BIM deepening design and construction collaborative integration system and method

Abstract

This invention relates to the field of building information and construction management technology, specifically disclosing a BIM-based integrated system and method for detailed design and construction collaboration. The system acquires time-series micro-deformation data and thermal diffusion image data of the waterproofing layer, fusing them to generate a hidden defect cloud map. It then associates the hidden defect cloud map with building information component identifiers, updating and generating dynamic building information objects. Based on defect information and real-time resources, it performs spatiotemporal coupled finite element simulation calculations to generate rectification task instructions. These instructions are pushed to the on-site terminal, receiving re-inspection data and comparing it with the hidden defect cloud map for verification, generating re-inspection qualification confirmation information. Finally, the entire process data is packaged and stored in a distributed ledger via a blockchain smart contract, generating a trusted traceability identifier. This invention achieves accurate identification of deep hidden defects, precise association of defect information with components, dynamic scheduling of rectification resources, and tamper-proof trusted traceability of the entire process data, solving technical problems in the quality control of concealed works.

Description

Technical Field

[0001] This invention relates to the field of building information and construction management technology, specifically to an integrated system and method for BIM-based detailed design and construction collaboration. Background Technology

[0002] In the construction process of building projects, the construction of waterproofing layers on basement floors, roofs, and bathrooms is a critical step affecting the durability and functionality of the building structure, and is an important component of concealed works. The quality of the waterproofing layer construction directly affects the risk of leakage after the building is put into use. Once leakage occurs, the later repair costs are high and difficult to completely eliminate. Therefore, strict quality inspection of the waterproofing layer before it is covered by subsequent construction processes is a fundamental requirement for project quality control.

[0003] The existing technology has the following shortcomings: Current methods for inspecting concealed waterproofing layers rely on manual visual inspection and tapping to detect surface defects. This approach only uncovers macroscopic surface defects and leaves blind spots for deeper, hidden defects such as voids and peeling beneath the adhesive layer, resulting in a large number of quality risks being overlooked. Furthermore, the inspection information is disconnected from the physical location of the building, making it impossible to accurately link defect data to specific building components. The rectification process also lacks dynamic awareness of real-time resource status, leading to delays in resource allocation. More seriously, the entire process of defect discovery, rectification, and re-inspection is data-driven, stored in paper or centralized electronic records, making it susceptible to tampering and difficult to trace. This results in the digital archives of concealed works lacking crucial physical field information about latent defects, failing to provide reliable underlying data support for the building's entire lifecycle maintenance. Summary of the Invention

[0004] The purpose of this invention is to provide an integrated system and method for BIM-based detailed design and construction collaboration to solve the problems mentioned above.

[0005] The objective of this invention can be achieved through the following technical solutions: The BIM-based integrated system and methodology for detailed design and construction collaboration includes the following steps: S1: Acquire the time-series micro-deformation data and thermal diffusion image data of the waterproof layer in the target construction area. Generate a micro-deformation field distribution map by performing interferometric radar analysis on the time-series micro-deformation data, and obtain a thermal anomaly area map by performing convolutional neural network semantic segmentation on the thermal diffusion image data. Spatially register and fuse the micro-deformation field distribution map and the thermal anomaly area map to obtain a hidden defect cloud map. S2: Associate the hidden defect cloud map with the preset building information component identifier, identify the defective components and sub-attribute partitions, and write the defect type, location and confidence information into the attribute fields of the corresponding component, and update and generate a dynamic building information object with defect tracing information. S3: Based on the defect information in the dynamic building information object, combined with the real-time resource location and status, a spatiotemporal coupled finite element simulation calculation is performed to generate rectification task instructions including the optimal path and resource matching; S4: Push the rectification task instruction to the on-site terminal, and receive the re-inspection images and micro-deformation data collected after the rectification is completed according to the rectification task instruction. Verify the rectification effect by comparing the re-inspection data with the hidden defect cloud map, and generate re-inspection qualification confirmation information. S5: Package time-series micro-deformation data, thermal diffusion image data, hidden defect cloud map, rectification task instructions and re-inspection qualification confirmation information into a data block, store it in the distributed ledger through blockchain smart contract, and generate an immutable and trustworthy traceability identifier corresponding to the building information component identifier.

[0006] As a further aspect of the present invention: S1 specifically includes: Deformation curvature analysis is performed on the micro-deformation field distribution map to identify the boundary of the region where the deformation gradient exceeds the preset threshold, and the boundary of the region is extracted as a set of spatial reference feature points. Geometric distortion parameters are calculated for the thermal anomaly region map, and spatial coordinate remapping is performed on the thermal anomaly region in the thermal anomaly region map based on the spatial reference feature point set to generate a thermal anomaly correction map with the same coordinate system as the micro-deformation field distribution map. The micro-deformation field distribution map and the thermal anomaly correction map are spatially registered at the pixel level. The registered dual-source data are then cascaded and weighted by confidence to generate a hidden defect cloud map.

[0007] As a further aspect of the present invention: S2 specifically includes: Morphological connectivity analysis is performed on continuous pixel regions in the hidden defect cloud map to extract the spatial contour and geometric center coordinates of each independent defect patch and generate a list of defect units. Traverse the list of defective units, perform a three-dimensional Boolean intersection operation between the spatial outline of each defective patch and the spatial occupancy boundary corresponding to the preset building information component identifier, and assign the defective patch to the corresponding component identifier based on the maximum intersection ratio. Cluster analysis was performed on each defect patch belonging to the same component identifier to calculate its intensity peak location and distribution density, and sub-attribute partitions were delineated under the corresponding component surface. The type, geometric center coordinates, and confidence level of the defect patch are written as attribute values ​​into the attribute field of the corresponding component, forming a dynamic building information object containing component identifier, sub-attribute partitions, and defect parameters.

[0008] As a further aspect of the present invention: the step of dividing the sub-attribute partitions under the surface of the corresponding component specifically includes: For each defect patch belonging to the same component identifier, the pixel intensity value and spatial coordinates of each defect patch are extracted to construct the spatial distribution field of defect intensity corresponding to the component. Local maxima search is performed on the spatial distribution field of defect intensity to obtain the coordinates of the intensity peak point of each defect patch, and the intensity gradient descent direction is calculated layer by layer from the intensity peak point outward. Based on the direction of intensity gradient descent and the preset intensity threshold, the continuously distributed defect area is divided into a core area containing the intensity peak point, a transition area with intensity below the threshold, and a defect boundary area. The core area, transition area, and defect boundary area are assigned different sub-attribute partition identifiers, and the spatial contour and strength characteristic parameters of each partition are written into the component attribute field.

