Subway secondary structure assembly type design method

Abstract

The invention relates to the technical field of rail traffic engineering, and discloses a subway secondary structure assembly type design method, which comprises the following steps of S1, collecting civil engineering structure size data, geological survey data and electromechanical pipeline layout data of subway stations and intervals, filling missing values in the data by adopting a cubic spline interpolation method, and obtaining a data matrix; performing maximum-minimum standardization processing on the continuous size data; s2, a three-dimensional grid topology model is constructed based on the installation point position coordinates of the subway secondary structure, nodes in the model correspond to the installation positions of the assembly type components, edges represent the space connection relation between the components, and the weight of the edges is calculated according to the space distance of component installation. According to the method, the data quality is guaranteed through multi-dimensional data standard processing, the assembly layout is accurately learned based on the three-dimensional model and the attention convolution network, node deviation is efficiently corrected in combination with an abnormal threshold value and a weighted median method, and design precision, intelligence and security are achieved.

Description

Technical Field

[0001] This invention relates to the field of rail transit engineering technology, specifically to a prefabricated design method for the secondary structure of subways. Background Technology

[0002] In the field of prefabricated design for subway secondary structures, traditional design methods suffer from several key problems: Firstly, data sources for subway station and section civil engineering dimensions, geological surveys, and electromechanical pipelines are scattered and prone to gaps. Furthermore, continuous data exhibits significant differences in dimensions and is difficult to quantify, resulting in inconsistent data quality and an inability to provide accurate and unified data support for design. Secondly, traditional designs often rely on manual experience to construct component assembly layouts, making it difficult to intuitively depict the spatial connections between components and to fully integrate multi-dimensional characteristics such as component material, dimensions, and load-bearing capacity, leading to insufficient layout rationality. In addition, there is a lack of objective and efficient identification standards and correction methods for abnormal nodes in the assembly layout. Manual judgment is highly subjective and has low correction accuracy, which not only affects design efficiency but may also pose hidden dangers to the safety and stability of subway secondary structures, failing to meet the demands of modern subway engineering for precise and intelligent prefabricated design.

[0003] To address this issue, we propose a prefabricated design method for the secondary structure of subway systems. Summary of the Invention

[0004] (a) Technical problems to be solved

[0005] To address the shortcomings of existing technologies, this invention provides a prefabricated design method for subway secondary structures, which solves the problems mentioned in the background section.

[0006] (II) Technical Solution

[0007] To achieve the above objectives, the present invention provides the following technical solution: a prefabricated design method for subway secondary structures, comprising the following steps:

[0008] Step S1: Collect the civil engineering structure dimensions, geological survey data, and electromechanical pipeline layout data of subway stations and sections. Fill the missing values ​​in the data with cubic spline interpolation and perform maximum-minimum standardization on continuous dimension data.

[0009] Step S2: Based on the installation point coordinates of the subway secondary structure, construct a three-dimensional mesh topology model. In the model, nodes correspond to the installation positions of prefabricated components, edges represent the spatial connection relationships between components, and edge weights are calculated based on the spatial distance of component installation.

[0010] Step S3: Construct a feature vector for each node of the 3D mesh topology model, fusing the material features, size features, and load-bearing features of the components;

[0011] Step S4: Design a convolutional neural network model based on the attention mechanism to learn the reasonable assembly layout pattern of the subway secondary structure in a self-supervised manner, and complete the model training by minimizing the layout deviation loss function;

[0012] Step S5: Calculate the layout deviation value of each node using the trained model, set an abnormal threshold based on the distribution of historical layout deviation data, and mark the installation nodes with abnormal layout.

[0013] Step S6: For installation nodes marked as abnormal, adjust the assembly parameters of the prefabricated components at the node using the weighted median method, taking into account the assembly information of its adjacent nodes.

[0014] Preferably, step S1 includes the following steps:

[0015] Step S11: Collect 3D point cloud data of the civil engineering structure of subway station platforms, concourses and tunnels using a 3D laser scanning device, and extract the length, width and height dimensional parameters of the structure;

[0016] Step S12: Extract soil type, foundation bearing capacity, and groundwater level data along the subway line from the geological survey report;

[0017] Step S13: Export the routing, diameter, and installation height data of the subway electromechanical pipelines using BIM modeling software;

[0018] Step S14: Fill in the missing values ​​in the civil engineering structure dimension data using cubic spline interpolation. For the missing dimension parameter point x... t Size value Where a t b t c t d t These are the cubic spline coefficients fitted based on adjacent valid data points;

[0019] Step S15: Perform maximum-minimum standardization on continuous features such as civil engineering structure dimensions and foundation bearing capacity to map their values ​​to the [0, 1] interval. For category features such as soil type and component material type, use one-hot encoding to convert them into numerical vectors.

