CN-121997446-A - Subway secondary structure assembly type design method

CN121997446ACN 121997446 ACN121997446 ACN 121997446ACN-121997446-A

Abstract

The invention relates to the technical field of rail transit engineering and discloses an assembly type design method of a secondary structure of a subway, which comprises the following steps of S1, collecting dimension data of a civil structure, geological survey data and electromechanical pipeline layout data of a subway station and a subway section, filling missing values in the data by adopting a cubic spline interpolation method, and carrying out maximum-minimum standardization processing on continuous dimension data, and S2, constructing a three-dimensional grid topological model based on installation point position coordinates of the secondary structure of the subway, wherein nodes in the model correspond to installation positions of assembly type components, edges represent space connection relations among the components, and edge weights are calculated according to space distances of the installation of the components. According to the method, the data quality is guaranteed through multi-dimensional data standard processing, the assembly layout is accurately learned by means of a three-dimensional model and an attention convolution network, and node deviation is efficiently corrected by combining an abnormal threshold value and a weighted median method, so that design accuracy, intellectualization and safety are realized.

Inventors

ZHANG FENG
LI CHENGHU
WANG YAO
ZHENG GUOXING
ZHAO HONGGANG
LI DONG

Assignees

中建海峡建设发展有限公司
福州地铁二号线东延线有限公司
福州地铁六号线东调段有限公司

Dates

Publication Date: 20260508
Application Date: 20251205

Claims (10)