[0009] As a further aspect of the present invention: S3 specifically includes: Extract the defect type, peak intensity, and spatial profile of each sub-attribute partition from the dynamic building information object, and map them to the boundary conditions and load parameters of the area to be repaired. Obtain the real-time resource location coordinates and resource status attributes within the construction area, calculate the spatial distance between the real-time resource location coordinates and the spatial outline of the area to be repaired, and generate a resource reachability time series matrix. The boundary conditions and load parameters are loaded into the digital simulation space of the construction area, and the response time of each resource point is solved by finite element iteration based on the resource accessibility time series matrix. Multiple candidate repair schemes that satisfy the structural stress constraints and their corresponding construction time series are output. A weighted comparison of construction path length and resource occupation time is performed on multiple candidate repair schemes, and the scheme with the lowest overall cost is selected to generate a rectification task instruction containing resource scheduling instructions and path navigation data.

[0010] As a further aspect of the present invention: the step of performing finite element iterative solution on the response time series of each resource point based on the resource reachability time series matrix specifically includes: Boundary conditions and load parameters are loaded into the digital simulation space, and the load application timing is set according to the timestamp of the earliest available resource point in the resource accessibility time series matrix. At the initial iteration moment, the stress field distribution in the digital simulation space is calculated, and the stress peak value and stress gradient direction within a preset range around the area to be repaired are extracted. The load application location and magnitude are corrected according to the stress gradient direction at the next iteration time, and the remaining available time of the corresponding resource point in the resource reachability time series matrix is ​​updated simultaneously. Repeat the stress field distribution calculation and load application position correction until the change in the stress peak obtained from two consecutive iterations is less than the preset convergence threshold. Then, output the load application sequence of each resource point at this time as the repair construction sequence.

[0011] As a further aspect of the present invention: S4 specifically includes: After the rectification task instruction is pushed to the on-site terminal, the re-inspection images and re-inspection micro-deformation data collected after the rectification area is repaired according to the rectification task instruction are obtained, and the re-inspection images and re-inspection micro-deformation data are spatially registered to generate a re-inspection registration dataset. The re-verification and registration dataset and the hidden defect cloud map are input into the comparison and verification space, pixel-level difference operation is performed, and the difference pixel areas with gray level differences exceeding the preset tolerance threshold are extracted to generate a rectification difference heat map. Morphological features are extracted from each continuous pixel region in the rectification difference heat map to obtain the area, perimeter and average gray level difference of each region, and then compared with the preset qualified acceptance threshold item by item. When the morphological features of all consecutive pixel regions are less than or equal to the acceptance threshold, a re-inspection acceptance confirmation information including a rectification difference heat map and comparison results is generated.

[0012] As a further aspect of the present invention: the step of inputting the re-verification registration dataset and the hidden defect cloud map into the comparison verification space and performing pixel-level difference operations specifically includes: The re-verification registration dataset and the hidden defect cloud map are respectively processed by pixel grayscale normalization, so that the re-verification registration dataset and the hidden defect cloud map are numerically aligned under the same grayscale scale, and normalized re-verification map and normalized defect map are generated. The gray values ​​of the normalized verification map and the normalized defect map are subtracted pixel by pixel to obtain the gray value difference at each pixel position, and an initial gray value difference matrix is ​​generated. The initial gray-level difference matrix is ​​subjected to neighborhood weighted smoothing filtering to eliminate isolated noise points caused by residual errors in pixel-level registration, and a filtered gray-level difference smoothing matrix is ​​generated. Traverse all pixels in the grayscale difference smoothing matrix, mark consecutive pixels with grayscale differences exceeding the preset tolerance threshold as difference pixel regions, extract the contour boundaries of all marked regions and assign region labels to generate a rectification difference heat map.

[0013] As a further aspect of the present invention: S5 specifically includes: The time-series micro-deformation data, thermal diffusion image data, hidden defect cloud map, rectification task instructions and re-inspection qualification confirmation information are linked together according to the timestamp order to generate a full-process traceability data chain for the waterproof layer in the construction area. Perform hash calculations on the entire process traceability data chain, extract the hash values ​​of each data node and construct a Merkle tree, and write the root hash value of the Merkle tree into the calling parameters of the blockchain smart contract; Trigger the execution of the storage operation of the blockchain smart contract, and return the Merkel root hash value and the corresponding block height to the local database as a data storage certificate for the waterproof layer in the construction area; A QR code is generated based on the building information component identifier and data storage certificate. The QR code is then physically bound to the building information component identifier, serving as a reliable traceability identifier for subsequent scanning to retrieve tamper-proof data.

[0014] As a further aspect of the present invention: the triggering of the blockchain smart contract to execute the storage operation, returning the Merkle root hash value and the corresponding block height to the local database as data storage evidence for the waterproof layer in the construction area, specifically includes: The Merkle root hash value and the data length of the end-to-end traceability data chain are encapsulated as call parameters of the smart contract, and a storage request is sent to the distributed ledger to trigger the smart contract to perform the data on-chain operation. Listen to the transaction receipts returned by the distributed ledger, parse the block number storing the Merkle root hash value and the corresponding block's position offset in the chain from the transaction receipts, and combine them to generate a storage position pointer containing the block number and position offset. The storage location pointer is associated with the Merkle root hash value and written to the index table of the local database in the form of key-value pairs, where the key is the Merkle root hash value and the value is the storage location pointer. Extract the record entries corresponding to the Merkel root hash value from the local database index table, and encapsulate the record entries and the current timestamp into a data storage certificate data package, which serves as the data storage certificate for the waterproof layer in the construction area.

[0015] The beneficial effects of this invention are: (1) This invention integrates interferometric radar micro-deformation monitoring with active thermal imaging semantic segmentation to generate a hidden defect cloud map that simultaneously characterizes deformation anomalies and thermal diffusion anomalies. This enables accurate identification and visual positioning of deep hidden defects such as voids and peeling under the adhesive layer of the waterproof layer. It overcomes the limitation that traditional visual observation can only detect surface wrinkles and hollows, and elevates defect detection from surface qualitative judgment to deep quantitative analysis, thereby improving the early detection rate of hidden engineering quality hazards.

[0016] (2) This invention dynamically binds defect information with building information component identification and divides the core area, transition area and boundary area based on the defect intensity distribution characteristics, thereby realizing component-level fine storage and traceable management of defect information; combined with pixel-level differential comparison and verification of re-inspection data and hidden defect cloud map, as well as on-chain storage of data throughout the process, an immutable and reliable traceability system is constructed from defect discovery, rectification instructions, on-site execution to re-inspection closed loop, ensuring that the hidden works acceptance data and the physical quality are of the same true origin. Attached Figure Description

[0017] The invention will now be further described with reference to the accompanying drawings.

