into a form that is easy for computer processing and analysis through automated means, providing a data basis for subsequent model training.

[0044] Optionally, in some embodiments, see Figure 3 As shown, in step S120, an original bridge section design diagram is generated based on the bridge section design parameters in the data set, and conditional feature data is generated based on the design conditions, including:

[0045] S121, using the matplotlib library to generate an original bridge section design diagram according to the bridge section design parameters; wherein the original bridge section design diagram is a symmetrical half section design diagram;

[0046] S123. Using a data processing tool, convert the design conditions into conditional characteristic data of a predetermined dimension; wherein the data format of the conditional characteristic data is a data format that can be recognized and processed by a computer.

[0047] In this embodiment, the graphics drawing library matplotlib in the Python programming language is used to write a specific function module, and the bridge section design parameters are used as input. Through the logic processing in the function module, the bridge section design parameters are directly converted into a bridge section design diagram to obtain the original bridge section design diagram, wherein the original bridge section design diagram is a symmetrical half-section design diagram; the data processing tool in the machine learning framework PyTorch is used to write a data processing script to convert the design conditions into feature data y of a predetermined dimension, that is, conditional feature data; the dimension of the conditional feature data, that is, the specific data y in the embodiment of the present invention is (1, 1, 7) to meet the input requirements of the subsequent model; the converted feature data y exists in a machine-readable form, providing necessary data support for subsequent model training. It should be noted that in view of the symmetry of the bridge section, in order to improve the calculation efficiency and simplify the problem, the embodiment of the present invention only considers the symmetrical half section for subsequent training and analysis. Doing so can not only effectively reduce the amount of calculation, but also maintain the integrity and accuracy of the design.

[0048] S130, constructing a bridge section intelligent generation model based on the CF-VAE-GAN algorithm, and training the bridge section intelligent generation model based on the original bridge section design sketch and conditional feature data.

[0049] In this step, the embodiment of the present invention introduces a generative intelligent design algorithm based on the CF-VAE-GAN algorithm, constructs a bridge section intelligent generation model, performs model training based on the original bridge section design diagram and conditional feature data, and fully learns the rules between the conditional feature data and the bridge section design diagram to realize automated intelligent design, so that the model can generate a bridge section design diagram for specific design conditions.

[0050] Optionally, in some embodiments, see Figure 4 As shown, step S130 constructs a bridge section intelligent generation model, and trains the bridge section intelligent generation model based on the original bridge section design diagram and conditional feature data, including:

[0051] S131, inputting the original bridge section design diagram and the conditional characteristic data into the bridge section intelligent generation model, and processing them through the encoder module to obtain hidden variables;

[0052] S133, inputting the latent variables and the conditional characteristic data into the generator module to obtain a new bridge section design diagram;

[0053] S135. Optimize the bridge section intelligent generation model based on the discriminator and similarity discriminator modules to obtain a trained bridge section intelligent generation model.

[0054] Optionally, in some embodiments, step S135 optimizes the bridge section intelligent generation model based on the discriminator and similarity discriminator modules to obtain a trained bridge section intelligent generation model, including: inputting the new bridge section design schematic, the conditional feature data and the original bridge section design schematic into the discriminator module, and using the discriminator module to distinguish the generated new bridge section design schematic from the real original bridge section design schematic; obtaining the feature tensor of the new bridge section design schematic based on the encoder module, and using the similarity discriminator module to evaluate the similarity between the feature tensor of the new bridge section design schematic and the feature tensor of the original bridge section design schematic; optimizing the bridge section intelligent generation model according to the evaluation result to obtain a trained bridge section intelligent generation model.

