Steel structure construction area division method based on space calculation

Abstract

The invention relates to the technical field of space calculation, and particularly discloses a steel structure construction area division method based on space calculation, which comprises the following steps: acquiring a design model of a steel structure and an initial space constraint condition of a construction site; the method comprises the following steps: acquiring deformation sensing data through a sensing network deployed on a sliding track, and calculating a dynamic deformation field of the sliding track based on the deformation sensing data; performing space mapping processing on the initial space constraint condition according to the dynamic deformation field to generate a dynamic security boundary; and outputting a construction area division map containing the dynamic safety boundary. The space coordination and the overall construction safety in the sliding assembly process of the steel structure are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of space computing, and in particular to a steel structure construction area division method based on space computing. BACKGROUND

[0002] Space computing is a technical method in the construction field that takes three-dimensional space as the core, integrates building information model, site environment data and construction element dynamic parameters. It converts the static space of design blueprint into dynamic logic of the construction process by analyzing spatial geometric relationships, mechanical constraints and timing coordination rules, supports decisions such as regional division, equipment operation range planning, and process connection boundary definition, and realizes precise matching of resources in the physical space and time dimension. It is a digital bridge connecting design intent and site implementation, and guarantees the orderly and efficient progress of construction.

[0003] When the assembly area, sliding area and positioning area are divided by space computing, the potential spatial interference risk is caused due to the failure to consider the dynamic contact deformation between the truss and the track during the sliding process. The core reason is that the calculation model usually regards the sliding track as a rigid body and divides the area boundary statically according to it; however, in actual construction, the huge self-weight of the truss will cause the track to produce contact elastic deformation, causing the actual sliding track to deviate from the design center line. With the increase of sliding distance, this deviation accumulates, causing the originally reserved safety distance to be gradually eroded, eventually leading to spatial interference between the newly assembled segmented structure and the structure that has completed sliding, seriously affecting the construction precision and safety. SUMMARY

[0004] The purpose of the present application is to provide a steel structure construction area division method based on space computing, which solves the above technical problems.

[0005] The purpose of the present application can be achieved by the following technical solutions:

[0006] A steel structure construction area division method based on space computing, comprising the following steps:

[0007] Obtaining a design model of a steel structure and initial spatial constraint conditions of a construction site, the design model being a digital three-dimensional model containing the geometric profile of a truss, and the initial spatial constraint conditions being a set of spatial coordinates of the boundaries of an assembly area, a sliding area and a positioning area;

[0008] Collecting deformation perception data through a sensor network deployed on the sliding track, and calculating a dynamic deformation field of the sliding track based on the deformation perception data;

[0009] Performing spatial mapping processing on the initial spatial constraint conditions according to the dynamic deformation field to generate a dynamic safety boundary;

[0010] The output comprises a construction area division map of the dynamic security boundary.

[0011] Preferably, the deformation sensing data is collected by a sensor network deployed on the sliding track, comprising:

[0012] The stress distribution of the sliding track under the load of the truss is calculated by a finite element method, and a high stress area on the sliding track is identified based on the stress distribution;

[0013] Points in the high stress area with stress values exceeding a preset threshold are taken as key support points, and a sensor node of the sensor network is deployed at each key support point, the sensor node comprising a laser range finder and an inclination sensor;

[0014] The vertical displacement change of the sliding track is collected by the laser range finder deployed at the key support point, and the local inclination change of the sliding track is collected by the inclination sensor deployed at the key support point;

[0015] All the vertical displacement changes and the local inclination changes are data fused based on a Kalman filtering algorithm to obtain the deformation sensing data.

[0016] Preferably, the deformation sensing data is collected by a sensor network deployed on the sliding track, comprising:

[0017] The stress distribution of the sliding track under the load of the truss is calculated by a finite element method, and a high stress area on the sliding track is identified based on the stress distribution;

[0018] Points in the high stress area with stress values exceeding a preset threshold are taken as key support points, and a sensor node of the sensor network is deployed at each key support point, the sensor node comprising a laser range finder and an inclination sensor;

[0019] The vertical displacement change of the sliding track is collected by the laser range finder deployed at the key support point, and the local inclination change of the sliding track is collected by the inclination sensor deployed at the key support point;

[0020] All the vertical displacement changes and the local inclination changes are data fused based on a Kalman filtering algorithm to obtain the deformation sensing data.