[0018] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Please see Figure 1 As shown, this invention is an integrated system and method for BIM detailed design and construction collaboration, comprising the following steps: S1: Acquire the time-series micro-deformation data and thermal diffusion image data of the waterproof layer in the target construction area. Generate a micro-deformation field distribution map by performing interferometric radar analysis on the time-series micro-deformation data, and obtain a thermal anomaly area map by performing convolutional neural network semantic segmentation on the thermal diffusion image data. Spatially register and fuse the micro-deformation field distribution map and the thermal anomaly area map to obtain a hidden defect cloud map. S2: Associate the hidden defect cloud map with the preset building information component identifier, identify the defective components and sub-attribute partitions, and write the defect type, location and confidence information into the attribute fields of the corresponding component, and update and generate a dynamic building information object with defect tracing information. S3: Based on the defect information in the dynamic building information object, combined with the real-time resource location and status, a spatiotemporal coupled finite element simulation calculation is performed to generate rectification task instructions including the optimal path and resource matching; S4: Push the rectification task instruction to the on-site terminal, and receive the re-inspection images and micro-deformation data collected after the rectification is completed according to the rectification task instruction. Verify the rectification effect by comparing the re-inspection data with the hidden defect cloud map, and generate re-inspection qualification confirmation information. S5: Package time-series micro-deformation data, thermal diffusion image data, hidden defect cloud map, rectification task instructions and re-inspection qualification confirmation information into a data block, store it in the distributed ledger through blockchain smart contract, and generate an immutable and trustworthy traceability identifier corresponding to the building information component identifier.

[0021] In S1, time-series micro-deformation data and thermal diffusion image data of the waterproof layer in the target construction area are acquired. Interferometric radar analysis is performed on the time-series micro-deformation data to generate a micro-deformation field distribution map. Convolutional neural network semantic segmentation is used on the thermal diffusion image data to obtain a thermal anomaly region map. Spatial registration and fusion of the micro-deformation field distribution map and the thermal anomaly region map are then performed to obtain a hidden defect cloud map. When detecting hidden defects in a waterproofing layer, it is first necessary to acquire time-series micro-deformation data and thermal diffusion image data of the waterproofing layer in the target construction area. The time-series micro-deformation data is acquired as follows: During the curing period after the waterproofing layer is constructed, at least four fixed reference points are set up around the target area. A vehicle-mounted millimeter-wave interferometric radar is used to continuously scan the surface of the waterproofing layer twice a day, continuously acquiring radar echo signals for at least seven days. By interferometric calculation of the phase difference of the radar echo signals at different time periods, the axial micro-deformation displacement of each sampling point on the surface of the waterproofing layer is obtained, forming a time-series micro-deformation dataset with timestamps. The thermal diffusion image data is acquired as follows: During a period of stable ambient temperature at night, a handheld active thermal imaging device is used to apply controllable thermal excitation to the surface of the waterproofing layer. Then, a thermal imager continuously acquires a sequence of surface temperature field changes within 30 seconds after the thermal excitation. Each sampling point is repeatedly acquired three times, and the average temperature is taken as the thermal diffusion characteristic value of that point, forming a thermal diffusion image dataset.

[0022] The collected time-series micro-deformation dataset is input into the data processing unit. Spatial interpolation is performed on the micro-deformation displacement of each sampling point to construct a continuously distributed micro-deformation field distribution map. Deformation curvature analysis is then performed on this micro-deformation field distribution map. Specifically, this is achieved by: traversing each pixel in the micro-deformation field distribution map, calculating the second-order difference value of the displacement between that pixel and its eight neighboring pixels, and using this second-order difference value as the deformation curvature of that point; setting a deformation gradient threshold, which is determined based on 60% of the ultimate elongation measured in the tensile test of waterproof materials, and marking consecutive pixels with deformation curvature exceeding this threshold as high-gradient regions; performing edge detection on the high-gradient regions, extracting their outer contour boundaries, and using these boundaries as the boundaries of deformed anomaly regions; and recording the coordinates of all extracted region boundaries as a spatial reference feature point set.

[0023] The collected thermal diffusion image dataset is input into the image processing unit. A pre-built encoder-decoder structure is used to perform semantic segmentation on the thermal diffusion sequence images. This encoder-decoder structure is trained using an image library containing various waterproofing layer defect samples. Its output is a thermal anomaly region map marked with thermal anomaly areas. Geometric distortion parameters are calculated from this thermal anomaly region map. Specifically, corner features are extracted from the thermal anomaly region map, and these corner features are matched with corresponding points in a spatial reference feature point set. Based on the spatial coordinate difference between the matched point pairs, a polynomial fitting method is used to solve for the geometric distortion coefficients of the thermal anomaly region map. These distortion coefficients include horizontal scaling coefficients, vertical scaling coefficients, and rotation angles. Based on the solved geometric distortion coefficients, spatial remapping calculations are performed on the pixel coordinates of each thermal anomaly region in the thermal anomaly region map to generate a thermal anomaly correction map with the same coordinate system and pixel resolution as the micro-deformation field distribution map.

[0024] It should be noted that the convolutional neural network adopts the U-Net architecture. Its encoder consists of four downsampling modules, each containing two 3×3 convolutional layers (ReLU activation) and one 2×2 max-pooling layer, used to extract multi-scale features from the image. The decoder consists of four upsampling modules, each containing one 2×2 deconvolutional layer and two 3×3 convolutional layers. Skip connections are used to concatenate the feature maps of corresponding layers in the encoder to recover detailed image information. The last layer of the network is a 1×1 convolutional layer, using a sigmoid activation function to output a probability map of each pixel belonging to a thermal anomaly region, i.e., a thermal anomaly region map. The network is trained using a training set containing 5000 images of thermal diffusion of the waterproofing layer, where defect regions were annotated at the pixel level by inspectors. Training uses the Adam optimizer with an initial learning rate of 0.001, and the loss function is a weighted sum of Dice Loss and binary cross-entropy loss. The training process lasted for 200 epochs, and early stopping was used to prevent overfitting. The final model achieved a pixel segmentation accuracy of over 95% on the validation set.

[0025] The micro-deformation field distribution map and the thermal anomaly correction map are input into the fusion processing unit for pixel-level spatial registration. Specifically, the micro-deformation field distribution map is used as a reference, and the thermal anomaly correction map is aligned with it pixel by pixel to ensure that points representing the same physical location in the two images have the same row and column coordinates. After registration, the dual-source data are cascaded and weighted by confidence. Specifically, the deformation displacement of each pixel in the micro-deformation field distribution map is extracted as the first feature channel, and the temperature difference of each pixel in the thermal anomaly correction map is extracted as the second feature channel. The two feature channels are then cascaded at the pixel level. A fusion weight is assigned to each feature channel, with the weight of the micro-deformation displacement feature set at 60% based on its signal-to-noise ratio, and the weight of the thermal anomaly feature set at 40% based on the inverse of the variance of repeated acquisitions. The values ​​of the two feature channels are multiplied by their respective weights and then summed to obtain the fused pixel value. The fused pixel value is then converted into image pixels according to a preset grayscale mapping relationship, generating a hidden defect cloud map that simultaneously represents the spatial distribution of deformation anomalies and thermal anomalies.