[0055] For details, see Figure 5 As shown, step S130 includes: first, the original bridge cross-section design diagram generated in step S120 is used as input data and sent to the image encoder E I In the image encoder E I Responsible for extracting key features from the input original bridge section diagram and generating a set of feature tensors f x ; Next, this set of feature tensors f x Together with the specific conditional feature data y, it is passed to the distribution encoder E Z , distributed encoder E Z Combine the input conditional feature data y with the feature tensor f x After stacking and splicing, it is compressed through a series of nonlinear transformations to finally generate a low-dimensional latent variable representation that meets specific conditions. These latent variables can capture the main features of the input data while removing some redundant information. It should be noted that the distribution encoder E Z Output two probability distribution parameters used to describe the latent variables, namely the mean and variance. These two parameters define a multidimensional Gaussian distribution N(µ,σ), which represents the conditional feature data y and the feature tensor f x representation in the latent space; then, from the distribution encoder E ZA specific latent variable value z is sampled from the output distribution, which is stacked and spliced ​​with the specific conditional feature data y to provide the complete information required by the generator module. The generator module uses these spliced ​​features to generate a new bridge section diagram under specific design conditions; then, the generated new bridge section diagram is combined with the conditional feature data y and the original bridge section diagram. Figure 1 The generated new bridge cross-section diagram is sent to the discriminator module for quality discrimination; the discriminator module is responsible for distinguishing the generated new bridge cross-section diagram from the real original bridge cross-section diagram, thereby providing feedback to optimize the generator module. In addition, in order to further improve the accuracy and authenticity of the generated new bridge cross-section diagram, the embodiment of the present invention also adopts a similarity discriminator to re-input the new bridge cross-section diagram generated by the generator module into the image encoder E I In the above example, we generate a new feature tensor f x '; The characteristic tensor f obtained from the original bridge section diagram x With f x ' are sent together to the similarity discriminator module for comparison. The similarity discriminator module is responsible for evaluating the similarity between the two feature tensors to determine the similarity between the bridge section diagram generated by the generator module and the original diagram. Through the collaborative optimization strategy of the discriminator and the similarity discriminator, the intelligent generation model of bridge sections can be trained and updated, and the performance of the encoder module, generator module and discriminator module can be gradually optimized, and finally a model that can generate clear, regular and highly compliant bridge section design diagrams according to specific design conditions is obtained.

[0056] In some embodiments, the bridge section intelligent generation model includes: an encoder module, a generator module, and a discriminator module;

[0057] The encoder module includes an image encoder and a distribution encoder; the image encoder is used to extract key features from the input original bridge section design diagram and output the feature tensor of the key features; the distribution encoder is used to compress the input feature tensor and conditional feature data into low-dimensional latent variables;

[0058] The generator module is used to process the low-dimensional latent variables and the conditional characteristic data using multiple network modules to generate a new bridge section design diagram;

[0059] The discriminator module is used to generate respective discriminant feature tensors based on the new bridge section design schematic and the original bridge section design schematic in combination with the conditional feature data to discriminate the new bridge section design schematic and the original bridge section design schematic.

[0060] Optionally, in some embodiments, the bridge section intelligent generation model also includes: a similarity discriminator module, which is used to discriminate based on the feature tensor obtained by the image encoder between the new bridge section design schematic and the original bridge section design schematic, and evaluate the similarity between the new bridge section design schematic and the original bridge section design schematic.

[0061] The encoder module includes an image encoder E I and distributed encoder E Z ; For image encoder E I The embodiment of the present invention provides an image encoder E based on an improved ViT (Vision Transformer, visual converter) model I , which is particularly suitable for feature extraction of bridge cross-section diagrams. Figure 6 As shown, the image encoder E of the embodiment of the present invention I While maintaining the core structure of the ViT model, the multi-layer perceptron head used for image classification in the original ViT model is removed through key adjustments to suit specific application requirements. This change makes the image encoder E I Instead of directly outputting the classification results of the image, it focuses on extracting and directly outputting the feature representation of the image from the input bridge cross-section diagram. This design optimization makes E I It is more suitable for application scenarios that require deep image features rather than direct classification output, especially in the field of automated intelligent generation of bridge section design.

