A BIM data-driven engineering cost adjustment method and system

Abstract

The present application relates to the technical field of engineering cost management, in particular to a kind of engineering cost adjustment method and system based on BIM data driving, comprising the following steps: reading model file extracts geometric coordinates and material parameters, constructs engineering entity topology network;Establish the mapping index of graphic element and quota library to generate benchmark list;Change difference matrix is generated based on Boolean difference operation, and the network is locked cascade affected component set is traversed;Combining real-time price data executes engineering quantity recalculation and incremental adjustment, and outputs final report.In the present application, by analyzing model geometric properties and topological relations, a dynamic mapping atlas is constructed, and a differentiated calculation logic is automatically triggered during changes, affected data is accurately updated, and the cost model is corrected, solving the problems of manual accounting lag and omission, ensuring data consistency under multi-specialty collaboration, and significantly improving cost control accuracy and dynamic adaptability.

Description

Technical Field

[0001] This invention relates to the field of engineering cost management technology, and in particular to a method and system for adjusting engineering costs based on BIM data. Background Technology

[0002] The field of construction cost management technology involves a series of management activities, including cost forecasting, planning, control, accounting, analysis, and evaluation, throughout the entire process of a construction project. Among these, the traditional BIM data-driven method for adjusting construction costs involves cost estimators manually entering or importing quantities from 2D CAD drawings or independent modeling software into a bill of quantities for pricing calculations. When design changes occur, manual comparison of the drawings is required to remeasure and adjust the affected items in the bill of quantities.

[0003] Existing technologies, when handling changes, rely on manual identification of drawing differences, resulting in delayed change response. Furthermore, when faced with modifications to complex spatial structures, it is difficult to accurately track cascading cost changes caused by alterations in geometric topology, which can easily lead to omissions or duplications in quantity calculations. It is also impossible to maintain the accuracy and consistency of cost data in real time in a multi-disciplinary collaborative environment, resulting in a lack of dynamic feedback mechanisms for project cost control and severely weakening the support of cost adjustment results for engineering decisions. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies by proposing a method and system for adjusting engineering costs based on BIM data.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a method for adjusting engineering costs based on BIM data, comprising the following steps: S1: Read BIM model files through a dedicated data interface, process building components using entity parsing algorithms, extract three-dimensional geometric coordinates and physical material parameters, parse the connection status between building components, extract component adjacency relationships and hierarchical dependency attributes, construct engineering entity topology network, and define primitive nodes. S2: Based on the standard engineering quantity calculation specifications, semantically associate the graphic element nodes with the cost quota database, generate a logical mapping index, execute the full statistical program, calculate the engineering quantity of each part, compare and filter the resource consumption standards, match the consumption quota, summarize various cost indicators, and generate a benchmark cost list. S3: In response to the model version update signal, compare the model data through Boolean difference operation to generate a change difference matrix, perform deep traversal analysis based on the engineering entity topology network, identify the affected nodes, aggregate the core changed components and derived components, and construct a cascaded affected component set; S4: Perform dynamic recalculation of the quantities of the change difference matrix and the cascaded affected component set, obtain the dynamic recalculation results, connect to the real-time material price data stream, calculate and update the comprehensive unit price value, use the comprehensive unit price value to implement incremental adjustment of the benchmark cost list, and output a cost adjustment report.

[0006] As a further aspect of the present invention, step S1 specifically comprises: S11: Obtain the BIM model file and start the entity parsing algorithm. Traverse all building components to perform multi-dimensional structural feature scanning operations. Use spatial vector matrix transformation logic to extract coordinates of spatial distribution dimensions and generate the three-dimensional geometric coordinates. Simultaneously use the material mapping engine to analyze the surface texture features, internal thermal conductivity and compressive strength index of structural elements. Perform structured cleaning and standardized alignment processing on the spatial coordinates and material information to generate the physical material parameters. S12: For building components, calculate the spatial intersection area and coplanar distance, set an adjacency determination threshold, perform invalid contact surface filtering operation according to the adjacency determination threshold, determine the spatial position compactness and extract the force transmission direction, generate the adjacency relationship of the components, extract the parent-child node nesting structure in the model structure tree, analyze the subordinate level of building components in the load-bearing system and spatial distribution, and generate the hierarchical dependency attribute. S13: Obtain the adjacency relationship of the components and the hierarchical dependency attribute, map the independent building components to the corresponding primitive nodes according to the principle of graph theory, connect the primitive nodes through the edges representing the degree of adjacency and dependency using the directed acyclic graph data structure, perform global topology verification to eliminate isolated unrelated nodes, and construct the engineering entity topology network.

[0007] As a further aspect of the present invention, step S2 specifically comprises: S21: Obtain the graphic element nodes in the engineering entity topology network and the preset cost quota database. Based on the standard engineering quantity calculation specification, extract the category labels and size features for the graphic element nodes. Use the semantic matching engine to search in the quota item system, obtain the quota number and work instruction code, and generate the logical mapping index. S22: Obtain the logical mapping index and load the full statistics program. Gather the volume, area and length data of the same type of primitive nodes one by one in the order of bottom to top of the network hierarchy. Deduct the duplicate engineering quantity caused by the spatial intersection of components. After accuracy verification and unit conversion, calculate the engineering quantity of the sub-items. S23: Obtain the resource consumption standards from the sub-item project quantity and cost quota database, use the quota matching algorithm to compare the project category and construction process requirements one by one, filter out the matching labor, material and mechanical operation consumption indicators, generate the consumption quota, and summarize the total consumption amount and the fee rate of each item to construct the benchmark cost list.