[0026] In S2, the hidden defect cloud map is associated with the preset building information component identifiers to identify defective components and sub-attribute partitions. The defect type, location, and confidence level information are written into the corresponding component's attribute fields, updating and generating a dynamic building information object with defect tracing information. Specifically, this includes: When associating defect information with building information components, it is first necessary to extract independent defect areas from the hidden defect cloud map. The hidden defect cloud map is input into the image processing unit, and morphological connectivity analysis is performed on the continuous pixel regions in the image. Specifically, the following steps are taken: A four-connected neighborhood judgment criterion is used to traverse each pixel in the hidden defect cloud map. When the gray value of a pixel is higher than a preset defect judgment threshold, the pixel is marked as a candidate defect pixel. The defect judgment threshold is set based on the gray value statistical mean of artificial defect samples of waterproof material standard specimens. Adjacent candidate defect pixels are merged into the same connected region. Boundary tracking is performed on each merged connected region to extract its outer contour pixel coordinate sequence. The area enclosed by this sequence is taken as the spatial contour of the defect patch. The geometric center coordinates of all pixels contained in the spatial contour are calculated. The geometric center coordinates are obtained by dividing the sum of the row coordinates of all pixels in the contour by the number of pixels to obtain the center row coordinates, and the sum of the column coordinates of all pixels by the number of pixels to obtain the center column coordinates. The spatial contour data and geometric center coordinate data of each defect patch are combined into a record. The set of records of all defect patches constitutes a defect unit list.

[0027] After obtaining the list of defective units, each defective patch needs to be assigned to a specific building information component. Each defective patch in the list is traversed, and its spatial contour data is read. Simultaneously, the component identifiers and their spatial occupancy boundaries corresponding to the construction area are retrieved from the preset building information database. These spatial occupancy boundaries are described by the component's three-dimensional geometric parameters, including the component's length, width, height, and its positioning coordinates in the global coordinate system. The spatial contour of the defective patch and the spatial occupancy boundary of each component identifier are input into a three-dimensional Boolean operation unit for intersection calculation. Specifically, the spatial contour of the defective patch is treated as a two-dimensional planar region, stretched vertically along the thickness range of the component to form a three-dimensional volume, and the overlap volume between this three-dimensional volume and the component's spatial occupancy boundary is calculated. The overlap volume is divided by the total volume of the three-dimensional volume corresponding to the defective patch to obtain the intersection ratio between the defective patch and the current component. After traversing all components, the component with the largest intersection ratio is selected as the component to which the defective patch is assigned. If the maximum intersection ratio is less than 50%, the defective patch is determined to be a boundary ambiguity defect, marked as pending review, and temporarily stored.

[0028] For each defect patch belonging to the same component identifier, its internal intensity distribution characteristics need further analysis. The pixel intensity values ​​and corresponding spatial coordinates of all defect patches belonging to the same component are extracted. The pixel intensity values ​​are taken from the grayscale value of the pixel location in the hidden defect cloud map, with a value range of 0 to 255. The pixel intensity values ​​of each defect patch are mapped onto a two-dimensional plane according to their spatial coordinates. The blank areas between adjacent pixels are filled using bilinear interpolation to construct a continuous spatial distribution field of defect intensity on the component surface. A local maximum search is performed on this spatial distribution field. Specifically, each grid point in the distribution field is traversed, and the intensity value of that grid point is compared with the intensity values ​​of its eight neighboring grid points. If the intensity value of that grid point is greater than the intensity values ​​of all eight neighboring points, then that grid point is marked as a local intensity peak point. The coordinates and intensity values ​​of all local intensity peak points are recorded as the intensity peak points of each defect patch.

[0029] Based on the intensity peak points obtained from the search, the area covered by each defect patch is partitioned into sub-attributes. Taking each intensity peak point as the center, the intensity gradient descent direction is calculated layer by layer outwards. The intensity gradient descent direction is calculated as follows: for the current grid point, the intensity change rate in the row direction and the intensity change rate in the column direction are calculated respectively. The row direction intensity change rate is obtained by subtracting the intensity value of the left adjacent point from the intensity value of the right adjacent point and dividing by 2; the column direction intensity change rate is obtained by subtracting the intensity value of the upper adjacent point from the intensity value of the lower adjacent point and dividing by 2. The row direction intensity change rate and the column direction intensity change rate are combined into a two-dimensional vector, and the opposite direction of this vector is the intensity gradient descent direction. Intensity thresholds are set, including a first intensity threshold and a second intensity threshold, wherein the first intensity threshold is set to 60% of the intensity value of the current peak point, and the second intensity threshold is set to 20% of the intensity value of the current peak point. Search outward point by point along the descent direction of the intensity gradient. Continuous regions with intensity values ​​higher than the first intensity threshold are classified as core regions. Continuous regions with intensity values ​​between the first and second intensity thresholds are classified as transition regions. Regions with intensity values ​​lower than the second intensity threshold but belonging to the same connected domain of the defect patch are classified as defect boundary regions.

[0030] After completing the sub-attribute partitioning, defect information needs to be written into the attribute fields of the corresponding components. For each defect patch, the defect type is read from the defect unit list to which it belongs. The defect type is determined based on the classification results during the generation of the hidden defect cloud map, including hollow defects, peeling defects, or micro-deformation defects. The geometric center coordinates of the patch are read. The confidence level is calculated based on the ratio of the core area of ​​the patch to the total area of ​​the patch. The confidence level is calculated by dividing the core area by the total area of ​​the patch and then multiplying by 100%. The component identifier corresponding to the component to which the defect patch belongs is found in the building information database. A new defect information sub-table is created under the attribute field of the component identifier. The defect type, geometric center coordinates, confidence level, and spatial contour data and strength characteristic parameters of the core area, transition area, and boundary area contained in the patch are written into this sub-table. After traversing all defect patches and completing the above writing operations, the updated building information data is output to form a dynamic building information object containing component identifiers, sub-attribute partitioning, and defect parameters.

[0031] In S3, based on defect information in dynamic building information objects, and combined with real-time resource location and status, spatiotemporal coupled finite element simulation calculations are performed to generate rectification task instructions including optimal paths and resource matching, specifically including: Before generating rectification task instructions, defect information must first be extracted from the dynamic building information object and converted into parameters required for simulation calculation. The sub-attribute partition data stored in the dynamic building information object are read. These sub-attribute partitions include a core area, a transition area, and a defect boundary area. For each sub-attribute partition, the defect type is extracted from its attribute fields. The defect type is determined based on the classification results when generating the hidden defect cloud map, including hollow defects, peeling defects, or micro-deformation defects. The intensity peak value of the partition is extracted, represented by a grayscale value ranging from 0 to 255. The spatial contour of the partition is extracted, described by a closed sequence of pixel coordinates. The boundary conditions of the area to be repaired are determined based on the defect type and the intensity peak value. The type of boundary condition is set according to the defect type: hollow defects correspond to displacement boundary conditions, peeling defects correspond to traction boundary conditions, and micro-deformation defects correspond to initial strain boundary conditions. The amplitude of the boundary conditions is linearly mapped based on the intensity peak value, with an intensity peak value of 255 corresponding to the maximum amplitude and an intensity peak value of 0 corresponding to zero amplitude. The load application location is determined based on the spatial contour of each sub-attribute partition. All nodes in the core area contour are marked as primary load application points, nodes in the transition area contour are marked as secondary load application points, and nodes in the defect boundary area contour are marked as constraint boundary nodes.