[0062] like Figure 6 As shown, in practical applications, the image encoder E I It accepts a picture of dimension 224×224×3 as input, passes it into the Linear Projection of Flattened Patches module, and divides it according to the fixed Patch size (32x32×3) to obtain multiple image blocks of dimension 32x32×3; then, these image blocks are converted into one-dimensional vectors through the Patch flattening operation. The flattened vectors are mapped to a vector of fixed length through linear transformation and convolution operations to form a Patch Embedding, and then enter the Patch+PositionEmbedding (addition of patch embedding and position embedding) module to add the corresponding position embedding (Position Embedding) to each Patch Embedding to retain the position information of the image block in the original image; then, the Patch Embedding with added position embedding is used as input, and deep feature extraction and transformation are performed through the Transformer model to finally obtain a low-dimensional feature tensor fx ∈R 256×256×3 , which contains the key feature information of the bridge cross-section diagram.

[0063] For the distributed encoder E Z ,like Figure 5 As shown in the figure, it is built with ResNet as the skeleton, and the conditional feature data y and the corresponding feature tensor f x Through the distributed encoder E Z is mapped to a continuous flow space described by an independent Gaussian distribution N(µ,σ), thereby ensuring the continuity and richness of feature representation and conditional information. Then, from the distribution encoder E Z Sampling from the output distribution obtains a specific latent variable value z, ; Among them, ε∈N(0,1) is a normally distributed noise, μ is the mean, and σ is the variance.

[0064] The generator module includes a generator G. For the generator G, its network architecture is as follows Figure 5 As shown in the figure, first, the latent variable value z is stacked with the specific conditional feature data y and used as the input of the generator G. Then, the input data passes through a network module composed of a deconvolution layer, a Batch Normalization layer and a ReLU activation function, and the output is a 7×7×256 feature map. Next, the network is stacked with four anti-residual blocks composed of deconvolution layers. ReLU is used as the activation function after each deconvolution layer, and a Batch Normalization layer is followed for batch normalization. Finally, a deconvolution layer with an activation function of tanh is used to output the generated new 224×224×3 bridge cross-section diagram.

[0065] The discriminator module includes a discriminator D. For the discriminator D, Figure 5As shown in the figure, the discriminator D accepts the newly generated bridge section diagram, conditional feature data y and the original bridge section diagram as input to discriminate the newly generated image from the original image. Specifically, the image and conditional feature data y are first mapped by the fully connected layer and reshaped, and then spliced ​​to output a 224×224×10 feature map. The feature map then enters a network structure consisting of a convolution layer, a BatchNormalization layer, a ReLU activation function and a downsampling layer. The following network is composed of three residual blocks consisting of convolution layers. Finally, a new feature tensor with a dimension of 1×1×256 is generated by the output of a convolution layer with an activation function of Sigmoid. The feature tensor is used to represent the discriminative features of the image. In the model optimization process, by comparing the discriminative features of the original bridge section image and the bridge section image generated by the generator G, the loss function is used to adjust the model parameters of the generator G and the discriminator D. In this way, the quality of the image generated by the generator G can be continuously improved, while the ability of the discriminator D to evaluate the image quality can be improved.

[0066] The similarity discriminator module includes a similarity discriminator C. For the similarity discriminator C, Figure 5 As shown, the similarity comparator C receives two inputs: one is the original bridge cross-section diagram after the image encoder E I The feature tensor f obtained after processing x , and secondly, the new bridge cross-section diagram generated by the generator G also passes through the image encoder E I The feature tensor f obtained after processing x '. These two feature tensors f x and f x 'Together as the input of the similarity discriminator C, it is used to evaluate the similarity between the two feature tensors. By comparing the two feature tensors, the similarity discriminator C can further determine the degree of proximity between the bridge cross-section diagram generated by the generator G and the original diagram in the feature space, and then evaluate the quality of the image generated by the generator G. Ultimately, the purpose of further improving the accuracy and authenticity of the generated bridge cross-section diagram is achieved.

