Rail transit mechanical and electrical installation method based on BIM, point cloud scanning and lofting robot

Abstract

The invention discloses a rail transit mechanical and electrical installation method based on BIM, point cloud scanning and a lofting robot, and belongs to the technical field of rail transit mechanical and electrical installation. The problems that in traditional mechanical and electrical installation, the adaptability of the prefabricated part is poor due to the deviation between the BIM model and the field structure, and secondary deviation is caused by the conversion error between the point cloud data and the lofting system data are solved. The method specifically comprises the steps that a three-dimensional BIM model is constructed, pipeline comprehensive arrangement and collision detection are carried out, point cloud data are collected and processed through point cloud scanning, deviation analysis is carried out through superposition comparison, adjustment is carried out to generate a deepening model, pipeline segmentation machining and two-dimensional code identification adding are carried out based on the deepening model, and pipeline processing is carried out. And finally, coordinate data are imported through the lofting robot for precise lofting installation. The method is mainly used for rail transit mechanical and electrical installation and has the advantages of high installation precision and efficiency and the like.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of rail transit electromechanical installation. More particularly, the present application relates to a rail transit electromechanical installation method based on BIM, point cloud scanning and lofting robot. BACKGROUND

[0002] Rail transit electromechanical installation engineering has the characteristics of dense pipelines, multi-specialty cross operation, complex space environment, etc. The traditional construction mode based on two-dimensional CAD drawings has the following defects: (1) pipeline integration relies on manual coordination, which is prone to collision conflicts, resulting in high on-site modification rate and large rework cost; (2) on-site measurement relies on manual lofting, and the accuracy is greatly affected by the operation experience (error usually ≥10mm), which is difficult to meet the high-precision installation demand; (3) there is a deviation between the design model and the actual structure on site, and there is a lack of effective data comparison and dynamic adjustment mechanism, resulting in poor adaptability of prefabricated components to the site; (4) the construction process is fragmented, and the data transmission between design, prefabrication and installation is not smooth, which affects the construction efficiency and quality control.

[0003] In the prior art, although BIM technology realizes three-dimensional visualization design, it lacks real-time fusion with actual data on site; point cloud scanning technology is only used for local size verification and has not formed full-process cooperation with prefabrication and on-site installation; and lofting robot application is limited to a single link and has not been deeply integrated with BIM model. Therefore, it is urgent to build an integrated technical system covering the whole process of "design-prefabrication-assembly". SUMMARY

[0004] The purpose of the present application is to provide a rail transit electromechanical installation method based on BIM, point cloud scanning and lofting robot to at least solve the above problems.

[0005] In order to achieve the purpose of the present application and other advantages, a rail transit electromechanical installation method based on BIM, point cloud scanning and lofting robot is provided, which comprises: constructing a three-dimensional BIM model containing civil structures and electromechanical pipelines, performing electromechanical pipeline comprehensive arrangement and collision detection, and generating an initial design model, wherein the three-dimensional BIM model adopts a coordinate system unified with field measurement; adopting a three-dimensional point cloud scanner to collect point cloud data of the structure of the construction site, the point position accuracy of the point cloud data being 1-2 mm, performing point cloud splicing and noise reduction processing through a point cloud processing software, the point cloud splicing error being controlled within a range of 2-4 mm, and generating a field point cloud model; superimposing and comparing the field point cloud model and the initial design model in the unified coordinate system, extracting structure feature points based on the point cloud data, and performing deviation analysis on the initial design model, identifying structure beam deviation and reserved hole position deviation, the deviation identification early warning value being 5-30 mm, adjusting the electromechanical pipeline arrangement and support hanger position in the initial design model according to the comparison result, and generating a deepened model; performing pipeline segmentation based on the deepened model, the segmented length being determined as 1-6 m according to transportation limitation and hoisting requirement, generating a prefabricated pipeline component processing list and assembly drawing, processing the prefabricated pipeline component through a numerical control plasma cutting machine, the processing accuracy being controlled within a range of 1-3 mm, and attaching a two-dimensional code mark containing component number and material information to the processed prefabricated pipeline component; importing the coordinate data in the deepened model into a lofting robot control system through a data verification interface, the data verification interface performing format conversion and accuracy verification on the coordinate data, the lofting robot control system receiving GPS signals and laser positioning data, analyzing the coordinate data and generating three-dimensional lofting points, the lofting point positioning accuracy being 1-2 mm, and an on-site operator installing support hangers and prefabricated pipeline components according to the lofting point position.

[0006] Preferably, when performing point cloud splicing and noise reduction processing, an automatic matching algorithm based on feature points is adopted to register the point cloud data of adjacent scanning stations, and after registration, a voxel grid downsampling method is adopted for point cloud density uniform processing, and the point cloud density is controlled within a range of 5000-8000 points per square meter.

[0007] Preferably, the automatic matching algorithm based on feature points is adopted to register the point cloud data of adjacent scanning stations, and specifically comprises: extracting corner points and edge feature points in the point cloud data of adjacent scanning stations as matching primitives; adopting an ISS feature point detection algorithm to calculate an FPFH feature descriptor of each feature point; establishing a preliminary correspondence between feature points of source point cloud and target point cloud based on the similarity of the FPFH feature descriptor; using a RANSAC algorithm to filter the preliminary correspondence and calculate a coarse registration transformation matrix; adopting an ICP algorithm to perform fine registration with the coarse registration result as an initial pose, and the iteration error threshold of fine registration being set as 2-4 mm. (to solve the problem of high failure rate of traditional single feature matching in repeated structure or low texture area registration).

[0008] Preferably, when performing deviation analysis, the structural beam deviation identification early warning value is set to 10 mm, and when the identified structural beam deviation exceeds the early warning value, a deviation adjustment scheme is automatically generated, which includes mechanical and electrical pipeline elevation adjustment and support hanger position correction.

[0009] Preferably, the specific process of automatically generating a deviation adjustment scheme includes: according to the difference ΔH between the structural beam deviation and the early warning value, calculating the mechanical and electrical pipeline elevation adjustment according to the pipeline slope compensation algorithm, elevation adjustment = ΔH + L × i min , wherein L is the length of the affected pipe section, i min is the minimum slope allowed by the pipeline system, with a value range of 0.3% to 0.5%; at the same time, the support hanger position correction is calculated according to the support hanger design distance D and the deviation direction, position correction = D × tanθ, wherein θ is the included angle between the deviation direction and the pipeline axis; the elevation adjustment calculated needs to ensure that the compensated pipeline slope is within the range of 0.3% to 1.0%, and the distance deviation of the support hanger position correction is controlled within ±50mm.