[0020] Preferably, step S2 includes the following steps:

[0021] Step S21: Define the node set V of the 3D mesh topology model, where each node v i For each prefabricated component installation point, the number of nodes N is the total number of prefabricated component installation points for the subway secondary structure.

[0022] Step S22: Construct edges between nodes using the spatial k-nearest neighbor method, setting k=8, and each node connects to the 8 installation nodes with the closest spatial distance;

[0023] Step S23: Node v i With node v j The edge weight w between ij Based on spatial Euclidean distance calculation, the formula is: In the formula, d ij γ is the spatial Euclidean distance between installation nodes i and j, and γ is the spatial scale adjustment parameter.

[0024] Step S24: Construct the adjacency matrix A of the 3D mesh topology model. If there is an edge between node i and node j, then A ij =w ij Otherwise, A ij =0.

[0025] Preferably, step S3 includes the following steps:

[0026] Step S31: Extract the material characteristics of the prefabricated components, including the numerical indicators corresponding to the concrete strength grade and steel grade.

[0027] Step S32: Extract the dimensional features of the component, including the component's length, thickness, and reserved hole size, and calculate the ratio between the dimensional parameters as derived features;

[0028] Step S33: Extract the load-bearing characteristics of the component, obtain the flexural and compressive bearing capacity values ​​of the component through structural mechanics calculations, and use the normalized values ​​as characteristic values;

[0029] Step S34: Concatenate the material features, dimensional features, and load-bearing features along the dimensions to form a node feature vector x. i ∈R m Then, construct the node feature matrix X∈R N×m .

[0030] Preferably, step S4 includes the following steps:

[0031] Step S41: The encoder uses a 3D convolutional layer with an attention mechanism to map the node feature matrix X and adjacency matrix A to the latent feature representation Z. The encoder layer is defined as follows: In the formula, The adjacency matrix for adding self-joins, where I is the identity matrix and D is... The degree matrix, W1 and W2 are the trainable weight matrices of the encoder, and Attention is the attention calculation module;

[0032] Step S42: The decoder uses a deconvolution layer to reconstruct the assembly layout features of the Z nodes from the latent feature representation. The decoder layer is defined as follows: In the formula, W3 is the trainable weight matrix of the decoder, and ConvTranspose3D is the three-dimensional deconvolution operation.

[0033] Step S43: The training process uses the layout bias loss function as the optimization objective, and the loss function is the mean absolute error.

[0034] The formula is: In the formula, N is the total number of nodes, x i Let be the original feature vector of the i-th node. Let be the reconstructed feature vector of the i-th node; train the model iteratively through backpropagation and stochastic gradient descent optimizer.

[0035] Preferably, step S5 includes the following steps:

[0036] Step S51: Using the trained model, calculate the layout deviation value e for each node of the newly input subway secondary structure assembly data. i The formula is:

[0037] Step S52: Based on the distribution of layout deviation values ​​in historical training data, set an abnormal threshold. Set as the 90th percentile of the layout deviation value;

[0038] Step S53: If the node layout deviation value Then mark the installation node as a layout anomaly node.

[0039] Preferably, step S6 includes the following steps:

[0040] Step S61: For installation nodes v marked as abnormal i Extract its neighboring node set M(i) in the three-dimensional mesh topology model;

[0041] Step S62: Adjust only the assembly parameters of the prefabricated components at the abnormal nodes, and calculate the adjusted parameter values ​​using the weighted median method. The formula is as follows: In the formula, Let x be the adjusted assembly parameter vector for the i-th node. j w is the assembly parameter vector of adjacent nodes. ij Let M(i) be the edge weight between nodes i and j, where j∈M(i) represents the weight of all adjacent nodes of node i.

[0042] Step S63: Update the assembly parameters of the abnormal installation node to... The basic characteristics of the component, such as its material and dimensions, remain unchanged.

[0043] Preferably, the spatial scale adjustment parameter γ in step S23 is in the range of 0.5m-2m, and the specific value is determined according to the component installation density of the subway secondary structure.

[0044] Preferably, the attention calculation module in step S41 adopts a multi-head attention mechanism, with the number of attention heads set to 4 and the feature dimension of each attention head being 16.