[0067] In some embodiments, the objective function of the CF-VAE-GAN algorithm is: ;

[0068] Among them, λ V and λ C is a hyperparameter for controlling the influence of the loss term, which is set to 0.5 in the embodiments of the present invention; V(G min ,D max ) is the adversarial loss between the generator G and the discriminator D, expressed as V G min , D max = E x~ P data (x) log D(x|y) + E x~ P z(z) log [1-D(G(z|y))] ; x is the original bridge section design diagram, y is the conditional characteristic data, and z is the hidden variable;

[0069] L C is the loss of the similarity discriminator C, expressed as ;f x f is the feature tensor obtained by inputting the original bridge section design diagram into the image encoder, x 'The generated new bridge cross-section diagram is re-input into the image encoder to obtain a new feature tensor;

[0070] D KL represents the KL divergence loss, which aims to make the encoding distribution as close to the standard normal distribution as possible. Its expression is ;

[0071] N(μ,σ) is an independent Gaussian distribution, μ is the mean and σ is the variance.

[0072] Specifically, the process of constructing and training a bridge section intelligent generation model based on the CF-VAE-GAN algorithm in the embodiment of the present invention includes:

[0073] Feature extraction: The original bridge section design diagram is used as input data and sent to the image encoder E I Image Encoder E I Responsible for extracting key features from the input original bridge section diagram and generating a set of feature tensors f x .

[0074] Conditional feature fusion: This set of feature tensors f x Together with the specific conditional feature data y, it is passed to the distribution encoder E z , the input conditional feature data y and the feature tensor f x After stacking and splicing, they are compressed through a series of nonlinear transformations to generate a low-dimensional latent variable representation that meets specific conditions.

[0075] Hidden variable generation and sampling: distribution encoder E z The output is used to describe the probability distribution parameters (i.e., mean and variance) of the latent variable, which define a multidimensional Gaussian distribution N(µ,σ). A specific latent variable value z is sampled from this distribution.

[0076] Generator G generates a new design: The sampled latent variable value z is stacked and spliced ​​with the specific conditional feature data y to provide complete information for generator G. Generator G uses this information to generate a new bridge section diagram under specific conditions.

[0077] Quality evaluation of discriminator D: The generated bridge section diagram is compared with the conditional feature data y and the original bridge section diagram. Figure 1 It is sent to the discriminator D for quality judgment to distinguish the generated bridge cross-section diagram from the real bridge cross-section diagram and provide feedback for the optimization generator.

[0078] The similarity discriminator C evaluates the similarity: the new bridge section diagram generated by the generator G is re-input into the image encoder E I In the above example, we generate a new feature tensor f x '; The characteristic tensor f obtained from the original bridge section diagram x With f x ' are sent together to the similarity discriminator module for comparison. The similarity discriminator module is responsible for evaluating the similarity between the two feature tensors to determine the similarity between the bridge section diagram generated by the generator module and the original diagram.

[0079] Collaborative optimization and cyclic learning: In order to improve the accuracy and authenticity of the generated bridge cross-section diagram, a collaborative optimization and cyclic learning strategy is adopted. The discriminant ability of the discriminator D ensures the basic quality of the generated image, and the detail evaluation ability of the similarity discriminator C ensures that the details of the generated image are reasonable. Combining the role of the discriminator D and the similarity discriminator C, the collaborative feedback loop promotes the distribution encoder E z , improvement of the performance of the generator G and the discriminator D.

[0080] The embodiment of the present invention uses a process to iterate and optimize the model multiple times to obtain a model that can generate a clear, regular and highly compliant bridge section design diagram according to specific conditions. This solves the defects that traditional section design usually relies on empirical formulas and finite element calculation software, has low design efficiency when manually adjusting design parameters, relies on the professional knowledge and experience of engineers, and has a low degree of intelligence in this field.

[0081] S140, inputting the target condition characteristic data into the trained bridge section intelligent generation model to obtain a bridge section design diagram for the target design conditions.