[0010] Preferably, when performing pipeline segmentation, it also includes: in the deepening model, adding a three-dimensional docking mark containing flange specifications and bolt hole positions at each pipe section connection; based on the transportation restriction condition, the maximum length of the pipe section is determined to be 6 meters, and based on the lifting capacity, the maximum weight of a single pipe section is determined to be 800 kilograms; finite element analysis is used to simulate the stress state of the pipe section during lifting, and the analysis indicators include maximum deflection and equivalent stress, wherein the maximum deflection limit is 1 / 400 of the pipe section length, and the equivalent stress limit is 60% of the pipe material yield strength; when the simulation result exceeds the limit, the following steps are performed: first, add ring-shaped reinforcing supports in the middle of the pipe section, with a support spacing of 2 to 3 meters, and a support cross-sectional moment of inertia not less than 1.2 times the pipe section body moment of inertia; re-perform finite element analysis verification; if the ring-shaped reinforcing supports still do not meet the requirements after being added, re-segment the pipe section in the middle, control the new segment maximum length within the range of 3 to 4 meters, and regenerate the segment interface flange design.

[0011] Preferably, the data verification interface adopts a CRC-32 verification algorithm to verify the transmitted coordinate data, and automatically triggers a retransmission mechanism when the error of single-point coordinate data exceeds 2 mm; the retransmission mechanism includes: resending the data packet after a 100 ms delay, and the maximum number of retransmissions is set to 3 times; if the data error still exceeds 2 mm after 3 retransmissions, the coordinate point is marked as an abnormal point, and the precast pipe component number and coordinate value of the point are recorded; a data integrity report is generated after the verification is passed, which records the transmission time, total data transmission amount, number of abnormal points, and specific coordinate value and corresponding precast pipe component number of each abnormal point; the lofting robot control system automatically skips the coordinate positions marked as abnormal points after receiving the data integrity report, and displays the abnormal point list and suggests manual measurement prompts on the operation interface.

[0012] Preferably, the lofting robot control system adopts a multi-source data fusion algorithm to comprehensively process GPS signals and laser positioning data when generating three-dimensional lofting points, and automatically switches to a full laser positioning mode when there is no GPS signal.

[0013] Preferably, when performing comprehensive arrangement and collision detection of mechanical and electrical pipelines, the Navisworks software is used for collision detection between the mechanical and electrical pipelines and the building structure, and the collision detection accuracy is set to 3-5 mm; for the detected collision points, automatic arrangement and optimization are performed according to the pipeline comprehensive principle, wherein the distance between adjacent pipelines is controlled to be 100-150 mm, and the distance between the pipeline and the building structure is controlled to be 50-100 mm.

[0014] The present application at least includes the following beneficial effects: First, through the cooperative application of high-precision point cloud scanning and a unified coordinate system, the high consistency of the BIM deepening model and the actual site condition is ensured, fundamentally solving the adaptability problem of precast components to the site; at the same time, the lofting coordinates are verified through the data verification interface, effectively blocking the data error transmission in the process from the model to the site lofting, avoiding rework caused by secondary deviation, and significantly improving the one-time success rate of mechanical and electrical installation and the overall construction efficiency.

[0015] Second, by using an automatic matching algorithm based on feature points for point cloud registration, the dependence on manual operation during multi-station scanning data splicing is greatly reduced, subjective errors and error accumulation problems introduced by human intervention are avoided, and the overall accuracy and reliability of the point cloud model are ensured; subsequent voxel grid downsampling processing unifies the point cloud density, optimizes the data amount while retaining key structural features, and provides an accurate and efficient data basis for subsequent model comparison and deviation analysis.

[0016] Third, by adopting the strategy of combining ISS feature point detection, FPFH feature descriptor matching and RANSAC rough registration, it can effectively deal with the low-texture areas such as repeated structures and smooth surfaces commonly found in construction sites, significantly improving the success rate and accuracy of feature point matching. On this basis, the ICP algorithm is used for fine registration, further controlling the registration error within a very low range, thereby ensuring the high precision of the final point cloud model and laying a solid technical foundation for subsequent accurate deviation identification.

[0017] Fourth, by setting scientific structural beam deviation warning values and automatically generating adjustment schemes, it realizes the rapid identification and intelligent response to major structural deviations on site, and changes the traditional inefficient mode of relying on manual measurement, calculation and decision-making. Not only does it significantly shorten the processing time for responding to sudden deviations, but also avoids omissions and errors that may occur during manual adjustment through systematic automatic calculation, ensuring the rationality and feasibility of the adjustment scheme.

[0018] Fifth, by conducting finite element simulation analysis of the pipe segment hoisting process in a virtual environment, it can predict and avoid the risk of plastic deformation or tearing of large-size thin-walled pipe segments during transportation and hoisting due to insufficient stiffness. According to the analysis results, intelligent optimization schemes such as adding reinforcement supports or re-segmentation are given to ensure the structural safety and hoisting feasibility of the prefabricated pipe segments from the design source, reducing safety risks and material losses during site construction.

[0019] Sixth, by introducing an efficient data verification and retransmission mechanism, it can identify and correct errors in coordinate data caused by interference during transmission in real time, ensuring the integrity and accuracy of the coordinate data imported into the lofting robot. The function of automatically marking abnormal points and generating detailed reports enables the operator to quickly locate the source of the problem and take remedial measures such as manual measurement, effectively preventing the lofting robot from working based on incorrect data and thereby avoiding the resulting chain installation errors.

[0020] Seventh, the lofting robot fuses multiple source positioning data such as GPS and laser, and intelligently switches the dominant positioning mode according to signal strength, ensuring that it can continuously obtain stable and reliable positioning information in different construction environments, especially in underground spaces without GPS signals. This adaptive capability ensures that the positioning accuracy of the lofting points does not degrade due to environmental changes, ensuring that the end lofting link of the entire mechanical and electrical installation process can always be efficient and accurate.

[0021] Eighth, by improving the accuracy of collision detection and strictly setting the minimum spacing control standard between the pipeline and the structure, the pipeline and the pipeline, the collision detection is no longer limited to finding hard collision, and the soft collision problem of insufficient installation operation space can be identified in advance, the deep optimization of the mechanical and electrical pipeline arrangement is realized, the scene is avoided in the traditional method, the installation and maintenance space is not reserved enough, and the scene is changed, the installation quality and the one-time optimization rate are significantly improved.