[0021] Preferably, the dynamic deformation field of the sliding track is calculated further comprising:

[0022] Prepare a training data set, the training data set contains a plurality of groups of samples, the samples are constructed based on historical construction process, each group of samples includes: historical input sequence, slip track material attribute parameters at target points and truss load distribution parameters and actual spatial displacement vectors of all target points;

[0023] Construct a time series-based recurrent neural network, initialize the weight parameters and bias parameters of the recurrent neural network, input the samples into the recurrent neural network, and obtain the predicted spatial displacement vectors of all target points through forward propagation calculation;

[0024] Calculate the loss function value based on the overall difference between the predicted spatial displacement vectors and the measured spatial displacement vectors at all target points, and the loss function adopts a mean square error function;

[0025] Calculate the gradient of the loss function with respect to each weight parameter of the recurrent neural network through the back propagation algorithm, and update the weight parameters of the recurrent neural network according to the gradient using the gradient descent optimization algorithm;

[0026] Repeat the above steps until the loss function value converges to a preset threshold, and obtain a trained deformation prediction model.

[0027] Preferably, generating a dynamic safety boundary comprises:

[0028] Based on the spatial displacement vectors of all target points in the dynamic deformation field, the spatial displacement vectors at each position of the assembly area boundary, the slip area boundary and the just-in-place area boundary are determined by a spatial interpolation algorithm;

[0029] Perform vector addition operation on the spatial coordinates of the assembly area boundary, the slip area boundary and the just-in-place area boundary and the corresponding spatial displacement vectors, and obtain the spatial coordinates of the modified assembly area boundary, the slip area boundary and the just-in-place area boundary through the vector addition operation, to form the dynamic safety boundary.

[0030] Preferably, after generating the dynamic safety boundary, a dynamic spatial interference detection step is further included:

[0031] Based on the truss geometric contour data in the design model, an outer envelope model of the truss is constructed, and the outer envelope model is the smallest convex polyhedron generated by calculating the convex hull of all vertices of the truss;

[0032] Calculate the minimum spatial distance between the outer envelope model and the dynamic safety boundary when the outer envelope model moves along the modified slip path, and the calculation process is completed by solving the Euclidean distance between the surface points of the outer envelope model and the surface points of the dynamic safety boundary and taking the minimum value;

[0033] comparing the minimum spatial distance with a preset safety threshold;

[0034] triggering a boundary re-partition procedure when the minimum spatial distance is not greater than the safety threshold.

[0035] Preferably, the boundary re-partition procedure comprises:

[0036] applying a set of known test loads on the sliding track, collecting corresponding deformation sensing data through the sensing network, and measuring spatial displacement vectors of all target points to form a calibration data set;

[0037] based on the calibration data set, establishing a response mapping function through multiple linear regression analysis, the response mapping function taking spatial displacement vectors of all target points as input and deformation sensing data as output;

[0038] taking the input sequence as a first input sequence and the corresponding dynamic deformation field as a first deformation field, and extracting spatial displacement vectors of all target points in the first deformation field;

[0039] inputting the extracted spatial displacement vectors of all target points into the response mapping function to obtain simulated deformation sensing data A;

[0040] removing deformation sensing data of the earliest time step in the first input sequence and taking the deformation sensing data A as new deformation sensing data of the nearest time step to combine into a new first input sequence B;

[0041] based on the first input sequence B, obtaining a new first deformation field B1, repeating the above steps until a specified number of times is reached, and constructing a dynamic deformation field evolution trend in chronological order from the new first deformation field obtained each time.

[0042] Preferably, the boundary re-partition procedure further comprises:

[0043] extracting a last first deformation field C from the dynamic deformation field evolution trend;

[0044] based on spatial displacement vectors of all target points in the first deformation field C, determining spatial displacement vectors at each position of the assembly area boundary through a spatial interpolation algorithm, denoted as first vectors;

[0045] calculating an arithmetic mean of the lengths of all first vectors as a boundary adjustment amount;

[0046] calculating a vector sum of all first vectors, and taking the direction of the unitized vector sum as a boundary adjustment direction;

[0047] based on the boundary adjustment amount and the boundary adjustment direction, constructing a translation transformation matrix;

[0048] Transform the spatial coordinates of all positions of the assembly area boundary through the translation transformation matrix to generate an updated dynamic safety boundary.