[0082] In this step, when generating a bridge section design sketch for specific design conditions, the bridge section intelligent generation model trained by the embodiment of the present invention is used to first input the target design conditions into a data processing script written in the machine learning framework PyTorch, and the target design conditions are converted into target condition feature data of a preset dimension. The target condition feature data is then input into the trained bridge section intelligent generation model to directly generate a bridge section design sketch for the target design conditions.

[0083] The intelligent generation method of bridge section design diagrams provided by the embodiment of the present invention builds a bridge numerical model based on a completed bridge, obtains a data set of design parameters and corresponding design conditions based on the bridge numerical model; generates an original bridge section design diagram based on the bridge section design parameters, and generates conditional characteristic data based on the design conditions; and uses a bridge section intelligent generation model based on the CF-VAE-GAN algorithm to generate a bridge section design diagram for target design conditions according to the rules between the conditional characteristic data and the bridge section design diagram. In this way, a bridge section design diagram for target design conditions is intelligently generated by the bridge section intelligent generation model to solve the problem of low efficiency caused by the traditional section design method relying on manual adjustment.

[0084] Embodiment 2

[0085] The present invention also provides a bridge section design diagram intelligent generation system according to an embodiment of the present invention. Figure 7 As shown, including:

[0086] The data acquisition module 710 is used to construct a bridge numerical model based on the completed bridge, and obtain a data set of design parameters and corresponding design conditions based on the bridge numerical model;

[0087] A data processing module 720, for generating an original bridge section design diagram based on the bridge section design parameters in the data set, and generating conditional characteristic data based on the design conditions;

[0088] A model training module 730 is used to construct a bridge section intelligent generation model based on the CF-VAE-GAN algorithm, and train the bridge section intelligent generation model based on the original bridge section design diagram and conditional feature data;

[0089] The design diagram generation module 740 is used to input the target condition characteristic data into the trained bridge section intelligent generation model to obtain a bridge section design diagram for the target design conditions.

[0090] Optionally, in some embodiments, the bridge section design parameters include: mid-span bridge deck size, mid-span steel box girder size, support bridge deck size and support steel box girder size; the design conditions include: construction method, number of lanes, lateral distribution coefficient, temperature change, ambient humidity and vehicle load level; the bridge section design parameters match the design conditions.

[0091] Optionally, in some embodiments, the data acquisition module is specifically used to: obtain the bridge section design parameters of the completed bridge and the design conditions corresponding to the section design parameters to establish a bridge numerical model; the bridge numerical model is used to simulate the structural performance of the bridge under different design conditions;

[0092] The bridge numerical model is used for simulation, and the design parameters of the bridge section are adjusted so that the bridge structure performance can meet the requirements of different design conditions;

[0093] According to the simulation results, a data set of design parameters and corresponding design conditions is generated; the data set is used to record the bridge section design parameters and bridge structure performance under different design conditions.

[0094] Optionally, in some embodiments, the data processing module is specifically used to generate an original bridge section design diagram according to the bridge section design parameters using a matplotlib library; wherein the original bridge section design diagram is a symmetrical half section design diagram;

[0095] The design conditions are converted into conditional characteristic data of a predetermined dimension using a data processing tool; wherein the data format of the conditional characteristic data is a data format that can be recognized and processed by a computer.

[0096] Optionally, in some embodiments, the bridge section intelligent generation model includes: an encoder module, a generator module, a discriminator module and a similarity discriminator module;

[0097] The encoder module includes an image encoder and a distribution encoder; the image encoder is used to extract key features from the input original bridge section design diagram and output the feature tensor of the key features; the distribution encoder is used to compress the input feature tensor and conditional feature data into low-dimensional latent variables;

[0098] The generator module is used to process the low-dimensional latent variables and the conditional characteristic data using multiple network modules to generate a new bridge section design diagram;

[0099] The discriminator module is used to generate respective discriminant feature tensors based on the new bridge section design diagram and the original bridge section design diagram in combination with the conditional feature data, so as to discriminate the new bridge section design diagram and the original bridge section design diagram;

[0100] The similarity discriminator module is used to discriminate based on the feature tensor obtained by the image encoder between the new bridge section design schematic and the original bridge section design schematic, and evaluate the similarity between the new bridge section design schematic and the original bridge section design schematic.