1. A subway secondary structure assembly design method is characterized by comprising the following steps: Step S1, acquiring civil structure size data, geological survey data and electromechanical pipeline layout data of subway stations and sections, filling missing values in the data by adopting a cubic spline interpolation method, and carrying out maximum-minimum standardization processing on continuous size data; S2, constructing a three-dimensional grid topological model based on installation point position coordinates of a subway secondary structure, wherein nodes in the model correspond to installation positions of assembly type components, edges represent space connection relations among the components, and edge weights are calculated according to space distances of component installation; S3, constructing a feature vector for each node of the three-dimensional grid topological model, and fusing the material features, the size features and the load bearing features of the components; S4, designing a convolutional neural network model based on an attention mechanism, learning a reasonable assembly layout mode of subway secondary structure assembly in a self-supervision mode, and completing model training by minimizing a layout deviation loss function; S5, calculating the layout deviation value of each node by using the trained model, setting an abnormal threshold according to the historical layout deviation data distribution, and marking the installation node with abnormal layout; and S6, aiming at the mounting node marked as abnormal, combining the assembly information of the adjacent nodes, and adopting a weighted median method to adjust the assembly parameters of the assembly type components at the node.
2. The method for designing the secondary structure of the subway as set forth in claim 1, wherein the step S1 includes the steps of: Step S11, collecting three-dimensional point cloud data of civil engineering structures of subway station platforms, station halls and interval tunnels through three-dimensional laser scanning equipment, and extracting length, width and height dimension parameters of the structures; S12, extracting soil type, foundation bearing capacity and groundwater level data along the subway line from a geological survey report; step S13, guiding out trend, pipe diameter and installation height data of the subway electromechanical pipeline through BIM modeling software; Step S14, filling missing values in the dimension data of the civil structure by using a cubic spline interpolation method, and filling the missing dimension parameter points x t with the dimension values Wherein a t 、b t 、c t 、d t is a cubic spline coefficient fitted based on adjacent valid data points; And S15, carrying out maximum-minimum standardization on continuous features such as civil engineering structure size, foundation bearing capacity and the like, mapping numerical values of the continuous features to a [0,1] interval, and converting single-heat codes into numerical value vectors for class features such as soil types, component material types and the like.
3. The method for designing the secondary structure of the subway as set forth in claim 1, wherein the step S2 includes the steps of: step S21, defining a node set V of a three-dimensional grid topological model, wherein each node V i corresponds to one assembly type component installation point, and the number N of nodes is the total number of assembly type component installation points of the subway secondary structure; step S22, constructing edges among nodes by adopting a space k nearest neighbor method, setting k=8, and connecting 8 mounting nodes with the nearest space distance to each node; Step S23, edge weight w ij between node v i and node v j is calculated based on the spatial Euclidean distance, and the formula is: Wherein d ij is the space Euclidean distance between the installation nodes i and j, and gamma is the space scale adjustment parameter; Step S24, constructing an adjacency matrix A of the three-dimensional grid topological model, wherein if an edge exists between a node i and a node j, A ij ＝w ij is formed, otherwise A ij =0.
4. The method for designing the secondary structure of the subway as set forth in claim 1, wherein the step S3 includes the steps of: S31, extracting material characteristics of the assembled component, wherein the material characteristics comprise numerical indexes corresponding to concrete strength grades and steel brands; S32, extracting dimensional characteristics of the component, including the length, the thickness and the reserved hole size of the component, and calculating the ratio of the dimensional parameters as derivative characteristics; s33, extracting load bearing characteristics of the component, calculating and obtaining the fracture resistance and compression resistance bearing capacity values of the component through structural mechanics, and adopting normalization processing as characteristic values; And step S34, splicing the material characteristics, the size characteristics and the load bearing characteristics according to dimensions to form a node characteristic vector X i ∈R m , and further constructing a node characteristic matrix X epsilon R N×m .
5. The method for designing the secondary structure of the subway as set forth in claim 1, wherein the step S4 includes the steps of: step S41, the encoder adopts a three-dimensional convolution layer of an attention mechanism to map a node characteristic matrix X and an adjacent matrix A into a potential characteristic representation Z, and the encoder layer is defined as: in the formula, To add a self-connecting adjacency matrix, I is an identity matrix, D is W 1 、W 2 is a trainable weight matrix of the encoder, and Attention is an Attention calculation module; step S42-the decoder uses deconvolution layers to represent the assembly layout features of the Z-reconstructed node from the potential features The decoder layer is defined as: Wherein W 3 is a trainable weight matrix of the decoder, and ConvTranspose D is three-dimensional deconvolution operation; Step S43, the training process takes the layout deviation loss function as an optimization target, the loss function adopts average absolute error, The formula is: Where N is the total number of nodes, x i is the original feature vector of the ith node, Reconstructing feature vectors for the ith node, and iteratively training the model by a back-propagation and random gradient descent optimizer.
6. The method for designing the secondary structure of the subway as set forth in claim 1, wherein the step S5 includes the steps of: Step S51, calculating a layout deviation value e i of each node by utilizing the trained model to assemble data of the newly input subway secondary structure, wherein the formula is as follows: step S52, based on the distribution of the layout deviation values of the historical training data, the abnormal threshold value is set Setting the layout deviation value as 90 minutes; step S53, if the layout deviation value of the node The installation node is marked as a layout exception node.
7. The method for designing the secondary structure of the subway as set forth in claim 1, wherein the step S6 includes the steps of: Step S61, extracting a neighboring node set M (i) of the installation node v i marked as abnormal in the three-dimensional grid topology model; Step S62, only adjusting assembly parameters of assembly type components at abnormal nodes, and calculating adjusted parameter values by adopting a weighted median method, wherein the formula is as follows: in the formula, For the adjusted assembly parameter vector of the ith node, x j is the assembly parameter vector of the adjacent node, w ij is the edge weight between the nodes i and j, and j epsilon M (i) is all the adjacent nodes traversing the node i; step S63, updating the assembly parameters of the abnormal installation node to The material and size basic characteristics of the component remain unchanged.
8. The method for designing the secondary structure of the subway as set forth in claim 3, wherein the value range of the spatial scale adjustment parameter γ in the step S23 is 0.5m-2m, and the specific value is determined according to the installation density of the components of the secondary structure of the subway.
9. The subway secondary structure assembly design method according to claim 5, wherein the attention calculating module in step S41 adopts a multi-head attention mechanism, the number of attention heads is set to 4, and the feature dimension of each attention head is 16.
10. The method according to claim 7, wherein the set of neighboring nodes M (i) extracted in the step S61 includes all installed nodes having a spatial distance within 3M from the abnormal node, and the number of the nodes is not less than 5.