[0032] After obtaining defect information, real-time resource status data within the construction area needs to be collected. Using positioning tags and a base station network deployed at the construction site, the location coordinates of various construction resources are collected in real time. These resources include waterproofing crew members, repair material transport equipment, and compaction tools. The location coordinates are represented in three-dimensional coordinates with an accuracy controlled within 5 centimeters. Simultaneously, the status attributes of each resource point are acquired, including whether it is currently idle, the remaining time of the current task, and the resource type. The remaining time of the current task is calculated by subtracting the current time from the planned completion time of the task currently being performed at the resource point. The spatial contour of the area to be repaired is converted into geometric center coordinates. The horizontal and vertical distance differences between the location coordinates of each resource point and the geometric center coordinates of the area to be repaired are calculated. The square root of the sum of the squares of the horizontal and vertical distance differences yields the straight-line distance between the resource point and the area to be repaired. This straight-line distance is divided by the average moving speed of the resource point to obtain its theoretical response time. The average moving speed is set according to the resource type: 1.5 meters per second for personnel and 0.8 meters per second for equipment. Add the theoretical response time of the resource point to the remaining time of the current task to obtain the available timestamp of the resource point. Arrange the available timestamps of all resource points in chronological order to construct a resource reachability time series matrix. Each row of the matrix corresponds to a resource point, and each column corresponds to a time node. The matrix elements indicate whether the resource point is available at the corresponding time node.

[0033] Boundary conditions and load parameters are loaded into the digital simulation space of the construction area. This digital simulation space is constructed based on the actual geometric dimensions and material parameters of the waterproofing layer in the construction area, and is discretized using a quadrilateral mesh with a mesh size of 5 mm x 5 mm. Based on the timestamp of the earliest available resource point in the resource accessibility time series matrix, the load application sequence for the initial iteration is set, and the load application position corresponding to the earliest available resource point is marked as the active load point for the initial iteration. At the initial iteration, the stress field distribution is calculated in the digital simulation space. Specifically, this is achieved by: constructing an overall stiffness matrix; substituting the displacement boundary conditions into the stiffness matrix for constraint processing to form a corrected stiffness matrix and load vector; solving the linear equations to obtain the displacement components of each node; calculating the stress value at the center point of each element based on the displacement components; and traversing all elements to extract the von Mises stress value at the center point of each element as the stress characterization at that location. The maximum stress value is searched within a preset range around the area to be repaired. The preset range is set as a circular area with a radius of 30 cm centered on the geometric center of the area to be repaired. The maximum stress value found is recorded as the stress peak value at the current iteration time. At the same time, the direction angle of the peak value point relative to the geometric center of the area to be repaired is recorded as the stress gradient direction.

[0034] The load application position and magnitude for the next iteration are corrected based on the stress gradient direction. The load application position is corrected by moving the current load application position in the opposite direction of the stress gradient by a preset step size, which is the distance between two grid nodes (1 cm). This new position is used as the load application point for the next iteration. The load magnitude is corrected by comparing the peak stress at the current iteration with 80% of the material's yield strength. If the peak stress is higher than 80% of the yield strength, the load magnitude for the next iteration is reduced by 10%; if the peak stress is lower than 60% of the yield strength, the load magnitude for the next iteration is increased by 10%; if the peak stress is between these values, the load magnitude remains unchanged. After the load correction is completed, the remaining available time of the corresponding resource point in the resource reachability time series matrix is ​​updated synchronously. The available timestamp of the currently used resource point is increased by the estimated time required for the resource point to perform the current load application task. The estimated time is set according to the load amplitude. For each unit increase in load amplitude, 10 seconds of construction time are added.

[0035] The stress field distribution calculation and load application position correction are repeated until the convergence condition is met. After each iteration, the absolute value of the difference between the stress peak value obtained in the current iteration and the stress peak value obtained in the previous iteration is calculated. This absolute value is divided by the stress peak value of the previous iteration to obtain the stress peak value change. A convergence threshold is set, which is determined based on the material fatigue properties and is set to 2%. When the stress peak value change obtained in two consecutive iterations is less than 2%, the iteration process is considered to have converged, and the iteration is stopped. From the records after the last iteration, the resource point identifier, load application position coordinates, load amplitude, and application time sequence corresponding to each iteration time are extracted. These records are arranged in chronological order to generate the load application sequence for each resource point. This sequence is the repair construction sequence, and each element in the sequence indicates at which time point, by which resource point, at which location, and with what load.

[0036] In S4, rectification task instructions are pushed to the field terminal, and re-inspection images and micro-deformation data collected after rectification is completed according to the rectification task instructions are received. The rectification effect is verified by comparing the re-inspection data with the hidden defect cloud map, and re-inspection qualification confirmation information is generated, specifically including: After the rectification task instruction is sent to the on-site terminal, the construction personnel complete the repair work on the defective areas according to the instructions. After the repair is completed, a re-inspection is required to verify the rectification effect. First, using the same equipment and methods as the initial inspection, data is collected again on the waterproof layer of the target construction area. The re-inspection images are acquired using active thermal imaging equipment. During acquisition, the thermal excitation power is maintained at 500 watts, the excitation duration is 5 seconds, the thermal imager shooting distance is 1.5 meters, and the viewing angle is perpendicular to the surface of the waterproof layer to ensure coverage of the entire rectified area. The re-inspection micro-deformation data is acquired using millimeter-wave interferometric radar. The radar scanning path is consistent with the initial inspection, the scan line spacing is set to 2 centimeters, and the sampling point density is 100 points per square meter. The acquired re-inspection images and re-inspection micro-deformation data are input into the data processing unit for spatial coordinate registration. The specific implementation of the registration is as follows: corner features are extracted from the re-examination image, and the Harris corner detection algorithm is used with a response threshold set to 0.01 to obtain no less than 20 corner coordinates; at the same time, the corresponding corner coordinates are read from the spatial reference feature point set saved during the initial detection; based on the matching point pairs, the least squares method is used to solve the rotation matrix R and translation vector T of the re-examination image relative to the initial coordinate system; the obtained rotation matrix and translation vector are applied to each sampling point of the re-examination micro-deformation data, so that the coordinates of all sampling points are transformed to the same global coordinate system as the hidden defect cloud map, generating a re-examination registration dataset, which contains the aligned re-examination image pixel matrix and the re-examination micro-deformation data point cloud.