[0101] Optionally, in some embodiments, the model training module is specifically used to input the original bridge section design diagram and the conditional feature data into the bridge section intelligent generation model, and process them through the encoder module to obtain latent variables;

[0102] Inputting the latent variables and the conditional characteristic data into the generator module to obtain a new bridge section design diagram;

[0103] The bridge section intelligent generation model is optimized based on the discriminator and similarity discriminator modules to obtain a trained bridge section intelligent generation model.

[0104] Embodiment 3

[0105] Figure 8 FIG. 1 is a schematic block diagram of an electronic device according to an embodiment of the present invention; based on the same technical concept as the first embodiment, the electronic device provided by the embodiment of the present invention is as follows: Figure 8 As shown, the step flow of any embodiment method described in Embodiment 1 of the present invention can be implemented.

[0106] The above-mentioned electronic device may include: a shell 81, a processor 82, a memory 83, a circuit board 84 and a power supply circuit 85, wherein the circuit board 84 is arranged inside the space enclosed by the shell 81, and the processor 82 and the memory 83 are arranged on the circuit board 84; the power supply circuit 85 is used to supply power to various circuits or devices of the above-mentioned electronic device; the memory 83 is used to store executable program codes; the processor 82 runs the program corresponding to the executable program code by reading the executable program code stored in the memory 83, so as to execute the intelligent generation method of the bridge section design schematic described in any of the aforementioned embodiments one.

[0107] The specific execution process of the above steps by the processor 82 and the steps further executed by the processor 82 by running the executable program code can be found in the description of the first embodiment of the present invention, which will not be repeated here.

[0108] The electronic device exists in many forms, including but not limited to:

[0109] (1) Mobile communication devices: These devices are characterized by their mobile communication functions and their main purpose is to provide voice and data communications. These terminals include: smart phones (such as iPhone), multimedia phones, feature phones, and low-end phones.

[0110] (2) Ultra-mobile personal computer devices: These devices fall into the category of personal computers, have computing and processing capabilities, and generally also have mobile Internet access features. These terminals include: PDA, MID and UMPC devices, such as iPad.

[0111] (3) Portable entertainment devices: These devices can display and play multimedia content. They include audio and video players (such as iPods), handheld game consoles, e-books, smart toys, and portable car navigation devices.

[0112] (4) Server: A device that provides computing services. The server consists of a processor, hard disk, memory, system bus, etc. The server is similar to a general-purpose computer architecture, but because it needs to provide highly reliable services, it has higher requirements in terms of processing power, stability, reliability, security, scalability, and manageability.

[0113] (5) Other electronic devices with data interaction functions.

[0114] Embodiment 4

[0115] An embodiment of the present invention also provides a computer-readable storage medium, which stores one or more programs. The one or more programs can be executed by one or more processors to implement the method for intelligently generating bridge section design schematics as described in any of the above-mentioned embodiments.

[0116] In summary, the embodiment of the present invention provides a method and system for intelligently generating a bridge section design diagram, which constructs a bridge numerical model based on a completed bridge, obtains a data set of design parameters and corresponding design conditions based on the bridge numerical model; generates an original bridge section design diagram based on the bridge section design parameters, and generates conditional characteristic data based on the design conditions; and uses a bridge section intelligent generation model based on the CF-VAE-GAN algorithm to generate a bridge section design diagram for target design conditions according to the rules between the conditional characteristic data and the bridge section design diagram. In this way, a bridge section design diagram for target design conditions is intelligently generated by the bridge section intelligent generation model to solve the problem of low efficiency caused by the traditional section design method relying on manual adjustment.

[0117] Furthermore, an embodiment of the present invention also provides a bridge section intelligent generation model based on the CF-VAE-GAN algorithm. This model not only improves the efficiency of bridge design, but also provides a powerful auxiliary tool to help quickly generate bridge section design schematics under various conditions.