[0049] The beneficial effects of the present application are that, compared with the prior art:

[0050] The present application realizes the adaptive adjustment of the assembly area, the sliding area and the in-place area in the steel structure construction process by introducing a dynamic area division method based on spatial calculation. By obtaining the deformation information of the sliding track in real time and establishing a dynamic deformation field, the boundaries of each area can be automatically corrected when the spatial constraint conditions change, so that the division result is consistent with the actual stress state. This method can reflect the elastic deformation of the track and the track offset caused by the elastic deformation in the process of truss sliding, effectively avoiding the spatial division error caused by the non-rigid characteristics of the track. Through the generation and real-time interference detection of the dynamic safety boundary, the construction area can be intelligently re-divided according to the stress state of the structure, ensuring that the truss sliding and assembly operations maintain a safe distance and prevent spatial interference. At the same time, the continuous evolution modeling of the boundary in the method makes the division result have time sequence continuity, ensuring the smoothness of the construction path and the positioning accuracy. In summary, the present application realizes the dynamic, accurate and safe division of the construction area, and improves the spatial coordination and overall construction safety of the steel structure sliding and assembly process. BRIEF DESCRIPTION OF DRAWINGS

[0051] The present application will be further described below in conjunction with the accompanying drawings.

[0052] Figure 1 is a flowchart of a steel structure construction area division method based on spatial calculation. DETAILED DESCRIPTION

[0053] The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0054] Please refer to Figure 1 The present application is a steel structure construction area division method based on spatial calculation, which comprises the following steps:

[0055] Obtain the design model of the steel structure and the initial spatial constraint conditions of the construction site. The design model is a digital three-dimensional model containing the geometric profile of the truss, and the initial spatial constraint conditions are a set of spatial coordinates of the boundaries of the assembly area, the sliding area and the in-place area defined in advance.

[0056] The deformation sensing data is collected by a sensor network deployed on the sliding track, and a dynamic deformation field of the sliding track is calculated based on the deformation sensing data.

[0057] In a preferred embodiment of the present application, the deformation sensing data collected by the sensor network deployed on the sliding track includes:

[0058] The stress distribution of the sliding track under the load of the truss is calculated by a finite element method, and a high stress area on the sliding track is identified based on the stress distribution.

[0059] Points with stress values exceeding a preset threshold in the high stress area are taken as key support points, and a sensor node of the sensor network is deployed at each key support point, the sensor node including a laser ranging device and an inclination sensor.

[0060] The vertical displacement variation of the sliding track is collected by the laser range finder deployed at the key support point, and the local inclination variation of the sliding track is collected by the inclination sensor deployed at the key support point.

[0061] Based on a Kalman filtering algorithm, all the vertical displacement variations and the local inclination variations are data fused to obtain the deformation sensing data.

[0062] It should be noted that the sliding track will appear non-uniform elastic deformation under the action of the self-weight and load of the truss, and if only the design value or single measurement point data is used, the overall stress and deformation characteristics cannot be reflected. By using the finite element method, the track structure can be discretized in the calculation model, the external load can be converted into the stress response of each element node, and then the stress concentration area, i.e. the high stress area, can be identified in the calculation result. The high stress area often corresponds to the position where the track structure is most likely to deform, so deploying sensor nodes in these areas can effectively capture the overall deformation trend with fewer sensor points. The laser range finder in the sensor node is used to measure the vertical displacement variation of the track, and the inclination sensor is used to measure the local angular offset of the track, which respectively represent the linear flexure and rotational inclination of the track.

[0063] Due to environmental noise, equipment precision and real-time dynamic loading and other factors, a single sensing quantity is difficult to directly reflect the real deformation state, and therefore a Kalman filtering algorithm is introduced to fuse multi-source sensing data. The Kalman filtering algorithm inputs the displacement change obtained by laser ranging and the angle change obtained by the tilt sensor as observation values, combines the prediction model and the covariance matrix of the system to perform weighted correction on each observation data, and thereby outputs the estimated value most consistent with the real state at each time step. The filtered result can smooth out random noise and retain the dynamic trend of the track deformation, and finally form continuous and reliable deformation sensing data. The data can accurately depict the space-time distribution of the force and deformation of the track during the truss sliding process, provide basic input for subsequent calculation of the dynamic deformation field, enable the construction area division to be dynamically adjusted according to the real structure response, ensure that the division boundary is consistent with the actual deformation state, and reduce the risk of spatial interference.

[0064] In another preferred embodiment of the present application, the dynamic deformation field of the sliding track comprises:

[0065] A series of fixed target points are defined on the surface of the sliding track, and the material property parameters and truss load distribution parameters of the sliding track at each target point are obtained, the material property parameters of the sliding track include the elastic modulus and the Poisson's ratio, and the truss load distribution parameters include the truss self-weight and the load action point position.