[0037] The re-registration dataset and the hidden defect cloud map are input into the comparison verification space, and pixel-level difference operations are performed. First, the pixel grayscale values ​​of the image portion in the re-registration dataset and the hidden defect cloud map are normalized to align their values ​​on the same grayscale scale. The normalization calculation formula is as follows: ; in, Represents the original image in pixel coordinates grayscale value at that location This is the minimum grayscale value of all pixels in the image. The maximum grayscale value. The normalized grayscale value ranges from 0 to 1. The normalized verification map and the normalized defect map are calculated using this formula. Next, the grayscale values ​​of the normalized verification map and the normalized defect map are subtracted pixel-by-pixel, and the absolute value is taken as the grayscale difference value at that location, generating the initial grayscale difference matrix. ; Because pixel-level registration may have residual errors, resulting in isolated noise points, a neighborhood-weighted smoothing filter is needed to smooth the initial grayscale difference matrix. The filter uses a 3×3 neighborhood-weighted average, with the weight coefficients set based on the distance between the pixel and the center pixel. The specific calculation formula is as follows: ; in, These are the weighting coefficients. Indicates the offset in the row direction. Indicates the offset in the column direction, center point weight. Weights of the four neighboring nodes (up, down, left, and right) Weights of the four diagonal points The denominator is the sum of all weights, 16. Pixels after filtering The grayscale difference smoothing value is calculated. A grayscale difference smoothing matrix is ​​obtained by iterating through all pixels. A preset tolerance threshold is set, determined based on the maximum allowable grayscale difference on the surface of the waterproof material. This threshold is calculated by statistically analyzing the grayscale differences of 10 defect-free samples, taking the mean plus three times the standard deviation, and finally setting it to 0.08 (after normalization). All pixels in the grayscale difference smoothing matrix are iterated through, and consecutive pixels with a grayscale difference exceeding 0.08 are marked as difference pixel regions. Adjacent pixels are merged using the 8-connectivity criterion. The contour boundaries of all marked regions are extracted, and each region is assigned a unique identifier to generate a rectification difference heatmap.

[0038] Morphological features were extracted from the generated heatmap of rectification differences. A connected component analysis algorithm was used to identify all continuous pixel regions in the image. For each region, the following morphological features were calculated: area (total number of pixels contained in the region); perimeter (number of pixels on the boundary of the region, obtained by tracing the region contour and summing the Euclidean distances of adjacent boundary pixels); and average gray-level difference (the average difference in gray-level difference across all pixels in the region within the gray-level difference smoothing matrix). The arithmetic mean of the corresponding values ​​is used. Preset acceptance thresholds are established. The area threshold is determined based on the maximum permissible size of residual defects in the waterproofing engineering acceptance specifications, taking the number of pixels corresponding to 10 square centimeters. Calculated at 100 pixels per square centimeter based on image resolution, the area threshold is set to 1000 pixels. The perimeter threshold is derived from the area threshold; assuming the defect shape is approximately circular, the perimeter threshold is set to the square root of the area threshold multiplied by 4. The average grayscale difference threshold is set based on the material's reflectivity, taking the upper limit of the normalized grayscale difference as 0.05. The area, perimeter, and average grayscale difference of each extracted region are compared item by item with their corresponding thresholds.

[0039] When the area of ​​all consecutive pixel regions in the rectification difference heatmap is less than or equal to 1000 pixels, the perimeter is less than or equal to 126 pixels, and the average grayscale difference is less than or equal to 0.05, the rectification of that region is deemed qualified. The rectification difference heatmap, the morphological feature values ​​of each region (including area, perimeter, and average grayscale difference), and the item-by-item comparison results (recording whether each threshold is met) are integrated into the re-inspection qualification confirmation information. This information also includes metadata such as the re-inspection timestamp, operator identification, and re-inspection equipment number, and the final output serves as the input data for step S5. If any indicator exceeds the threshold, the rectification is deemed unqualified, and the system will automatically re-trigger the rectification task instruction until the re-inspection is qualified.

[0040] In S5, time-series micro-deformation data, thermal diffusion image data, hidden defect cloud maps, rectification task instructions, and re-inspection qualification confirmation information are packaged into data blocks. These blocks are then stored in a distributed ledger via a blockchain smart contract, generating an immutable and trustworthy traceability identifier corresponding to the building information component identifier. Specifically, this includes: After the re-inspection and confirmation are completed, all relevant data from the entire construction process need to be integrated and stored. First, all data that needs to be stored is collected from each processing unit, including time-series micro-deformation data and thermal diffusion image data obtained in the initial inspection stage, hidden defect cloud maps generated by fusion processing, rectification task instructions output in step S3, and re-inspection and confirmation information generated in step S4. These data are concatenated in chronological order, with timestamps accurate to the millisecond level. The specific implementation of data concatenation is as follows: the system time recorded when each data point is generated is read, and the earliest time-series micro-deformation data is used as the starting node of the data chain. Subsequently, thermal diffusion image data, hidden defect cloud map, rectification task instructions, and re-inspection qualification confirmation information are arranged in chronological order. Separator markers are inserted between adjacent data. The separator marker consists of three parts: a start flag, a data length field, and an end flag. The start flag is fixed at the hexadecimal value 0x7E, the data length field records the number of bytes of subsequent data, and the end flag is fixed at the hexadecimal value 0x7F. All data and their separator markers are concatenated sequentially to form a complete end-to-end traceability data chain. This data chain is stored in binary format for subsequent hash operations.

[0041] A Merkle tree is constructed for the entire traceability data chain to ensure that the integrity of each data node in the chain can be independently verified. First, the entire traceability data chain is divided into fixed-size data blocks, each 1024 bytes in size. If the last data block is less than 1024 bytes, it is padded with zeros to bring it to 1024 bytes. A hash operation is performed on each data block using the SHA-256 algorithm. The input data block yields a 256-bit hash value, which serves as the leaf node of the Merkle tree. If the number of leaf nodes is odd, the last leaf node is copied and paired with itself. The hash values ​​of adjacent leaf nodes are concatenated by appending the 256-bit hash value of the previous leaf node to the 256-bit hash value of the next leaf node, forming a 512-bit input data. Another SHA-256 hash operation is performed to obtain the parent node's hash value. This pairing and hashing process is repeated layer by layer upwards until a unique top-level hash value is obtained; this value is the root hash value of the Merkle tree. The Merkel root hash value is extracted and used as a unique digital fingerprint for the entire end-to-end traceability data chain, which will then be used for subsequent blockchain evidence storage.

[0042] The Merkle root hash value and the data length of the end-to-end traceability data chain are encapsulated as call parameters for the smart contract. The data length, in bytes, is obtained by calculating the total length of the binary data in the end-to-end traceability data chain. A storage request is sent to the distributed ledger network, triggering the smart contract to perform the data uploading operation. The distributed ledger network uses blockchain platforms such as Ethereum. The smart contract predefines the data storage logic, including writing the Merkle root hash value from the call parameters into the block body and recording the data length as metadata. After sending the storage request, the system enters a waiting state, listening for the transaction receipt returned by the distributed ledger network. The transaction receipt is a data structure containing the transaction execution result. From this, the block number storing the Merkle root hash value is parsed out. The block number is an integer starting from 0 and incrementing, indicating which block the transaction is recorded in. Simultaneously, the offset of the transaction within the corresponding block is parsed out. The offset, in bytes, represents the starting position of the transaction data within the block body. Combine the block number with the location offset to generate a storage location pointer. The pointer format is the block number followed by a hyphen and then the location offset. For example, "128-4096" represents the 4096th byte of block number 128.