Description

Subway secondary structure assembly type design method Technical Field The invention relates to the technical field of rail transit engineering, in particular to an assembly type design method for a secondary structure of a subway. Background In the field of subway secondary structure assembly design, the traditional design method has various key problems that on one hand, the data sources such as the civil structure size, geological investigation, electromechanical pipelines and the like of subway stations and intervals are scattered and easy to lose, meanwhile, continuous data dimension differences are large, category characteristics are difficult to quantify, so that data quality is uneven, accurate and unified data support cannot be provided for the design, on the other hand, the traditional design relies on manual experience to construct component assembly layout, space connection relation among components is difficult to intuitively describe, multidimensional characteristics such as component materials, sizes and load bearing cannot be comprehensively fused, the problem of insufficient layout rationality easily occurs, and moreover, objective and efficient identification standards and correction means are lacked on abnormal nodes in the assembly layout, so that the subjectivity is high in manual judgment, the correction precision is low, the design efficiency is influenced, and hidden hazards of safety and stability of the subway secondary structure are also possible, and the requirements of modern subway engineering on assembly design are difficult to meet. Therefore, we propose a subway secondary structure assembly design method to solve the problem. Disclosure of Invention (One) solving the technical problems Aiming at the defects of the prior art, the invention provides a subway secondary structure assembly type design method, which solves the problems in the background art. (II) technical scheme In order to achieve the purpose, the invention provides the following technical scheme that the subway secondary structure assembly type design method comprises the following steps: Step S1, acquiring civil structure size data, geological survey data and electromechanical pipeline layout data of subway stations and sections, filling missing values in the data by adopting a cubic spline interpolation method, and carrying out maximum-minimum standardization processing on continuous size data; S2, constructing a three-dimensional grid topological model based on installation point position coordinates of a subway secondary structure, wherein nodes in the model correspond to installation positions of assembly type components, edges represent space connection relations among the components, and edge weights are calculated according to space distances of component installation; S3, constructing a feature vector for each node of the three-dimensional grid topological model, and fusing the material features, the size features and the load bearing features of the components; S4, designing a convolutional neural network model based on an attention mechanism, learning a reasonable assembly layout mode of subway secondary structure assembly in a self-supervision mode, and completing model training by minimizing a layout deviation loss function; S5, calculating the layout deviation value of each node by using the trained model, setting an abnormal threshold according to the historical layout deviation data distribution, and marking the installation node with abnormal layout; and S6, aiming at the mounting node marked as abnormal, combining the assembly information of the adjacent nodes, and adopting a weighted median method to adjust the assembly parameters of the assembly type components at the node. Preferably, the step S1 includes the following steps: Step S11, collecting three-dimensional point cloud data of civil engineering structures of subway station platforms, station halls and interval tunnels through three-dimensional laser scanning equipment, and extracting length, width and height dimension parameters of the structures; S12, extracting soil type, foundation bearing capacity and groundwater level data along the subway line from a geological survey report; step S13, guiding out trend, pipe diameter and installation height data of the subway electromechanical pipeline through BIM modeling software; Step S14, filling missing values in the dimension data of the civil structure by using a cubic spline interpolation method, and filling the missing dimension parameter points x t with the dimension values Wherein a t、bt、ct、dt is a cubic spline coefficient fitted based on adjacent valid data points; And S15, carrying out maximum-minimum standardization on continuous features such as civil engineering structure size, foundation bearing capacity and the like, mapping numerical values of the continuous features to a [0,1] interval, and converting single-heat codes into numerical value vectors for class features such as