[0066] The deformation sensing data collected at the current and historical continuous multiple time steps are combined in time sequence as an input sequence, and the input sequence, the material property parameters and the load distribution parameters of all target points are jointly input into the pre-trained deformation prediction model, and the spatial offset vector of all target points is synchronously output through the deformation prediction model, and the spatial offset vector is a vector containing size and direction information.

[0067] The set of all target points and their spatial offset vectors is taken as the dynamic deformation field.

[0068] In a preferred case of the present embodiment, the dynamic deformation field of the sliding track further comprises:

[0069] A training data set is prepared, the training data set contains multiple samples, the samples are constructed based on the historical construction process, and each sample includes: the historical input sequence, the material property parameters and the load distribution parameters of the sliding track at the target points, and the actual spatial offset vector of all target points.

[0070] A recurrent neural network based on time series is constructed, the weight parameters and bias parameters of the recurrent neural network are initialized, the samples are input into the recurrent neural network, and the predicted spatial offset vector of all target points is synchronously obtained through forward propagation calculation.

[0071] The loss function value is calculated based on the overall difference between the predicted spatial displacement vector and the measured spatial displacement vector on all target points, and the loss function adopts a mean square error function.

[0072] The gradient of the loss function with respect to each weight parameter of the recurrent neural network is calculated by a back propagation algorithm, and the weight parameters of the recurrent neural network are updated according to the gradient using a gradient descent optimization algorithm.

[0073] The above steps are repeated until the loss function value converges to a preset threshold, and a trained deformation prediction model is obtained.

[0074] It can be understood that the core of the above process is to establish a mapping relationship between the stress state of the sliding track and the spatial deformation by the deformation prediction model, so that the deformation of the track under the action of dynamic load can be inferred and reconstructed in a time sequence manner, thereby forming a continuous dynamic deformation field.

[0075] The deformation of the sliding track is not a static quantity, but a spatial process that evolves with time and load, so it is necessary to define a number of target points on the track surface as discrete deformation observation units to describe the continuous deformation characteristics of the entire track. The material property parameters of each target point determine its response capability under external force, while the load distribution parameters reflect the local action strength of the truss structure on the track, both of which jointly affect the deformation amplitude and direction of the target point; Specifically, the elastic modulus represents the ability of the sliding track material to resist deformation after being stressed, which is a quantitative indicator of material stiffness, and the larger the value, the smaller the strain produced under the same stress; Poisson's ratio describes the proportion of transverse deformation perpendicular to the direction of tension or compression of the material, which reflects the volume deformation characteristics of the material, and the two parameters together determine the elastic response characteristics of the track under stress. The self-weight of the truss refers to the constant distributed load on the sliding track due to the gravity of the truss structure, reflecting the basic stress that the track has been subjected to for a long time; the load action point position represents the specific spatial coordinate position of the load applied to the track by the truss or external equipment, determining the distribution form and size of the stress on the track. The four together describe the basic physical relationship between track stress and deformation, providing the necessary mechanical constraints for the deformation prediction model, so that the calculated dynamic deformation field can conform to the material properties and actual load distribution law.

[0076] By introducing input data in the form of time series, the state evolution law of the track between different time steps can be captured. The reason why recurrent neural networks are suitable for this problem is that they have memory cells inside, which can save the information of the previous state in the time dimension, so that the continuous influence of historical loading on the system is considered when predicting the deformation at the current time.

[0077] In the training stage, the model learns the nonlinear mapping relationship between the input variables and the target point space displacement vector through a large number of historical samples, and the mean square error is used as the loss function to quantify the overall deviation between the predicted results and the measured values. By calculating the gradient through back propagation and constantly adjusting the network weights, the model gradually approaches the real dynamic response of the track. After training is completed, the model can output the spatial displacement vector of each target point after inputting new perception data in real time, and these displacement vectors constitute the dynamic deformation field of the sliding track. The deformation field can reflect the deformation direction and degree of each part of the track in space and reflect the continuous evolution of the deformation in time, providing accurate basic information for the generation of the dynamic safety boundary, so that the construction area division can be consistent with the actual structure deformation state, ensuring the spatial precision and safety of the sliding process.

[0078] According to the dynamic deformation field, the initial spatial constraint condition is subjected to spatial mapping processing to generate a dynamic safety boundary.