[0022] Other advantages, objects, and features of the application will be in part apparent and in part pointed out below. DETAILED DESCRIPTION

[0023] The application will be further described in detail below with reference to examples, so that those skilled in the art can implement the application according to the description.

[0024] It should be noted that the experimental methods in the following examples are conventional methods, and the reagents and materials are commercially available unless otherwise specified.

[0025] According to one embodiment of the present application, a rail transit electromechanical installation method based on BIM, point cloud scanning and lofting robots comprises: constructing a three-dimensional BIM model containing civil structures and electromechanical pipelines, performing electromechanical pipeline comprehensive arrangement and collision detection, and generating an initial design model, wherein the three-dimensional BIM model adopts a coordinate system unified with field measurement; using a three-dimensional point cloud scanner to collect point cloud data of the structure of the construction site, the point position accuracy of the point cloud data being 1-2 mm, performing point cloud splicing and noise reduction processing through point cloud processing software, the point cloud splicing error being controlled within a range of 2-4 mm, and generating a field point cloud model; superimposing and comparing the field point cloud model and the initial design model in the unified coordinate system, extracting structure feature points based on the point cloud data, performing deviation analysis on the initial design model, identifying structure beam deviation and reserved hole position deviation, the deviation identification early warning value being 5-30 mm, adjusting the electromechanical pipeline arrangement and support hanger position in the initial design model according to the comparison result, and generating a deepened model; performing pipeline segmentation based on the deepened model, the segmentation length being determined as 1-6 m according to transportation limitation and hoisting requirement, generating a prefabricated pipeline component processing list and assembly drawing, processing the prefabricated pipeline component through a numerical control plasma cutting machine, the processing accuracy being controlled within a range of 1-3 mm, and attaching a two-dimensional code mark containing component number and material information to the processed prefabricated pipeline component; importing the coordinate data in the deepened model into a lofting robot control system through a data verification interface, the data verification interface performing format conversion and accuracy verification on the coordinate data, the lofting robot control system receiving GPS signals and laser positioning data, analyzing the coordinate data and generating three-dimensional lofting points, the lofting point positioning accuracy being 1-2 mm, and field operators installing support hangers and prefabricated pipeline components according to the lofting point positions.

[0026] In the above embodiment, the three-dimensional BIM model of the civil structure and the electromechanical pipeline can be constructed by using the BIM modeling software on the computer. The modeling process needs to adopt a coordinate system unified with the field total station measurement, such as a local coordinate system or a construction independent coordinate system. After completing the model building, the software collision detection function needs to be used to check the interference between the electromechanical pipeline and the building structure, and the collision detection accuracy can be set to 3 mm. For the detected collision points, manual or automatic adjustment should be performed according to the pipeline comprehensive avoidance principle, and the distance between adjacent metal pipelines can be controlled to 100 mm and the distance between the metal pipeline and the non-metal pipeline can be controlled to 150 mm, so as to generate a collision-free initial design model, which will serve as the basis for subsequent work.

[0027] At the construction site, a phase three-dimensional laser scanner can be used to collect point cloud data, and the single-point ranging accuracy can reach 1 mm. When scanning, a target ball needs to be placed at each station as a public matching point. Point cloud data processing can be done in a point cloud processing software. First, the automatic matching algorithm based on the target ball is used to splice the scanning data of multiple stations into a whole, and the splicing error can be controlled within 2 mm. Then, the voxel grid downsampling method is used to denoise and thin the spliced massive point cloud, and the point cloud density is uniformly reduced to about 7000 points per square meter, and finally a high-precision on-site point cloud model is generated, which truly reflects the actual situation of the construction site.

[0028] The on-site point cloud model and the initial design model are imported into the same comparison software, and the spatial alignment function of the software is used to accurately superimpose the two in the same coordinate system. The software can automatically extract feature points such as structural beam boundary lines, floor elevations, and reserved hole centers based on point cloud data, and perform deviation analysis with the corresponding positions in the model. When the displacement of the structural beam exceeds the 10 mm warning value, the system can automatically generate an adjustment scheme, such as compensatory adjustment of pipeline elevation and corresponding correction of support hanger position. Engineers update the initial design model based on this scheme to obtain a deepening model that closely matches the site, which will be directly used to guide prefabrication and installation.

[0029] In the deepening model, according to the size limit of the transport vehicle cargo box, the pipe segment length is determined to be 1 m, 3 m or 6 m, and a processing drawing containing the accurate size and slope of each pipe segment is generated. A numerical control plasma cutting machine can be selected for cutting and beveling of carbon steel pipes, with a processing accuracy of 2 mm. After each pipe segment is processed, a two-dimensional code label can be attached to its surface, which contains the component number, specification, and installation position information. Then, the support hanger coordinate points and pipe interface coordinate points in the deepening model are imported into the control system of the lofting robot through data lines, and the lofting robot uses its internal total station module to accurately project these three-dimensional coordinate points to the actual position on site, with a lofting point positioning accuracy of 1 mm. Construction personnel can install support hangers and prefabricated pipes according to these light spot indicators.

[0030] Specifically, the construction team first created a three-dimensional BIM model containing all civil and mechanical and electrical specialties in a unified coordinate system, and performed collision detection to generate an initial design model. Subsequently, a high-precision three-dimensional laser scanner was used to scan the completed tunnel section, obtaining millimeter-level point cloud data. After software processing, an accurate on-site point cloud model was generated. By superimposing and comparing the point cloud model with the initial BIM model, the system automatically identified deviations in the pipe gallery roof elevation and design at multiple locations. If the deviation exceeds the warning value, technical personnel will adjust the elevation of all water pipes and air pipes below and the installation position of the corresponding hangers in the model based on the results, generating a deepening model that perfectly fits the site. Based on the deepening model, the pipes are segmented and prefabricated, and each prefabricated pipe component has a two-dimensional code containing identity information. Finally, the coordinate data in the deepening model is directly imported into the lofting robot, which automatically and accurately projects the installation position of the hangers and pipe fittings interfaces on site. Workers install according to the laser point indication, and all prefabricated pipes are installed in place at one time without any on-site modifications. The entire process is efficient and accurate.