[0045] Preferably, the set of adjacent nodes M(i) extracted in step S61 includes all installed nodes whose spatial distance from the abnormal node is within 3m, and the number of nodes is not less than 5.

[0046] (III) Beneficial Effects

[0047] Compared with the prior art, the present invention provides a prefabricated design method for the secondary structure of subways, which has the following beneficial effects:

[0048] This invention ensures data quality through multi-dimensional data standardization processing, accurately learns assembly layout based on 3D models and attention convolutional networks, and efficiently corrects node deviations by combining anomaly thresholds and weighted median method, thereby achieving precise, intelligent and safe design. Detailed Implementation

[0049] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0050] A prefabricated design method for secondary structures of subways is proposed, comprising the following steps:

[0051] Step S1: Collect the civil engineering structure dimensions, geological survey data, and electromechanical pipeline layout data of subway stations and sections. Fill the missing values ​​in the data with cubic spline interpolation and perform maximum-minimum standardization on continuous dimension data.

[0052] Step S11: Collect 3D point cloud data of the civil engineering structure of subway station platforms, concourses and tunnels using a 3D laser scanning device, and extract the length, width and height dimensional parameters of the structure;

[0053] Step S12: Extract soil type, foundation bearing capacity, and groundwater level data along the subway line from the geological survey report;

[0054] Step S13: Export the routing, diameter, and installation height data of the subway electromechanical pipelines using BIM modeling software;

[0055] Step S14: Fill in the missing values ​​in the civil engineering structure dimension data using cubic spline interpolation. For the missing dimension parameter point x... t Size value Where a t b t c t d t These are the cubic spline coefficients fitted based on adjacent valid data points;

[0056] Step S15: Perform maximum-min standardization on continuous features such as civil engineering structure dimensions and foundation bearing capacity to map their values ​​to the [0, 1] interval. For category features such as soil type and component material type, use one-hot encoding to convert them into numerical vectors.

[0057] Step S2: Based on the installation point coordinates of the subway secondary structure, construct a three-dimensional mesh topology model. In the model, nodes correspond to the installation positions of prefabricated components, edges represent the spatial connection relationships between components, and edge weights are calculated based on the spatial distance of component installation.

[0058] Step S21: Define the node set V of the 3D mesh topology model, where each node v i For each prefabricated component installation point, the number of nodes N is the total number of prefabricated component installation points for the subway secondary structure.

[0059] Step S22: Construct edges between nodes using the spatial k-nearest neighbor method, setting k=8, and each node connects to the 8 installation nodes with the closest spatial distance;

[0060] Step S23: Node v i With node v j The edge weight w between ij Based on spatial Euclidean distance calculation, the formula is: In the formula, d ij The spatial Euclidean distance between installation nodes i and j is γ, which is a spatial scale adjustment parameter. The value of the spatial scale adjustment parameter γ ranges from 0.5m to 2m, and the specific value is determined according to the installation density of the components of the subway secondary structure.

[0061] Step S24: Construct the adjacency matrix A of the 3D mesh topology model. If there is an edge between node i and node j, then A ij =w ij Otherwise, A ij =0;

[0062] Step S3: Construct a feature vector for each node of the 3D mesh topology model, fusing the material features, size features, and load-bearing features of the components;

[0063] Step S31: Extract the material characteristics of the prefabricated components, including the numerical indicators corresponding to the concrete strength grade and steel grade.

[0064] Step S32: Extract the dimensional features of the component, including the component's length, thickness, and reserved hole size, and calculate the ratio between the dimensional parameters as derived features;

[0065] Step S33: Extract the load-bearing characteristics of the component, obtain the flexural and compressive bearing capacity values ​​of the component through structural mechanics calculations, and use the normalized values ​​as characteristic values;

[0066] Step S34: Concatenate the material features, dimensional features, and load-bearing features along the dimensions to form a node feature vector x. i ∈R m Then, construct the node feature matrix X∈R N×m ;

[0067] Step S4: Design a convolutional neural network model based on the attention mechanism to learn the reasonable assembly layout pattern of the subway secondary structure in a self-supervised manner, and complete the model training by minimizing the layout deviation loss function;

[0068] Step S41: The encoder uses a 3D convolutional layer with an attention mechanism to map the node feature matrix X and adjacency matrix A to the latent feature representation Z. The encoder layer is defined as follows: In the formula, The adjacency matrix for adding self-joins, where I is the identity matrix and D is... The degree matrix, W1 and W2 are the trainable weight matrices of the encoder, and Attention is the attention calculation module;

[0069] Step S42: The decoder uses a deconvolution layer to reconstruct the assembly layout features of the Z nodes from the latent feature representation. The decoder layer is defined as follows: In the formula, W3 is the trainable weight matrix of the decoder, and ConvTranspose3D is the three-dimensional deconvolution operation.