[0079] In another preferred embodiment of the present application, generating a dynamic safety boundary comprises:

[0080] Based on the spatial displacement vectors of all target points in the dynamic deformation field, the spatial displacement vectors at each position of the assembly area boundary, the sliding area boundary and the in-place area boundary are determined through a spatial interpolation algorithm.

[0081] The spatial coordinates of the assembly area boundary, the sliding area boundary and the in-place area boundary are subjected to vector addition operation with the corresponding spatial displacement vectors, and the spatial coordinates of the corrected assembly area boundary, the sliding area boundary and the in-place area boundary are obtained through vector addition operation to constitute the dynamic safety boundary.

[0082] It should be noted that the real deformation result of the track reflected in the dynamic deformation field is converted into the dynamic correction of the spatial boundary, so that the division of the construction area is no longer based on the ideal static model, but can be updated in real time with the structure deformation.

[0083] The spatial offset vector of each target point in the dynamic deformation field represents the local displacement and direction change of the track under the action of force, but the boundary itself is usually not directly laid out with sensing points, so it is necessary to establish a continuous displacement distribution relationship between the target points through a spatial interpolation algorithm. The principle of spatial interpolation is to use the offset data of known points to calculate the offset of unknown points through mathematical functions or weighted calculation methods, so that the deformation distribution of the entire boundary region is expanded from discrete points to a continuous field. In this way, the spatial correction amount of any point on the boundary can be ensured to be consistent with the actual deformation trend of the track. After obtaining the offset vectors of the points on the boundary, vector addition is performed on these vectors and the spatial coordinates of the original boundary, which is equivalent to translating each boundary point in the three-dimensional space according to its deformation direction and amplitude to form a new modified boundary. This process realizes the spatial mapping conversion of the design boundary to the boundary in the actual deformation state, so that the boundary line or boundary surface of the assembly area, the sliding area and the in-place area can be adjusted adaptively with the track deformation. Through this processing, the division result can accurately reflect the spatial offset caused by the stress change of the structure, thereby maintaining the consistency of the construction safety zone and the actual movement path, and avoiding spatial interference caused by boundary distortion of the truss during the sliding or assembly process due to track deformation.

[0084] In another preferred embodiment of the present application, after generating the dynamic safety boundary, a dynamic spatial interference detection step is further included:

[0085] An outer envelope model of the truss is constructed based on the geometric contour data of the truss in the design model, and the outer envelope model is the smallest convex polyhedron generated by calculating the convex hull of all vertices of the truss;

[0086] The minimum spatial distance between the outer envelope model moving along the modified sliding path and the dynamic safety boundary is calculated, and the calculation process is completed by solving the Euclidean distance between the surface points of the outer envelope model and the surface points of the dynamic safety boundary and taking the minimum value.

[0087] The minimum spatial distance is compared with a preset safety threshold.

[0088] When the minimum spatial distance is not greater than the safety threshold, a boundary re-division process is triggered.

[0089] It is worth noting that after generating the dynamic safety boundary, the spatial relationship between the truss and the boundary is continuously checked, the truss geometry profile in the design model is constructed as a minimum convex polyhedron envelope in a convex hull manner, so that the actual space occupied by the truss has clear and conservative shape limits in calculation; then the envelope is moved along the corrected sliding path, the Euclidean distance between the surface points of the envelope and the surface points of the dynamic safety boundary is calculated at each position and the minimum value is taken, so that the most unfavorable approaching degree of the truss movement in the whole process is obtained. The minimum distance is compared with the preset safety threshold, and the potential interference trend can be identified in time when the distance is reduced to the threshold and below, so that the boundary redivision can be triggered according to the established process. The convex hull shape is used for unified expression of the truss profile, the minimum distance is used for quantization of the spatial margin of the truss and the boundary, and the threshold criterion is used to form a clear decision trigger condition, so that the regional boundary can be updated in time when the risk is detected, the division result can be kept consistent with the actual movement path under dynamic deformation, and interference caused by excessive spatial approach in the sliding process is avoided.

[0090] In a preferred case of the embodiment, after generating the dynamic safety boundary, a dynamic spatial interference detection step is further included:

[0091] An envelope model of the truss is constructed based on the truss geometry profile data in the design model, and the envelope model is a minimum convex polyhedron generated by calculating the convex hull of all vertices of the truss.

[0092] The minimum spatial distance between the envelope model and the dynamic safety boundary is calculated when the envelope model moves along the corrected sliding path, and the calculation process is completed by solving the Euclidean distance between the surface points of the envelope model and the surface points of the dynamic safety boundary and taking the minimum value.