[0031] This embodiment ensures the high consistency of the design model with the actual site through the deep integration of high-precision point cloud scanning and BIM model, effectively solving the problem of prefabricated components that cannot be installed due to civil construction deviations. At the same time, through standardized data processes and precise automatic lofting, data conversion and transmission errors that may occur in multiple links from design to construction are eliminated, significantly reducing on-site cutting, welding, and other secondary operations, and improving the overall precision and work efficiency of mechanical and electrical installation.

[0032] According to another embodiment of the present application, when performing point cloud stitching and noise reduction processing, an automatic matching algorithm based on feature points is used to register the point cloud data of adjacent scanning stations. After registration, a voxel grid downsampling method is used for point cloud density uniform processing, and the point cloud density is controlled at 5000 to 8000 points per square meter.

[0033] In the above embodiment, when performing three-dimensional laser scanning on site, it is necessary to set up stations at different stations, and the single-point ranging accuracy of the three-dimensional laser scanner can be 2 mm. In order to splice the point cloud data obtained by multiple stations into a complete whole, the operator can select an automatic matching algorithm based on feature points in the point cloud processing software. The algorithm automatically searches for common features in the point clouds of adjacent two stations, such as the edge lines of the corners of the wall or the centers of the pre-pasted target balls, and calculates the rotation and translation matrix based on this to complete the preliminary alignment of the data. This process is called coarse registration. Coarse registration can roughly align the positions of the data of two stations, laying a foundation for subsequent fine registration. On the basis of coarse registration, the software further performs fine registration. The iterative closest point algorithm used in fine registration continuously fine-tunes the transformation matrix to minimize the average distance between corresponding points in the two pieces of point cloud. The iteration error threshold of this process can be set to 3 mm, and the algorithm stops iteration when the error is lower than this value, thereby ensuring that the final point cloud splicing error is strictly controlled within a relatively low range. After the registration of all stations is completed, the software optimizes the large point cloud data set by using the voxel grid downsampling method, sets the voxel edge length to 5 mm, and keeps only one representative point in each voxel space, thereby controlling the overall point cloud density to about 7000 points per square meter.

[0034] The embodiment significantly reduces the subjective error and error accumulation caused by manual selection of matching points in the multi-station point cloud data splicing process through automatic feature matching and fine registration algorithm, and ensures the overall accuracy and reliability of the point cloud model. The subsequent voxel downsampling process effectively reduces the occupation of massive point cloud data on computing resources on the premise of preserving key structural features, improves the efficiency of subsequent model comparison and analysis, and provides quality assurance for generating high-precision on-site digital models.

[0035] According to another embodiment of the application, the point cloud data of adjacent scanning stations is registered by using an automatic matching algorithm based on feature points, specifically including: extracting corner points and edge feature points in the point cloud data of adjacent scanning stations as matching primitives; calculating the FPFH feature descriptor of each feature point by using the ISS feature point detection algorithm; establishing a preliminary correspondence between the feature points of the source point cloud and the target point cloud based on the similarity of the FPFH feature descriptor; screening the preliminary correspondence and calculating the coarse registration transformation matrix by using the RANSAC algorithm; and performing fine registration by using the ICP algorithm with the coarse registration result as the initial pose, and the iteration error threshold of the fine registration is set to 2 mm to 4 mm. (to solve the problem of high failure rate of traditional single feature matching in repeated structure or low texture area registration).

[0036] In the above embodiment, in the point cloud processing software, after preprocessing the source point cloud and target point cloud data that need to be spliced, the feature points are extracted first. The software can call the internally integrated ISS algorithm module, which evaluates the saliency of the points by establishing a local coordinate system around each point and calculating the eigenvalues of the covariance matrix thereof, so as to screen out stable corner points and edge points such as wall corners and equipment edges as the elements for subsequent matching. These feature points can better represent the geometric structure of the scanned scene and provide a reliable matching basis for the association of different point cloud data.

[0037] After obtaining the feature point set, a high-dimensional feature description vector needs to be calculated for each feature point to represent its local geometric properties. The system can use the FPFH algorithm to complete this step, which calculates the normal vector angle and other parameters of all points in a certain neighborhood around the feature point, and generates a 33-dimensional floating-point vector as the descriptor of the feature point. This descriptor is not sensitive to the rotation and translation changes of the point cloud and has good robustness. After calculating the FPFH descriptor of all feature points, the software will evaluate their similarity by calculating the Euclidean distance between the descriptor vectors, and establish a preliminary corresponding relationship set for the feature points of the source point cloud and the target point cloud.

[0038] Since the preliminary matching will contain a large number of false correspondences, the RANSAC algorithm needs to be used for screening next. This algorithm will randomly extract three pairs of matching points to calculate a transformation matrix, then verify the degree of agreement of other matching points under this matrix, and find the optimal transformation model with the most inliers through multiple iterations, so as to calculate an accurate coarse registration transformation matrix. On this basis, the software will further perform ICP fine registration. The ICP algorithm will take the coarse registration result as the initial value, iteratively find the nearest point pairs between the two point clouds, and calculate the optimal rigid body transformation to minimize the mean square error. The iteration error threshold of this process can be set to 3 mm, and when the average distance error of the transformed point cloud is lower than this threshold, the registration is completed, and finally the high-precision automatic alignment of the two point clouds is realized.

[0039] This embodiment can effectively capture and describe stable and unique geometric features in the point cloud through advanced feature detection and description algorithms such as ISS and FPFH, significantly improving the success rate of feature matching in repeated structures or construction areas lacking texture. Combined with the robust estimation of RANSAC and the high-precision iterative optimization of ICP, a complete automatic registration process is formed, greatly reducing the dependence on manual intervention and ensuring the automation degree, accuracy and reliability of the final result of the multi-point cloud data splicing.

[0040] According to another embodiment of the present application, when performing deviation analysis, the structural beam deviation identification early warning value is set to 10 mm, and when the identified structural beam offset exceeds the early warning value, an automatic deviation adjustment scheme is generated, which includes mechanical and electrical pipeline elevation adjustment amount and support hanger position correction amount.

[0041] In the above embodiment, after the point cloud model is superimposed and compared with the initial design model, the system starts the deviation analysis program. The program checks the deviation value of the actual position of the structural beam and the model position one by one. The user can previously set the identification early warning value of the structural beam deviation to 10 mm in the parameter setting interface of the software. The threshold value can be determined according to the allowable deviation value of the cast-in-place structure section size in the common concrete structure construction acceptance specification. When the system detects that the elevation or plane position deviation of the structural beam at a certain place exceeds the early warning value of 10 mm, the component at the place is highlighted in the model interface, and a visual alarm is sent to the user, prompting that there is a significant deviation that needs attention at the place.