[0118] It should be noted that, in this article, relational terms such as first and second, etc. 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 variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0119] Each embodiment in this specification is described in a related manner, and the same or similar parts between the embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

[0120] For the convenience of description, if it involves a system, server, etc., it may be described separately by dividing the functions into various units / modules. Of course, when implementing the present invention, the functions of each unit / module can be implemented in the same or multiple software and / or hardware.

[0121] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes of the embodiments of the above-mentioned methods. The storage medium can be a disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM), etc.

[0122] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed by the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. A method for intelligently generating a bridge cross-section design diagram, characterized in that: The method comprises: constructing a bridge numerical model based on the completed bridge, and obtaining a data set of design parameters and corresponding design conditions based on the bridge numerical model; generating an original bridge section design diagram based on the bridge section design parameters in the data set, and generating conditional characteristic data based on the design conditions; Constructing a bridge section intelligent generation model based on the CF-VAE-GAN algorithm, and training the bridge section intelligent generation model based on the original bridge section design diagram and conditional feature data; Input the target condition characteristic data into the trained bridge section intelligent generation model to obtain a bridge section design diagram for the target design conditions; The bridge section intelligent generation model based on the CF-VAE-GAN algorithm includes: an encoder module, a generator module, a discriminator module and a similarity discriminator module; The encoder module includes an image encoder and a distribution encoder; the image encoder is used to extract key features from the input original bridge section design diagram and output the feature tensor of the key features; the distribution encoder is used to compress the input feature tensor and conditional feature data into low-dimensional latent variables; The generator module is used to process the low-dimensional latent variables and the conditional characteristic data using multiple network modules to generate a new bridge section design diagram; The discriminator module is used to generate respective feature tensors based on the new bridge section design diagram and the original bridge section design diagram in combination with the conditional feature data to discriminate the new bridge section design diagram from the original bridge section design diagram; The similarity discriminator module is used to discriminate based on the feature tensor obtained by the image encoder between the new bridge section design diagram and the original bridge section design diagram, and evaluate the similarity between the new bridge section design diagram and the original bridge section design diagram; The method of constructing a bridge section intelligent generation model based on the CF-VAE-GAN algorithm and training the bridge section intelligent generation model based on the original bridge section design diagram and conditional feature data includes: Inputting the original bridge section design diagram and the conditional characteristic data into the bridge section intelligent generation model, and processing them through the encoder module to obtain hidden variables; Inputting the latent variables and the conditional characteristic data into the generator module to obtain a new bridge section design diagram; The bridge section intelligent generation model is optimized based on the discriminator and similarity discriminator modules to obtain a trained bridge section intelligent generation model.

2. The intelligent generation method of bridge cross-section design diagram according to claim 1 is characterized in that: The bridge section design parameters include: mid-span bridge deck size, mid-span steel box girder size, support bridge deck size and support steel box girder size; the design conditions include: construction method, number of lanes, lateral distribution coefficient, temperature change, ambient humidity and vehicle load level; the bridge section design parameters match the design conditions.

3. The intelligent generation method of bridge cross-section design diagram according to claim 1 is characterized in that: The step of constructing a bridge numerical model based on the completed bridge and obtaining a data set of design parameters and corresponding design conditions based on the bridge numerical model includes: Obtaining the bridge section design parameters of the completed bridge and the design conditions corresponding to the section design parameters to construct a bridge numerical model; the bridge numerical model is used to simulate the structural performance of the bridge under different design conditions; The bridge numerical model is used for simulation, and the design parameters of the bridge section are adjusted so that the bridge structure performance can meet the requirements of different design conditions; According to the simulation results, a data set of design parameters and corresponding design conditions is generated; the data set is used to record the bridge section design parameters and bridge structure performance under different design conditions.