[0093] The minimum spatial distance is compared with a preset safety threshold.

[0094] When the minimum spatial distance is not greater than the safety threshold, the boundary redivision process is triggered.

[0095] In a preferred embodiment of the application, the boundary redivision process includes:

[0096] A set of known test loads are applied on the sliding track, corresponding deformation sensing data are collected through a sensing network, and spatial displacement vectors of all target points are measured to form a calibration data set.

[0097] Based on the calibration data set, a response mapping function is established through multivariate linear regression analysis, the response mapping function takes the spatial displacement vectors of all target points as input and takes the deformation sensing data as output.

[0098] The input sequence is taken as a first input sequence, the corresponding dynamic deformation field is taken as a first deformation field, and the spatial displacement vectors of all target points in the first deformation field are extracted.

[0099] The spatial offset vectors of all extracted target points are input into the response mapping function to obtain simulated deformation sensing data A.

[0100] Remove the deformation sensing data from the earliest time step in the first input sequence, and use deformation sensing data A as the new deformation sensing data from the most recent time step to form a new first input sequence B.

[0101] Based on the first input sequence B, a new first deformation field B1 is obtained. The above steps are repeated until a specified number of times are reached. The new first deformation fields obtained each time are arranged in chronological order to form the dynamic deformation field evolution trend.

[0102] It should be noted that the boundary re-delineation process also includes:

[0103] Extract the last first deformation field C from the evolution trend of the dynamic deformation field.

[0104] Based on the spatial offset vectors of all target points in the first deformation field C, the spatial offset vectors at each point on the boundary of the assembly area are determined by a spatial interpolation algorithm and denoted as the first vector.

[0105] Calculate the arithmetic mean of the magnitudes of all first vectors as the boundary adjustment.

[0106] Calculate the vector sum of all first vectors, and use the normalized direction of the vector sum as the boundary to adjust the direction.

[0107] A translation transformation matrix is ​​constructed based on the boundary adjustment amount and the boundary adjustment direction.

[0108] The spatial coordinates of all positions on the assembly area boundary are transformed by a translation transformation matrix to generate an updated dynamic safety boundary.

[0109] It should be noted that in the boundary re-division process, by calibrating the track's response under known test loads, a statistical correlation is established between the deformation data collected by the sensor network and the actual spatial offset of the target point, thereby obtaining a response mapping function that reflects the mechanical response law of the track. Multiple linear regression plays a role in extracting parameter relationships here; it searches for the linear mapping in the high-dimensional input space that best describes the deformation change trend, enabling the model to predict the corresponding deformation data based on the spatial offset of the target point.

[0110] Then, based on this function, the input sequence is iteratively replaced and updated. By continuously calculating new deformation fields, a dynamic trend of deformation evolution over time is formed. This iterative process is essentially a state recursion, enabling the model to simulate the development direction and rate of change of track deformation in future moments. As the evolution trend is established, the offset vector of the last deformation field represents the system reaching a new stable state. Using these offset vectors, the continuous offset distribution of the assembly area boundary can be obtained through spatial interpolation. The average magnitude of all offset vectors is calculated to obtain the overall boundary translation scale, and the direction of the vector sum is normalized to reflect the overall offset direction.

[0111] The resulting translation transformation matrix mathematically equates to a complete translation correction of the boundary along the principal force direction, ensuring that the correction amount matches the actual structural deformation. The transformed dynamic safety boundary re-aligns with the current stress state, maintaining a reasonable safety distance even after significant changes in structural response, thus guaranteeing continuous spatial coordination during truss sliding and assembly processes.

[0112] Output a construction zone delineation map that includes dynamic safety boundaries.

[0113] It is worth noting that the calculated dynamic safety boundary is presented as a construction area division map in a visual manner, transforming the dynamic spatial calculation results into spatial information that can be directly used for construction guidance. The principle is to map the corrected boundary spatial coordinates onto a three-dimensional coordinate system and display the assembly area, sliding area, and positioning area separately, thus intuitively reflecting the real-time positional relationship of each area under stress and deformation. By outputting this division map, construction personnel can accurately grasp the safe operating range and structural sliding path, enabling effective connection between spatial calculation results and construction control, ensuring consistency between area division and the dynamic state of the structure.