[0042] Once it is confirmed that the deviation exceeds the early warning value, the system automatically triggers the adjustment scheme generation module. The module has embedded adjustment logic and will first calculate the specific value and direction of the deviation. For the adjustment of the mechanical and electrical pipeline elevation, the module will calculate the height value that needs to be raised or lowered as a whole according to the length of the affected pipe section and the minimum slope required by the system, to ensure that the pipe slope always meets the requirements of drainage or air flow. For the support hanger, the module will calculate the horizontal distance that each support hanger needs to move according to its design spacing and the offset direction of the pipeline, so as to ensure that the support hanger can still uniformly and effectively support the adjusted pipeline. Finally, the system generates a list containing all the necessary adjustment amounts for the designer to review and apply.

[0043] By setting a clear deviation early warning threshold, the present embodiment can quickly screen out the civil structure deviation that has a substantial impact on mechanical and electrical installation, avoiding the inefficient work of checking all minor deviations one by one. The further automatic adjustment scheme generation function converts the complex judgment and calculation process that traditionally relies on the personal experience of engineers into systematic automatic processing, significantly improving the response speed and processing efficiency of major site deviations, while reducing the risk of inconsistent treatment schemes or calculation errors caused by human subjective factors, ensuring the rationality and feasibility of the subsequent deepening design scheme.

[0044] According to another embodiment of the present application, the specific process of automatically generating a deviation adjustment scheme includes: according to the difference ΔH between the structural beam offset and the early warning value, calculating the mechanical and electrical pipeline elevation adjustment amount according to the pipeline slope compensation algorithm, elevation adjustment amount = ΔH + L × i min , where L is the length of the affected pipe section, and i minThe minimum slope allowed for the pipeline system is 0.3% to 0.5%. Meanwhile, the position correction of the support and hanger is calculated according to the design distance D and the offset direction. The position correction is D x tan θ, where θ is the angle between the offset direction and the pipeline axis. The elevation adjustment calculated should ensure that the pipeline slope after compensation is within the range of 0.3% to 1.0%, and the distance deviation of the support and hanger after position correction should be controlled within ±50 mm.

[0045] In the above embodiment, when the system identifies that the offset of a structural beam at a certain location exceeds the 10 mm early warning value, it automatically calculates the difference ΔH. Subsequently, the system calls the built-in slope compensation algorithm to calculate the elevation adjustment of the electrical pipeline. The algorithm reads the length L of the affected pipeline section, which can be automatically obtained from the BIM model. At the same time, the algorithm calls a preset minimum slope value i min of the system, which can be set to 0.4%. The calculation formula for the elevation adjustment is ΔH plus the product of L and i min . This calculation not only compensates for the structural deviation, but also ensures that the pipeline still meets the minimum slope requirement at the new elevation, preventing reverse slope or insufficient slope.

[0046] For the correction of the support and hanger position, the system first obtains the design distance D of the support and hanger on the pipeline from the deepening model, which can be 1.5 meters. At the same time, the system determines the offset direction of the structural beam according to the point cloud comparison result, and calculates the angle θ between the offset direction and the pipeline axis direction. The support and hanger position correction is calculated by multiplying the design distance D by the tangent of the angle θ. This allows each support and hanger to be accurately adjusted in proportion to its specific position, ensuring that the support and hanger is always located directly below the adjusted pipeline, providing effective support.

[0047] After all calculations are completed, the system automatically verifies the adjustment results. It reviews the adjusted pipeline slope to ensure that it is within the reasonable range of 0.3% to 1.0%, meeting functional requirements while avoiding excessive slope changes. At the same time, it checks the corrected support and hanger distance to ensure that the deviation from the design distance is controlled within ±50 mm, ensuring the uniformity and reliability of the support. Only when all parameters pass the verification, the automatically generated adjustment scheme will be finally confirmed and applied to the deepening model.

[0048] This embodiment automatically calculates the adjustment amount through the embedded standardized algorithm, converting the complex manual judgment and calculation process into an efficient and accurate automated process, significantly improving the processing efficiency. This method ensures the scientificity and systematicness of pipeline slope adjustment, avoiding the discontinuity or errors of manual calculation, while ensuring the uniformity of the support and hanger after correction and the effectiveness of the load, improving the quality and reliability of the deepening design scheme as a whole.

[0049] According to another embodiment of the application, when the pipeline is segmented, a three-dimensional docking mark containing the flange specifications and bolt hole positions is added to each pipe segment connection in the deepening model. The maximum length of the pipe segment is determined to be 6 meters based on the transportation restrictions, and the maximum weight of a single pipe segment is determined to be 800 kilograms based on the lifting capacity. Finite element analysis is used to simulate the stress state of the pipe segment during lifting, and the analysis indicators include the maximum deflection and equivalent stress, where the maximum deflection limit is 1 / 400 of the length of the pipe segment, and the equivalent stress limit is 60% of the yield strength of the pipe material. When the simulation results exceed the limits, the following steps are performed: first, add a ring-shaped reinforcing support to the middle of the pipe segment, with a support spacing of 2-3 meters, and a support cross-sectional moment of inertia not less than 1.2 times the pipe segment body moment of inertia; re-perform finite element analysis verification; if the ring-shaped reinforcing support is still not satisfied after adding, re-segment the middle of the pipe segment, with the new segment length controlled within 3-4 meters, and regenerate the segment interface flange design.

[0050] In the above embodiment, during the deepening design phase, engineers use three-dimensional design software to segment the pipeline. The software can automatically generate a three-dimensional docking mark for each flange connection between pipe segments, which clearly displays the nominal diameter, pressure rating, and number and orientation of bolt holes. When segmenting, the system will follow the pre-set rules to limit the maximum length of the pipe segment to 6 meters to adapt to the transportation capacity of a conventional truck; at the same time, the maximum weight of a single pipe segment is limited to 800 kilograms to match the rated lifting capacity of the site tower crane. These parameters can be pre-set in the software according to the performance indicators of the transportation tools and lifting equipment actually used in the project.