[0070] Step S43: The training process uses the layout bias loss function as the optimization objective, and the loss function is the mean absolute error.

[0071] The formula is: In the formula, N is the total number of nodes, x i Let be the original feature vector of the i-th node. The reconstructed feature vector of the i-th node is used; the model is iteratively trained through backpropagation and stochastic gradient descent optimizer.

[0072] Step S5: Calculate the layout deviation value of each node using the trained model, set an abnormal threshold based on the distribution of historical layout deviation data, and mark the installation nodes with abnormal layout.

[0073] Step S51: Using the trained model, calculate the layout deviation value e for each node of the newly input subway secondary structure assembly data. i The formula is:

[0074] Step S52: Based on the distribution of layout deviation values ​​in historical training data, set an abnormal threshold. Set as the 90th percentile of the layout deviation value;

[0075] Step S53: If the node layout deviation value Then mark the installation node as a layout anomaly node;

[0076] Step S6: For installation nodes marked as abnormal, adjust the assembly parameters of the prefabricated components at the node using the weighted median method, taking into account the assembly information of its adjacent nodes.

[0077] Step S61: For installation nodes v marked as abnormal i Extract the set of neighboring nodes M(i) in the three-dimensional mesh topology model. The extracted set of neighboring nodes M(i) includes all installed nodes whose spatial distance from the abnormal node is within 3m, and the number of nodes is not less than 5.

[0078] Step S62: Adjust only the assembly parameters of the prefabricated components at the abnormal nodes, and calculate the adjusted parameter values ​​using the weighted median method. The formula is as follows: In the formula, Let x be the adjusted assembly parameter vector for the i-th node. j w is the assembly parameter vector of adjacent nodes. ij Let M(i) be the edge weight between nodes i and j, where j∈M(i) represents the weight of all adjacent nodes of node i.

[0079] Step S63: Update the assembly parameters of the abnormal installation node to... The basic characteristics of the component, such as its material and dimensions, remain unchanged.

[0080] 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 alterations 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 prefabricated design method for secondary structures of subways, characterized in that: Includes the following steps: Step S1: Collect the civil engineering structure dimensions, geological survey data, and electromechanical pipeline layout data of subway stations and sections. Fill the missing values ​​in the data with cubic spline interpolation and perform maximum-minimum standardization on continuous dimension data. Step S2: Based on the installation point coordinates of the subway secondary structure, construct a three-dimensional mesh topology model. In the model, nodes correspond to the installation positions of prefabricated components, edges represent the spatial connection relationships between components, and edge weights are calculated based on the spatial distance of component installation. Step S3: Construct a feature vector for each node of the 3D mesh topology model, fusing the material features, size features, and load-bearing features of the components; Step S4: Design a convolutional neural network model based on the attention mechanism to learn the reasonable assembly layout pattern of the subway secondary structure in a self-supervised manner, and complete the model training by minimizing the layout deviation loss function; Step S5: Calculate the layout deviation value of each node using the trained model, set an abnormal threshold based on the distribution of historical layout deviation data, and mark the installation nodes with abnormal layout. Step S6: For installation nodes marked as abnormal, adjust the assembly parameters of the prefabricated components at the node using the weighted median method, taking into account the assembly information of its adjacent nodes.

2. The prefabricated design method for subway secondary structures according to claim 1, characterized in that, Step S1 includes the following steps: Step S11: Collect 3D point cloud data of the civil engineering structure of subway station platforms, concourses and tunnels using a 3D laser scanning device, and extract the length, width and height dimensional parameters of the structure; Step S12: Extract soil type, foundation bearing capacity, and groundwater level data along the subway line from the geological survey report; Step S13: Export the routing, diameter, and installation height data of the subway electromechanical pipelines using BIM modeling software; Step S14: Fill in the missing values ​​in the civil engineering structure dimension data using cubic spline interpolation. For the missing dimension parameter point x... t Size value Where a t b t c t d t These are the cubic spline coefficients fitted based on adjacent valid data points; Step S15: Perform maximum-minimum standardization on continuous features such as civil engineering structure dimensions and foundation bearing capacity to map their values ​​to the [0, 1] interval. For category features such as soil type and component material type, use one-hot encoding to convert them into numerical vectors.