[0114] The overall concept of this invention is to use spatial computation as the core, transforming the division of the steel structure construction area from a static geometric constraint to a dynamic adjustment process based on the actual structural response. Traditional division methods assume the sliding track is a rigid body, ignoring the elastic deformation of the track during truss sliding, causing the actual sliding path to gradually deviate from the design centerline, ultimately leading to spatial interference. This invention first acquires track deformation information in real time through a sensor network, constructing a dynamic deformation field that reflects changes in the stress state. Then, through spatial mapping and boundary re-division steps, this mechanical response is fed back to the region boundary, thereby forming a dynamic safety boundary that can automatically correct itself according to changes in the structural state, ensuring that the spatial relationship between the assembly area, sliding area, and positioning area always matches the actual structural state.

[0115] This invention achieves dynamic spatial partitioning through a three-stage closed loop of "sensing-computation-correction". In the sensing stage, finite element analysis is used to identify high-stress regions and deploy sensor nodes. A multi-source fusion algorithm is used to obtain the deformation information of the track in multi-dimensional space, ensuring data integrity and accuracy. In the computation stage, a time-series model is introduced to model the evolution of deformation over time, enabling the system to not only reflect the current state but also predict the trend of the track under continuous stress. In the correction stage, spatial interpolation and vector operations are used to transform the impact of track deformation on the region boundary into coordinate changes, geometrically achieving dynamic updates of the region boundary. In this way, the originally static boundary model is endowed with time-varying characteristics, and the system can automatically adjust the safety zone range according to structural stress and deformation, fundamentally eliminating the problem of safety clearance reduction caused by the accumulation of track deformation.

[0116] The key to this invention lies in the combination of deformation sensing and prediction models. Real-time measurements alone are insufficient to capture the continuous evolution of deformation, while simple prediction models lack real data support. The combination of both ensures the system possesses both physical realism and temporal continuity. Kalman filtering is used to achieve multi-source data fusion, eliminating environmental noise and maintaining dynamic smoothness. A recurrent neural network is used to establish a deformation prediction model, enabling the calculation of the deformation field to consider historical loading effects and avoiding boundary misjudgments caused by transient data fluctuations. Furthermore, dynamic spatial interferometry detection and boundary re-division mechanisms ensure that the system can automatically identify and correct boundaries when the safety margin decreases, maintaining the stability and safety of the boundary division results.

[0117] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the present invention should still fall within the scope of the present invention.

Claims

1. A method for dividing steel structure construction areas based on spatial calculation, characterized in that, Includes the following steps: Obtain the design model of the steel structure and the initial spatial constraints of the construction site. The design model is a digital three-dimensional model containing the geometric contour of the truss, and the initial spatial constraints are a set of predefined spatial coordinates of the assembly area boundary, the sliding area boundary, and the positioning area boundary. Deformation sensing data is collected by a sensor network deployed on the sliding track, and the dynamic deformation field of the sliding track is calculated based on the deformation sensing data. The initial spatial constraints are spatially mapped based on the dynamic deformation field to generate a dynamic safety boundary. Output a construction area delineation map that includes the dynamic safety boundary.

2. The method for dividing steel structure construction areas based on spatial calculation according to claim 1, characterized in that, Deformation sensing data is collected through a sensor network deployed on the sliding track, including: The stress distribution of the sliding track under the load of the truss is calculated using the finite element method, and high-stress areas on the sliding track are identified based on the stress distribution. Points in the high-stress region where the stress value exceeds a preset threshold are designated as key support points. Sensor nodes of the sensor network are deployed at each of the key support points. Each sensor node includes a laser rangefinder and a tilt sensor. The vertical displacement change of the sliding track is collected by the laser rangefinder deployed at the key support point, and the local tilt angle change of the sliding track is collected by the tilt sensor deployed at the key support point. Based on the Kalman filter algorithm, all the vertical displacement changes and the local tilt angle changes are fused to obtain the deformation sensing data.

3. The method for dividing steel structure construction areas based on spatial calculation according to claim 1, characterized in that, The calculation of the dynamic deformation field of the sliding track includes: A series of fixed target points are defined on the surface of the sliding track. The material property parameters of the sliding track and the load distribution parameters of the truss at each target point are obtained. The material property parameters of the sliding track include the elastic modulus and Poisson's ratio. The load distribution parameters of the truss include the truss self-weight and the position of the load application point. The deformation sensing data collected at current and historical consecutive time steps are combined into an input sequence in chronological order. The input sequence, material property parameters and load distribution parameters of all target points are input into a pre-trained deformation prediction model. The deformation prediction model synchronously outputs the spatial offset vector of all target points, which is a vector containing magnitude and direction information. The set of all target points and their spatial offset vectors is taken as the dynamic deformation field.