[0051] After completing the initial segmentation, engineers will import the pipe segment model into the finite element analysis software for lifting simulation. During analysis, the yield strength of the pipe material can be set to 235 MPa. The software simulates the stress state of the pipe segment under the influence of lifting with both ends hanging, and calculates its maximum deflection value and equivalent stress value. The system sets the criteria that the maximum deflection must not exceed 1 / 400 of the length of the pipe segment, and the equivalent stress must not exceed 141 MPa. If the analysis results show that the maximum deflection of a certain large-diameter thin-wall water pipe exceeds the limit value during lifting, the system will prompt that the pipe segment is insufficient in rigidity and there is a risk of permanent deformation.

[0052] For pipe segments that do not meet the requirements, the first solution is to add a ring-shaped reinforcing support for optimization. A first ring-shaped reinforcing ring can be added 2.5 meters from the end of the pipe segment, and the material can be Q235B carbon steel, with a cross-sectional moment of inertia calculated to be 1.3 times the moment of inertia of the pipe segment body. After completing the model modification, it needs to be submitted again for finite element analysis verification. If the calculation results show that after adding a reinforcing ring, the maximum deflection of the pipe segment has decreased to within the limit value and the equivalent stress also meets the requirements, the reinforcing scheme is approved and the model is updated.

[0053] If the pipe segment still fails to pass the verification after adding a reinforcing ring, the system will suggest performing re-segmentation. The software will automatically split the original pipe segment at a middle position, generate a new flange connection interface, and control the maximum length of the new segment within the range of 3-4 meters. The designer needs to specify the flange specifications and design the bolt hole positions for the newly generated interface. The pipe segment model after re-segmentation needs to be re-calculated for weight and simulated for lifting to ensure that it meets all safety requirements, thereby fundamentally avoiding structural risks during lifting.

[0054] By introducing finite element simulation analysis, the present embodiment can detect potential stiffness deficiency and strength risks of large-size thin-walled pipe segments during lifting before prefabrication and processing, thereby providing an opportunity for optimization at the design stage. The method provides a systematic solution from adding reinforcing supports to re-segmentation, effectively avoiding potential safety accidents such as pipe body permanent deformation or cracking during on-site lifting through virtual verification, ensuring construction safety and engineering quality, and reducing economic losses and schedule delays caused by component damage.

[0055] According to another embodiment of the present application, the data verification interface uses the CRC-32 verification algorithm to verify the transmitted coordinate data, and automatically triggers the retransmission mechanism when the single-point coordinate data error exceeds 2 mm. The retransmission mechanism includes: resending the data packet after a 100 ms delay, and setting the maximum retransmission number to 3 times. If the data error still exceeds 2 mm after retransmission for 3 times, the coordinate point is marked as an abnormal point, and the precast pipe component number and coordinate value of the point are recorded. After the verification is passed, a data integrity report is generated, which records the transmission time, data transmission amount, number of abnormal points, and specific coordinate values and corresponding precast pipe component numbers of each abnormal point. The lofting robot control system automatically skips the coordinate positions marked as abnormal points after receiving the data integrity report, and displays the abnormal point list and suggests manual measurement prompts on the operation interface.

[0056] In the above embodiment, during the process of importing the coordinate data in the deepened model into the lofting robot control system, the data is transmitted through a dedicated verification interface. This interface can be integrated into the data transmission software, which internally uses the CRC-32 cyclic redundancy check algorithm to verify each transmitted coordinate data packet. During verification, the system compares the calculated check code at the receiving end with the check code attached by the sending end. The system's preset error tolerance is 2 mm. When the system detects that the data of a certain coordinate point exceeds this 2 mm threshold due to transmission interference, it automatically triggers the retransmission mechanism to ensure the absolute accuracy of the data.

[0057] After the retransmission mechanism is triggered, the system does not immediately resend the data. It can wait for a 100-millisecond delay to avoid possible transient network interference, and then resend the data packet after the communication link is restored to stability. Such retry operation will be performed at most three times. If after three retransmissions, the data error of the coordinate point is still greater than 2 mm, the system will determine that the point is an unreliable abnormal point. At this time, the system will automatically record the three-dimensional coordinate value of the abnormal point, and find and record the unique number of the prefabricated pipe component corresponding to the coordinate point through data association, and store these information into a temporary log file.

[0058] When the entire data set is transmitted and verified, the data verification interface will automatically generate a data integrity report. The contents of this report include the starting time, ending time, total number of successfully transmitted coordinate points, and the number of abnormal points marked for this data transmission. For each abnormal point, the report will list its X, Y, Z coordinate values and the associated prefabricated component number in detail. This report will be automatically sent to the control system of the lofting robot in the form of a text file as a complete record of this data receiving process.

[0059] The control system of the lofting robot will first read the data integrity report before starting the lofting operation. The control system software will automatically exclude all abnormal point coordinates listed in the report from the current lofting task queue, ensuring that the robot will not go to these positions that may have errors for lofting. At the same time, the control system will display the component number and coordinates of all abnormal points on its touch screen operation interface in a prominent list, and also provide a prompt for the operator to use tools such as total station to manually recheck these specific points, thereby guiding human intervention to complete the final positioning.

[0060] This embodiment effectively guarantees the integrity and accuracy of coordinate data transmission from the design end to the lofting end through strict data verification and retransmission mechanism, preventing the deviation of lofting points caused by data transmission errors. The automatic abnormal point identification and recording function can quickly locate the problem data and isolate it, avoiding the execution of incorrect instructions by the lofting robot. Finally, through the generation of detailed reports and clear operation prompts, the construction personnel are provided with clear guidance, ensuring that individual data point abnormalities will not affect the overall work process, forming a reliable data quality control closed loop.

[0061] According to another embodiment of the present application, the control system of the lofting robot uses a multi-source data fusion algorithm to process GPS signals and laser positioning data when generating three-dimensional lofting points, and automatically switches to a full laser positioning mode when there is no GPS signal.

[0062] In the above embodiment, the lofting robot control system internally integrates a multi-source data fusion positioning module. The hardware basis of the module can include a built-in GPS receiver module for receiving satellite signals and a built-in laser tracker for receiving positioning signals emitted by a laser emitter arranged on a known reference point at the construction site. During operation, the control system synchronously receives the two signals and runs a Kalman filtering algorithm to perform weighted fusion calculation on the absolute coordinates provided by the GPS and the relative coordinates provided by the laser ranging, thereby generating an optimally estimated three-dimensional lofting point coordinate. This fusion strategy can utilize the global nature of GPS and the local high precision of laser, complement each other, and provide continuous and stable position reference for the autonomous navigation and lofting of the robot.