[0043] The storage location pointer is associated with the Merkle root hash value and written to the index table of the local database. The local database uses a relational database, such as SQLite or MySQL. A pre-created storage index table is used, with two fields: a key field storing the Merkle root hash value and a value field storing the location pointer. The write operation is implemented as follows: using the Merkle root hash value as the primary key, it checks if the same key-value pair already exists in the index table. If not, an insert operation is performed, writing the key-value pair to the table; if it exists, an update operation is performed, overwriting the original value with the new storage location pointer. After writing, the record corresponding to the current Merkle root hash value is retrieved from the local database index table, and the complete content of the entry is extracted, including the values ​​of the key and value fields. The current system time, accurate to milliseconds, is obtained as a timestamp. The record entry and the current timestamp are encapsulated into a data packet for evidence storage. The data packet format is as follows: it begins with a 16-byte credential identifier, fixed as the hexadecimal value 0x5A; followed by a 32-byte Merkle root hash value; then a variable-length string of storage location pointers; and finally, an 8-byte timestamp. This data packet serves as evidence of the waterproofing layer in the construction area, proving that all inspection and rectification data for that area has been written to the blockchain at a specific point in time and is tamper-proof.

[0044] A QR code identifier is generated based on the Building Information Component (BIC) identifier and the data storage certificate. First, the BIC identifier corresponding to the waterproofing layer of the construction area is extracted from the dynamic building information object. This identifier is a globally unique identifier, formatted as a 32-bit hexadecimal string. The BIC identifier is then concatenated with the Merkle root hash value and storage location pointer from the data storage certificate. The concatenation format is: component identifier followed by a vertical bar separator, then the Merkle root hash value, then another vertical bar separator, and finally the storage location pointer. The concatenated string is input into a QR code encoder, using the QR code encoding standard, version 10, and error correction level set to H, generating a QR code image with a resolution of 300 pixels by 300 pixels. This QR code image is printed on a waterproof self-adhesive label, with a label size of 5 cm by 5 cm. The printed QR code label is then affixed to a prominent location in the corresponding waterproofing layer area at the construction site, such as the side wall of a sump or a structural column, achieving physical binding between the QR code identifier and the BIC identifier. Subsequent supervisors or maintenance personnel can scan the QR code with their mobile phones to retrieve the corresponding end-to-end traceability data chain from the distributed ledger, verify the integrity and authenticity of the data, and achieve tamper-proof and reliable traceability.

[0045] The working principle of this invention is as follows: Temporal micro-deformation data and thermal diffusion image data of the waterproofing layer in the target construction area are acquired. Interferometric radar analysis is performed on the temporal micro-deformation data to generate a micro-deformation field distribution map. Convolutional neural network semantic segmentation is performed on the thermal diffusion image data to obtain a thermal anomaly region map. The micro-deformation field distribution map and the thermal anomaly region map are spatially registered and fused to obtain a hidden defect cloud map. The hidden defect cloud map is associated with preset building information component identifiers to identify defective components and sub-attribute partitions. The defect type, location, and confidence level information are written into the attribute fields of the corresponding components, updating and generating a dynamic building information object with defect tracing information. Based on the dynamic building information... The defect information in the object is combined with the real-time resource location and status for spatiotemporal coupled finite element simulation calculation to generate rectification task instructions including optimal path and resource matching; the rectification task instructions are pushed to the field terminal, and the re-inspection images and re-inspection micro-deformation data collected after the rectification is completed according to the rectification task instructions are received. The rectification effect is verified by comparing the re-inspection data with the hidden defect cloud map, and re-inspection qualification confirmation information is generated; the time-series micro-deformation data, thermal diffusion image data, hidden defect cloud map, rectification task instructions and re-inspection qualification confirmation information are packaged into a data block, stored in the distributed ledger through a blockchain smart contract, and an immutable and trustworthy traceability identifier corresponding to the building information component identifier is generated.

[0046] 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 claims of this invention should still fall within the patent coverage of this invention.

Claims

1. A BIM deepening design and construction collaboration integration system and method, characterized in that, Includes the following steps: S1: Acquire the time-series micro-deformation data and thermal diffusion image data of the waterproof layer in the target construction area. Generate a micro-deformation field distribution map by performing interferometric radar analysis on the time-series micro-deformation data, and obtain a thermal anomaly area map by performing convolutional neural network semantic segmentation on the thermal diffusion image data. Spatially register and fuse the micro-deformation field distribution map and the thermal anomaly area map to obtain a hidden defect cloud map. S2: Associate the hidden defect cloud map with the preset building information component identifier, identify the defective components and sub-attribute partitions, and write the defect type, location and confidence information into the attribute fields of the corresponding component, and update and generate a dynamic building information object with defect tracing information. S3: Based on the defect information in the dynamic building information object, combined with the real-time resource location and status, a spatiotemporal coupled finite element simulation calculation is performed to generate rectification task instructions including the optimal path and resource matching; S4: Push the rectification task instruction to the on-site terminal, and receive the re-inspection images and micro-deformation data collected after the rectification is completed according to the rectification task instruction. Verify the rectification effect by comparing the re-inspection data with the hidden defect cloud map, and generate re-inspection qualification confirmation information. S5: Package time-series micro-deformation data, thermal diffusion image data, hidden defect cloud map, rectification task instructions and re-inspection qualification confirmation information into a data block, store it in the distributed ledger through blockchain smart contract, and generate an immutable and trustworthy traceability identifier corresponding to the building information component identifier. 2.The BIM deep design and construction collaboration integration system and method according to claim 1, characterized in that, S1 specifically includes: Deformation curvature analysis is performed on the micro-deformation field distribution map to identify the boundary of the region where the deformation gradient exceeds the preset threshold, and the boundary of the region is extracted as a set of spatial reference feature points. Geometric distortion parameters are calculated for the thermal anomaly region map, and spatial coordinate remapping is performed on the thermal anomaly region in the thermal anomaly region map based on the spatial reference feature point set to generate a thermal anomaly correction map with the same coordinate system as the micro-deformation field distribution map. The micro-deformation field distribution map and the thermal anomaly correction map are spatially registered at the pixel level. The registered dual-source data are then cascaded and weighted by confidence to generate a hidden defect cloud map.

3. The BIM-based integrated system and method for detailed design and construction collaboration according to claim 1, characterized in that, S2 specifically includes: Morphological connectivity analysis is performed on continuous pixel regions in the hidden defect cloud map to extract the spatial contour and geometric center coordinates of each independent defect patch and generate a list of defect units. Traverse the list of defective units, perform a three-dimensional Boolean intersection operation between the spatial outline of each defective patch and the spatial occupancy boundary corresponding to the preset building information component identifier, and assign the defective patch to the corresponding component identifier based on the maximum intersection ratio. Cluster analysis was performed on each defect patch belonging to the same component identifier to calculate its intensity peak location and distribution density, and sub-attribute partitions were delineated under the corresponding component surface. The type, geometric center coordinates, and confidence level of the defect patch are written as attribute values ​​into the attribute field of the corresponding component, forming a dynamic building information object containing component identifier, sub-attribute partitions, and defect parameters.