4. The intelligent generation method of bridge cross-section design diagram according to claim 1 is characterized in that: The generating of the original bridge section design diagram based on the bridge section design parameters in the data set, and the generating of conditional characteristic data based on the design conditions, include: Generate an original bridge section design diagram according to the bridge section design parameters using the matplotlib library; wherein the original bridge section design diagram is a symmetrical half section design diagram; The design conditions are converted into conditional characteristic data of a predetermined dimension using a data processing tool; wherein the data format of the conditional characteristic data is a data format that can be recognized and processed by a computer.

5. The intelligent generation method of bridge cross-section design diagram according to claim 1 is characterized in that: The objective function of the CF-VAE-GAN algorithm is: ； Among them, λ V and λ C is a hyperparameter that controls the impact of the loss term; V(G min ,D max ) is the adversarial loss between the generator G and the discriminator D, expressed as ; x is the original bridge section design diagram, y is the conditional characteristic data, and z is the hidden variable; L C is the loss of the similarity discriminator C, expressed as ;f x f is the feature tensor obtained by inputting the original bridge section design diagram into the image encoder, x 'The generated new bridge cross-section diagram is re-input into the image encoder to obtain a new feature tensor; D KL represents the KL divergence loss, the purpose is to make the encoding distribution as close to the standard normal distribution as possible, and its expression is ; N(μ,σ) is an independent Gaussian distribution, μ is the mean and σ is the variance.

6. A bridge section design diagram intelligent generation system, characterized in that: include: A data acquisition module, used to construct a bridge numerical model based on the completed bridge, and obtain a data set of design parameters and corresponding design conditions based on the bridge numerical model; A data processing module, used to generate an original bridge section design diagram based on the bridge section design parameters in the data set, and to generate conditional characteristic data based on the design conditions; A model training module, used to construct a bridge section intelligent generation model based on the CF-VAE-GAN algorithm, and to train the bridge section intelligent generation model based on the original bridge section design diagram and conditional feature data; A design diagram generation module, which is used to input the target condition characteristic data into the trained bridge section intelligent generation model to obtain a bridge section design diagram for the target design conditions; The bridge section intelligent generation model based on the CF-VAE-GAN algorithm includes: an encoder module, a generator module, a discriminator module and a similarity discriminator module; The encoder module includes an image encoder and a distribution encoder; the image encoder is used to extract key features from the input original bridge section design diagram and output the feature tensor of the key features; the distribution encoder is used to compress the input feature tensor and conditional feature data into low-dimensional latent variables; The generator module is used to process the low-dimensional latent variables and the conditional characteristic data using multiple network modules to generate a new bridge section design diagram; The discriminator module is used to generate respective feature tensors based on the new bridge section design diagram and the original bridge section design diagram in combination with the conditional feature data to discriminate the new bridge section design diagram from the original bridge section design diagram; The similarity discriminator module is used to discriminate based on the feature tensor obtained by the image encoder between the new bridge section design diagram and the original bridge section design diagram, and evaluate the similarity between the new bridge section design diagram and the original bridge section design diagram; The model training module is specifically used to input the original bridge section design diagram and the conditional feature data into the bridge section intelligent generation model, and process them through the encoder module to obtain hidden variables; Inputting the latent variables and the conditional characteristic data into the generator module to obtain a new bridge section design diagram; The bridge section intelligent generation model is optimized based on the discriminator and similarity discriminator modules to obtain a trained bridge section intelligent generation model.

7. An electronic device, characterized in that: The electronic device comprises: a shell, a processor, a memory, a circuit board and a power supply circuit, wherein the circuit board is placed inside the space enclosed by the shell, and the processor and the memory are arranged on the circuit board; the power supply circuit is used to supply power to various circuits or devices of the above-mentioned electronic device; the memory is used to store executable program codes; the processor runs a program corresponding to the executable program code by reading the executable program code stored in the memory, so as to execute the intelligent generation method of bridge section design schematics described in any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the method for intelligently generating bridge section design schematics as described in any one of claims 1 to 5.

Images