[0063] The control system monitors the signal strength parameter output by the GPS receiver module in real time. This monitoring process is continuous, and the sampling frequency can be 1 time per second. The system has a preset signal strength threshold, such as 28 dB-Hz. When it is monitored that the GPS signal strength has been continuously below the threshold for 5 seconds, the system logic determines that it is currently in an environment with severe satellite signal attenuation or loss, such as an underground pipe gallery or a densely built area. At this time, the control system automatically performs switching operation, stops calling GPS data stream in the fusion algorithm, and completely relies on the signal provided by the laser positioning system for positioning mode, thereby entering the full laser positioning mode, ensuring that the lofting operation will not be interrupted due to satellite signal failure.

[0064] This embodiment can obtain stable and reliable global coordinates for the lofting robot in most open sites by fusing positioning signals of different principles, and can seamlessly switch to rely on local high-precision laser positioning network in complex environments with limited satellite signals, ensuring the continuity and reliability of the positioning service. This automatic switching mechanism enables a set of equipment to adapt to various construction environments from open air to underground, avoids interruption of operation due to signal loss, expands the application range of the lofting robot, and ensures the construction efficiency of the end link of the entire mechanical and electrical installation process.

[0065] According to another embodiment of the present application, preferably, when performing mechanical and electrical pipeline comprehensive arrangement and collision detection, Navisworks software is used for collision detection between the mechanical and electrical pipelines and the building structure, and the collision detection accuracy is set to 3-5 mm; for the detected collision points, automatic arrangement optimization is performed according to the pipeline comprehensive principle, wherein the distance between adjacent pipelines is controlled to be 100-150 mm, and the distance between the pipeline and the building structure is controlled to be 50-100 mm.

[0066] In the above embodiment, after the three-dimensional BIM model is completed, the designer can import the model file created by Revit or other software into Navisworks software for in-depth collision detection. In the software clash detective module, the detection rules need to be set. The detection range should include the interference check between all mechanical and electrical pipelines and between the pipelines and the building structure. The precision parameter of the collision detection can be set to 5 mm, which means that the software will identify all potential conflicts with a spacing of less than 5 mm. The software will run the detection algorithm to generate a detailed conflict report, which includes not only hard collisions but also soft collisions that cannot be constructed due to insufficient spacing.

[0067] After the software detects the collision points, the designer can start the automatic arrangement optimization function. This function is embedded with common pipeline comprehensive avoidance principles, such as letting pressure pipes let non-pressure pipes, letting small pipes let large pipes, and letting metal pipes let non-metal pipes. The optimization algorithm will automatically adjust the position of the colliding pipeline segments according to these principles and in combination with the preset spacing requirements. The adjustment methods can include lifting or lowering the pipeline in the vertical direction, or translating it horizontally, so as to re-plan a reasonable pipeline arrangement path in three-dimensional space and eliminate all identified collision points.

[0068] During the optimization arrangement process, the software will strictly implement the preset minimum spacing control standard. For adjacent mechanical and electrical pipelines arranged side by side, whether they are air pipes, water pipes, or bridge frames, the system will ensure that the minimum clear spacing between them is maintained at 100 mm. For the distance between the outer wall of the pipeline and the inner surface of the building structure, such as the distance between the top of the pipeline and the bottom of the beam, the system will ensure that the minimum clear spacing is maintained at 50 mm. These spacing values reserve the necessary space for subsequent installation operations, insulation construction, and daily maintenance, avoiding on-site modifications due to insufficient operation space from the design source.

[0069] This embodiment improves the collision detection accuracy and strictly implements the installation and maintenance spacing standards, so that the collision detection range is expanded from simple geometric body interference to construction feasibility and convenience. This method can identify and eliminate soft collision problems that cannot be installed or maintained due to insufficient spacing, even if they do not directly contact, at the design stage, thereby generating a deepened design model that is truly constructible and easy to maintain, significantly reducing on-site modifications, rework, and material waste due to space conflicts during the construction phase, and improving installation quality and efficiency.

[0070] The rail transit electromechanical installation method based on BIM, point cloud scanning and lofting robot of the application is applied to a certain rail transit electromechanical installation project. The project profile is as follows: the total length of the line is 22.43 km, including 20 underground stations, involving the installation of multi-specialty electromechanical pipelines such as ventilation, water supply and drainage, cable bridge, etc. The results show that the electromechanical installation period of a single station is shortened from 45 days to 28 days; the cost of dismantling and modification is reduced by 2.1 million yuan, and the labor cost is reduced by 40%; the qualification rate of pipeline installation verticality and flatness is improved from 82% to 100%.

[0071] The number of devices and the scale of processing described here are used to simplify the description of the application. The application, modification and variation of the rail transit electromechanical installation method based on BIM, point cloud scanning and lofting robot of the application to those skilled in the art are obvious.

[0072] Although the embodiments of the application have been disclosed as above, they are not limited to the application and implementation listed in the specification and embodiments, and can be fully applied to various fields suitable for the application, and additional modifications can be easily realized by those skilled in the art, therefore the application is not limited to specific details without departing from the general concept defined by the claims and equivalent scope.