4. The BIM-based integrated system and method for detailed design and construction collaboration according to claim 3, characterized in that, The process of dividing the corresponding component surface into sub-attribute partitions specifically includes: For each defect patch belonging to the same component identifier, the pixel intensity value and spatial coordinates of each defect patch are extracted to construct the spatial distribution field of defect intensity corresponding to the component. Local maxima search is performed on the spatial distribution field of defect intensity to obtain the coordinates of the intensity peak point of each defect patch, and the intensity gradient descent direction is calculated layer by layer from the intensity peak point outward. Based on the direction of intensity gradient descent and the preset intensity threshold, the continuously distributed defect area is divided into a core area containing the intensity peak point, a transition area with intensity below the threshold, and a defect boundary area. The core area, transition area, and defect boundary area are assigned different sub-attribute partition identifiers, and the spatial contour and strength characteristic parameters of each partition are written into the component attribute field.

5. The BIM-based integrated system and method for detailed design and construction collaboration according to claim 1, characterized in that, S3 specifically includes: Extract the defect type, peak intensity, and spatial profile of each sub-attribute partition from the dynamic building information object, and map them to the boundary conditions and load parameters of the area to be repaired. Obtain the real-time resource location coordinates and resource status attributes within the construction area, calculate the spatial distance between the real-time resource location coordinates and the spatial outline of the area to be repaired, and generate a resource reachability time series matrix. The boundary conditions and load parameters are loaded into the digital simulation space of the construction area, and the response time of each resource point is solved by finite element iteration based on the resource accessibility time series matrix. Multiple candidate repair schemes that satisfy the structural stress constraints and their corresponding construction time series are output. A weighted comparison of construction path length and resource occupation time is performed on multiple candidate repair schemes, and the scheme with the lowest overall cost is selected to generate a rectification task instruction containing resource scheduling instructions and path navigation data.

6. The BIM-based integrated system and method for detailed design and construction collaboration according to claim 5, characterized in that, The step of performing finite element iterative solution on the response time series of each resource point based on the resource reachability time series matrix specifically includes: Boundary conditions and load parameters are loaded into the digital simulation space, and the load application timing is set according to the timestamp of the earliest available resource point in the resource accessibility time series matrix. At the initial iteration moment, the stress field distribution in the digital simulation space is calculated, and the stress peak value and stress gradient direction within a preset range around the area to be repaired are extracted. The load application location and magnitude are corrected according to the stress gradient direction at the next iteration time, and the remaining available time of the corresponding resource point in the resource reachability time series matrix is ​​updated simultaneously. Repeat the stress field distribution calculation and load application position correction until the change in the stress peak obtained from two consecutive iterations is less than the preset convergence threshold. Then, output the load application sequence of each resource point at this time as the repair construction sequence.

7. The BIM-based integrated system and method for detailed design and construction collaboration according to claim 1, characterized in that, S4 specifically includes: After the rectification task instruction is pushed to the on-site terminal, the re-inspection images and re-inspection micro-deformation data collected after the rectification area is repaired according to the rectification task instruction are obtained, and the re-inspection images and re-inspection micro-deformation data are spatially registered to generate a re-inspection registration dataset. The re-verification and registration dataset and the hidden defect cloud map are input into the comparison and verification space, pixel-level difference operation is performed, and the difference pixel areas with gray level differences exceeding the preset tolerance threshold are extracted to generate a rectification difference heat map. Morphological features are extracted from each continuous pixel region in the rectification difference heat map to obtain the area, perimeter and average gray level difference of each region, and then compared with the preset qualified acceptance threshold item by item. When the morphological features of all consecutive pixel regions are less than or equal to the acceptance threshold, a re-inspection acceptance confirmation information including a rectification difference heat map and comparison results is generated.

8. The BIM-based integrated system and method for detailed design and construction collaboration according to claim 7, characterized in that, The step of inputting the re-verification registration dataset and the hidden defect cloud map into the comparison verification space and performing pixel-level difference operations specifically includes: The re-verification registration dataset and the hidden defect cloud map are respectively processed by pixel grayscale normalization, so that the re-verification registration dataset and the hidden defect cloud map are numerically aligned under the same grayscale scale, and normalized re-verification map and normalized defect map are generated. The gray values ​​of the normalized verification map and the normalized defect map are subtracted pixel by pixel to obtain the gray value difference at each pixel position, and an initial gray value difference matrix is ​​generated. The initial gray-level difference matrix is ​​subjected to neighborhood weighted smoothing filtering to eliminate isolated noise points caused by residual errors in pixel-level registration, and a filtered gray-level difference smoothing matrix is ​​generated. Traverse all pixels in the grayscale difference smoothing matrix, mark consecutive pixels with grayscale differences exceeding the preset tolerance threshold as difference pixel regions, extract the contour boundaries of all marked regions and assign region labels to generate a rectification difference heat map.

9. The BIM-based integrated system and method for detailed design and construction collaboration according to claim 1, characterized in that, S5 specifically includes: The time-series micro-deformation data, thermal diffusion image data, hidden defect cloud map, rectification task instructions and re-inspection qualification confirmation information are linked together according to the timestamp order to generate a full-process traceability data chain for the waterproof layer in the construction area. Perform hash calculations on the entire process traceability data chain, extract the hash values ​​of each data node and construct a Merkle tree, and write the root hash value of the Merkle tree into the calling parameters of the blockchain smart contract; Trigger the execution of the storage operation of the blockchain smart contract, and return the Merkel root hash value and the corresponding block height to the local database as a data storage certificate for the waterproof layer in the construction area; A QR code is generated based on the building information component identifier and data storage certificate. The QR code is then physically bound to the building information component identifier, serving as a reliable traceability identifier for subsequent scanning to retrieve tamper-proof data.

10. The BIM-based integrated system and method for detailed design and construction collaboration according to claim 9, characterized in that, The triggering of the blockchain smart contract to execute the storage operation, returning the Merkle root hash value and the corresponding block height to the local database as data storage evidence for the waterproofing layer in the construction area, specifically includes: The Merkle root hash value and the data length of the end-to-end traceability data chain are encapsulated as call parameters of the smart contract, and a storage request is sent to the distributed ledger to trigger the smart contract to perform the data on-chain operation. Listen to the transaction receipts returned by the distributed ledger, parse the block number storing the Merkle root hash value and the corresponding block's position offset in the chain from the transaction receipts, and combine them to generate a storage position pointer containing the block number and position offset. The storage location pointer is associated with the Merkle root hash value and written to the index table of the local database in the form of key-value pairs, where the key is the Merkle root hash value and the value is the storage location pointer. Extract the record entries corresponding to the Merkel root hash value from the local database index table, and encapsulate the record entries and the current timestamp into a data storage certificate data package, which serves as the data storage certificate for the waterproof layer in the construction area.

Images