3. The prefabricated design method for subway secondary structures according to claim 1, characterized in that, Step S2 includes the following steps: Step S21: Define the node set V of the 3D mesh topology model, where each node v i For each prefabricated component installation point, the number of nodes N is the total number of prefabricated component installation points for the subway secondary structure. Step S22: Construct edges between nodes using the spatial k-nearest neighbor method, setting k=8, and each node connects to the 8 installation nodes with the closest spatial distance; Step S23: Node v i With node v j The edge weight w between ij Based on spatial Euclidean distance calculation, the formula is: In the formula, d ij γ is the spatial Euclidean distance between installation nodes i and j, and γ is the spatial scale adjustment parameter. Step S24: Construct the adjacency matrix A of the 3D mesh topology model. If there is an edge between node i and node j, then A ij =w ij Otherwise, A ij =0.

4. The prefabricated design method for subway secondary structures according to claim 1, characterized in that, Step S3 includes the following steps: Step S31: Extract the material characteristics of the prefabricated components, including the numerical indicators corresponding to the concrete strength grade and steel grade. Step S32: Extract the dimensional features of the component, including the component's length, thickness, and reserved hole size, and calculate the ratio between the dimensional parameters as derived features; Step S33: Extract the load-bearing characteristics of the component, obtain the flexural and compressive bearing capacity values ​​of the component through structural mechanics calculations, and use the normalized values ​​as characteristic values; Step S34: Concatenate the material features, dimensional features, and load-bearing features along the dimensions to form a node feature vector x. i ∈R m Then, construct the node feature matrix X∈R N×m .

5. The prefabricated design method for subway secondary structures according to claim 1, characterized in that, Step S4 includes the following steps: Step S41: The encoder uses a 3D convolutional layer with an attention mechanism to map the node feature matrix X and adjacency matrix A to the latent feature representation Z. The encoder layer is defined as follows: In the formula, The adjacency matrix for adding self-joins, where I is the identity matrix and D is... The degree matrix, W1 and W2 are the trainable weight matrices of the encoder, and Attention is the attention calculation module; Step S42: The decoder uses a deconvolution layer to reconstruct the assembly layout features of the Z nodes from the latent feature representation. The decoder layer is defined as follows: In the formula, W3 is the trainable weight matrix of the decoder, and ConvTranspose3D is the three-dimensional deconvolution operation. Step S43: The training process uses the layout bias loss function as the optimization objective, and the loss function is the mean absolute error. The formula is: In the formula, N is the total number of nodes, x i Let be the original feature vector of the i-th node. Let be the reconstructed feature vector of the i-th node; train the model iteratively through backpropagation and stochastic gradient descent optimizer.

6. The prefabricated design method for subway secondary structures according to claim 1, characterized in that, Step S5 includes the following steps: Step S51: Using the trained model, calculate the layout deviation value e for each node of the newly input subway secondary structure assembly data. i The formula is: Step S52: Based on the distribution of layout deviation values ​​in historical training data, set an abnormal threshold. Set as the 90th percentile of the layout deviation value; Step S53: If the node layout deviation value Then mark the installation node as a layout anomaly node.

7. The prefabricated design method for subway secondary structures according to claim 1, characterized in that, Step S6 includes the following steps: Step S61: For installation nodes v marked as abnormal i Extract its neighboring node set M(i) in the three-dimensional mesh topology model; Step S62: Adjust only the assembly parameters of the prefabricated components at the abnormal nodes, and calculate the adjusted parameter values ​​using the weighted median method. The formula is as follows: In the formula, Let x be the adjusted assembly parameter vector for the i-th node. j w is the assembly parameter vector of adjacent nodes. ij Let M(i) be the edge weight between nodes i and j, where j∈M(i) represents the weight of all adjacent nodes of node i. Step S63: Update the assembly parameters of the abnormal installation node to... The basic characteristics of the component, such as its material and dimensions, remain unchanged.

8. The prefabricated design method for subway secondary structures according to claim 3, characterized in that, The spatial scale adjustment parameter γ in step S23 has a range of 0.5m-2m, and the specific value is determined according to the component installation density of the subway secondary structure.

9. The prefabricated design method for subway secondary structures according to claim 5, characterized in that, The attention calculation module in step S41 adopts a multi-head attention mechanism, with the number of attention heads set to 4 and the feature dimension of each attention head being 16.

10. The prefabricated design method for subway secondary structures according to claim 7, characterized in that, The set of adjacent nodes M(i) extracted in step S61 includes all installed nodes within a 3m spatial distance from the abnormal node, and the number of nodes is no less than 5.

Images