Claims

1. A rail transit electromechanical installation method based on BIM, point cloud scanning and lofting robots, characterized in that, The application relates to a method for constructing a three-dimensional BIM model of a building structure and mechanical and electrical pipelines, performing mechanical and electrical pipeline comprehensive arrangement and collision detection, and generating an initial design model, wherein the three-dimensional BIM model adopts a coordinate system unified with field measurement. The method comprises the following steps: collecting point cloud data of the structure of the construction site by using a three-dimensional point cloud scanner, wherein the point position accuracy of the point cloud data is 1-2 mm; performing point cloud splicing and noise reduction processing by using a point cloud processing software, wherein the point cloud splicing error is controlled within a range of 2-4 mm, and a field point cloud model is generated; superimposing and comparing the field point cloud model and the initial design model in the unified coordinate system; extracting structure feature points based on the point cloud data and performing deviation analysis on the initial design model; identifying structure beam deviation and reserved hole position deviation; setting a structure beam deviation identification early warning value as 5-30 mm; adjusting the arrangement of the mechanical and electrical pipelines and the position of the support hangers in the initial design model according to the comparison result; and generating a deepening model. The method further comprises the following steps: performing pipeline segmentation based on the deepening model, wherein the segmentation length is determined as 1-6 m according to transportation limitation and hoisting requirement; generating a prefabricated pipeline component processing list and assembly drawing; processing the prefabricated pipeline component by using a numerical control plasma cutting machine, wherein the processing accuracy is controlled within a range of 1-3 mm, and a two-dimensional code mark containing component number and material information is additionally attached to the processed prefabricated pipeline component; and importing the coordinate data in the deepening model into a lofting robot control system through a data verification interface, wherein the data verification interface performs format conversion and accuracy verification on the coordinate data, the lofting robot control system receives GPS signals and laser positioning data, analyzes the coordinate data and generates three-dimensional lofting points, and the positioning accuracy of the lofting points is 1-2 mm; and a field operator installs the support hangers and the prefabricated pipeline component according to the position of the lofting points. During the point cloud splicing and noise reduction processing, a feature point-based automatic matching algorithm is used to register the point cloud data of adjacent scanning stations, and after registration, a voxel grid downsampling method is used for point cloud density uniform processing, and the point cloud density is controlled within a range of 5000-8000 points per square meter. The feature point-based automatic matching algorithm comprises the following steps: extracting corner points and edge feature points in the point cloud data of adjacent scanning stations as matching primitives; calculating an FPFH feature descriptor of each feature point by using an ISS feature point detection algorithm; establishing a preliminary correspondence between the feature points of the source point cloud and the target point cloud based on the similarity of the FPFH feature descriptors; screening the preliminary correspondence and calculating a coarse registration transformation matrix by using a RANSAC algorithm; and performing fine registration by using an ICP algorithm, taking the coarse registration result as an initial pose, and setting an iteration error threshold value of the fine registration as 2-4 mm. During the deviation analysis, the structure beam deviation identification early warning value is set as 10 mm, and when the identified structure beam deviation exceeds the early warning value, a deviation adjustment scheme is automatically generated, and the adjustment scheme comprises a mechanical and electrical pipeline elevation adjustment amount and a support hanger position correction amount.

2. The rail transit electromechanical installation method based on BIM, point cloud scanning and lofting robot according to claim 1, characterized in that, ​ 3. The rail transit electromechanical installation method based on BIM, point cloud scanning and lofting robot according to claim 2, characterized in that, ​ 4. The rail transit electromechanical installation method based on BIM, point cloud scanning and lofting robot according to claim 1, characterized in that, ​ 5. The rail transit electromechanical installation method based on BIM, point cloud scanning and lofting robot according to claim 4, characterized in that, The specific process of automatically generating the deviation adjustment scheme comprises: according to the difference ΔH between the structural beam offset and the early warning value, calculating the pipeline elevation adjustment amount according to the pipeline slope compensation algorithm, the elevation adjustment amount = ΔH + L × i min , wherein L is the length of the affected pipe section, i min is the minimum slope allowed by the pipeline system, and the value range is 0.3% to 0.5%; at the same time, according to the support hanger design distance D and the offset direction, calculating the support hanger position correction amount, the position correction amount = D × tan θ, wherein θ is the included angle between the offset direction and the pipeline axis; the elevation adjustment amount calculated needs to ensure that the compensated pipeline slope is within the range of 0.3% to 1.0%, and the distance deviation of the support hanger position correction is controlled within ±50mm.

6. The rail transit electromechanical installation method based on BIM, point cloud scanning and lofting robot according to claim 1, characterized in that, When the pipeline is segmented, the following steps are further included: in the deepening model, a three-dimensional docking mark containing the flange specifications and bolt hole positions is added to each pipe segment connection; the maximum length of the pipe segment is determined to be 6 meters based on the transportation restrictions, and the maximum weight of a single pipe segment is determined to be 800 kg based on the lifting capacity; the finite element analysis is used to simulate the stress state of the pipe segment during lifting, and the analysis indicators include the maximum deflection and the equivalent stress, wherein the maximum deflection limit is 1 / 400 of the length of the pipe segment, and the equivalent stress limit is 60% of the yield strength of the pipe material; when the simulation result exceeds the limit, the following steps are performed: first, add a ring-shaped reinforcing support in the middle of the pipe segment, the support spacing is set to 2-3 meters, and the support sectional moment of inertia is not less than 1.2 times the moment of inertia of the pipe segment body; re-perform finite element analysis verification; if the ring-shaped reinforcing support is still not satisfied after adding, re-segment the middle of the pipe segment, and the new segment length is controlled within 3-4 meters, and the segment interface flange design is regenerated.

7. The rail transit electromechanical installation method based on BIM, point cloud scanning and lofting robot according to claim 1, characterized in that, The data verification interface uses the CRC-32 verification algorithm to verify the transmitted coordinate data, and when the single-point coordinate data error exceeds 2 mm, the automatic retransmission mechanism is triggered; the retransmission mechanism includes: resending the data packet after a 100 ms delay, and the maximum retransmission number is set to 3 times; if the data error still exceeds 2 mm after retransmission 3 times, the coordinate point is marked as an abnormal point, and the precast pipe component number and coordinate value of the point are recorded; a data integrity report is generated after the verification is passed, which records the transmission time, data transmission amount, number of abnormal points, and specific coordinate value and corresponding precast pipe component number of each abnormal point; the lofting robot control system automatically skips the coordinate position marked as an abnormal point after receiving the data integrity report, and displays the abnormal point list and suggests manual measurement prompt on the operation interface.

8. The rail transit electromechanical installation method based on BIM, point cloud scanning and lofting robot according to claim 1, characterized in that, When the lofting robot control system generates three-dimensional lofting points, a multi-source data fusion algorithm is used to comprehensively process GPS signals and laser positioning data, and when there is no GPS signal, it automatically switches to a full laser positioning mode.

9. The rail transit electromechanical installation method based on BIM, point cloud scanning and lofting robot according to claim 1, characterized in that, When performing comprehensive arrangement and collision detection of mechanical and electrical pipelines, the Navisworks software is used for collision detection of mechanical and electrical pipelines and between the pipelines and the building structure, and the collision detection accuracy is set to 3-5 mm; for the detected collision points, automatic arrangement optimization is performed according to the pipeline comprehensive principle, wherein the distance between adjacent pipelines is controlled to be 100-150 mm, and the distance between the pipeline and the building structure is controlled to be 50-100 mm.