4. The method for dividing steel structure construction areas based on spatial calculation according to claim 3, characterized in that, Calculating the dynamic deformation field of a sliding track also includes: Prepare a training dataset containing multiple sets of samples. The samples are constructed based on historical construction processes. Each set of samples includes: historical input sequences, material property parameters of the sliding track at the target point, truss load distribution parameters, and actual spatial offset vectors of all target points. A time-series-based recurrent neural network is constructed, the weight parameters and bias parameters of the recurrent neural network are initialized, the samples are input into the recurrent neural network, and the predicted spatial offset vectors of all target points are obtained synchronously through forward propagation calculation. The loss function value is calculated based on the overall difference between the predicted spatial offset vector and the measured spatial offset vector at all target points, and the loss function adopts the mean square error function. The gradient of the loss function with respect to each weight parameter of the recurrent neural network is calculated using the backpropagation algorithm, and the weight parameters of the recurrent neural network are updated using the gradient descent optimization algorithm based on the gradient. Repeat the above steps until the loss function value converges to a preset threshold to obtain the trained deformation prediction model.

5. The method for dividing steel structure construction areas based on spatial calculation according to claim 4, characterized in that, Generating dynamic security boundaries includes: Based on the spatial offset vectors of all target points in the dynamic deformation field, the spatial offset vectors at each of the assembly area boundary, the sliding area boundary, and the positioning area boundary are determined by a spatial interpolation algorithm. The spatial coordinates of the assembly area boundary, the sliding area boundary, and the positioning area boundary are added to the corresponding spatial offset vectors to obtain the corrected spatial coordinates of the assembly area boundary, the sliding area boundary, and the positioning area boundary, which constitute the dynamic safety boundary.

6. The method for dividing steel structure construction areas based on spatial calculation according to claim 5, characterized in that, After generating the dynamic safety boundary, a dynamic spatial interference detection step is also included: Based on the truss geometric contour data in the design model, an outer envelope model of the truss is constructed. The outer envelope model is a minimal convex polyhedron generated by calculating the convex hull of all vertices of the truss. The minimum spatial distance between the outer envelope model and the dynamic safety boundary is calculated when the outer envelope model moves along the modified sliding path. The calculation process is completed by solving the Euclidean distance between the surface points of the outer envelope model and the surface points of the dynamic safety boundary and taking the minimum value. Compare the minimum spatial distance with a preset safety threshold; When the minimum spatial distance is not greater than the safety threshold, the boundary re-division process is triggered.

7. The method for dividing steel structure construction areas based on spatial calculation according to claim 6, characterized in that, The boundary re-division process includes: A set of known test loads are applied on the sliding track, and the corresponding deformation sensing data is collected through the sensor network. At the same time, the spatial offset vectors of all target points are measured to form a calibration dataset. Based on the calibration dataset, a response mapping function is established through multiple linear regression analysis. The response mapping function takes the spatial offset vector of all target points as input and the deformation sensing data as output. The input sequence is used as the first input sequence, and the corresponding dynamic deformation field is used as the first deformation field. The spatial offset vectors of all target points in the first deformation field are extracted. The spatial offset vectors of all extracted target points are input into the response mapping function to obtain simulated deformation sensing data A; Remove the deformation sensing data of the earliest time step in the first input sequence, and use the deformation sensing data A as the new deformation sensing data of the most recent time step to form a new first input sequence B; Based on the first input sequence B, a new first deformation field B1 is obtained. The above steps are repeated until a specified number of times are reached. The new first deformation fields obtained each time are arranged in chronological order to form the dynamic deformation field evolution trend.

8. The method for dividing steel structure construction areas based on spatial calculation according to claim 7, characterized in that, The boundary re-partitioning process also includes: Extract the last first deformation field C from the evolution trend of the dynamic deformation field; Based on the spatial offset vectors of all target points in the first deformation field C, the spatial offset vectors at each point on the boundary of the assembly area are determined by a spatial interpolation algorithm and denoted as the first vector. Calculate the arithmetic mean of the magnitudes of all first vectors as the boundary adjustment amount; Calculate the vector sum of all first vectors, and use the normalized direction of the vector sum as the boundary adjustment direction; Based on the boundary adjustment amount and the boundary adjustment direction, a translation transformation matrix is ​​constructed; The spatial coordinates of all positions on the boundary of the assembly area are transformed by the translation transformation matrix to generate an updated dynamic safety boundary.

Images