[0008] As a further aspect of the present invention, step S3 specifically comprises: S31: Obtain the model version update signal from external input, obtain historical model data and the latest model data, perform Boolean difference operation on primitive attributes and geometric contours within the same spatial coordinate range for both, extract geometric shape addition and deletion and material replacement status, perform numerical conversion calculation based on the degree of difference, and generate the change difference matrix; S32: Obtain the change difference matrix and the engineering entity topology network, anchor the difference points to specific primitive nodes, start a breadth-first depth traversal analysis along the topology edge with the change node as the starting point, determine whether the connection strength and dependency weight exceed the preset wave propagation threshold, and identify the affected node. S33: Obtain the affected nodes and their associated edge data in the engineering entity topology network, extract the structural force transmission path and spatial conflict avoidance conditions, screen out the derivative components that must be simultaneously modified in design and adjusted in engineering quantity, aggregate the core changed components and derivative components at the system level, and establish the cascaded affected component set.

[0009] As a further aspect of the present invention, step S4 specifically comprises: S41: Obtain the change difference matrix and the cascaded affected component set, re-extract the geometric parameters for the components that have undergone shape modification or material replacement, use the local update algorithm to replace the global recalculation, calculate the net engineering quantity changes of the increment, reduction and remodeling parts, and obtain the dynamic recalculation result; S42: Obtain real-time material price data stream and historically stored price archives, use time series smoothing technology to perform smoothing calculations on discrete price data, calculate the current market average price and procurement channel weighted cost for core building materials, replace the original static quota unit price, and generate the comprehensive unit price value; S43: Obtain the dynamic recalculation result, the comprehensive unit price value, and the benchmark cost list; perform matrix multiplication on the changed quantity of work and the updated unit price to determine the increase or decrease in cost; add the increase or decrease to the benchmark cost item to implement incremental adjustment; and compile the cost adjustment report.

[0010] As a further aspect of the present invention, the process of obtaining the three-dimensional geometric coordinates specifically includes: For the surface of building components, obtain the coordinates of discrete sampling points, calculate the spatial curvature change gradient and normal vector deflection angle between adjacent sampling points, and dynamically adjust the sampling density according to the curvature change to preserve edge feature details; For the coordinates of discrete sampling points, the distribution density and noise interference weights are obtained. A Gaussian mixture model is used to perform nonlinear regression fitting on the original coordinate data. Spatially isolated outliers are removed based on the fitting residuals, and the core skeleton information of the main components is retained to obtain the core skeleton information. Obtain the core skeleton information using a coordinate transformation matrix: ; Perform translation and rotation alignment of the global coordinate system, and output the three-dimensional geometric coordinates; in, This represents the transformed standard three-dimensional coordinate vector. Represents the spatial rotation transformation matrix. This represents the original local coordinate vector before the transformation. This represents the translation offset vector.

[0011] As a further aspect of the present invention, the process of establishing the logical mapping index specifically includes: For the primitive nodes, category labels, material identifiers and spatial size features are obtained. The applicable scope and process feature text of the quota items are extracted from the cost quota database. Natural language processing technology is used to perform part-of-speech tagging on the process feature text to obtain word segmentation results. Obtain the word segmentation results and the category labels, calculate the cosine similarity between the two in the semantic vector space and set a numerical judgment threshold, filter out candidate quota items with cosine similarity higher than the numerical judgment threshold, analyze the labor consumption weight and mechanical construction operation coefficient of the candidate quota items, and obtain the labor consumption weight and mechanical construction operation coefficient. Obtain the labor consumption weight, the mechanical construction operation coefficient, and the category label; establish the association between candidate quota items and graphic element nodes; use the conflict resolution mechanism to eliminate mutually exclusive quota association items; and generate the logical mapping index.

[0012] As a further aspect of the present invention, the process of generating the change difference matrix specifically includes: Obtain historical model data and the latest model data, extract the overlapping areas of their spatial locations to obtain voxel mesh units, compare the material status and existence markers of voxel mesh units under the same spatial coordinates to determine whether there is an incremental reduction of geometric volume or a complete replacement of material properties, and identify the changed voxel mesh. The changed voxel mesh is obtained, and the volume change, surface area change ratio and material unit price difference coefficient are calculated. Various change indicators are converted into dimensionless characteristic values ​​with unified dimensions to obtain dimensionless characteristic values. The dimensionless feature values ​​are obtained, and the voxel-level difference data are aggregated and summarized upwards according to the component affiliation relationship. A multidimensional matrix structure is constructed with the component unique identifier code as the row index and the change index type as the column index, and the change difference matrix is ​​generated.

[0013] As a further aspect of the present invention, the process of generating the comprehensive unit price specifically includes: Based on the records of real-time material price data streams over a past observation window, discrete price data is obtained, and the variance of price fluctuations and the distribution of purchasing frequency for various materials at different time points are calculated. Based on the degree of fluctuation, smoothing weights are dynamically allocated to obtain the average price and market share. Obtain the average price and the market share ratio, and then use the weighted fusion formula: ; The base price of the material is calculated; in, This represents the weighted average benchmark price of the materials. Representing the Market share of each supplier Representing the The average price offered by each supplier This represents a price risk compensation coefficient set in conjunction with macroeconomic inflation expectations. Represents the total number of suppliers; Obtain the benchmark price of the material, include it in the preset labor cost and machinery depreciation and amortization expenses for construction operations, and perform additional calculations according to the rate standards promulgated by the cost management department of the project location to generate the comprehensive unit price value.

[0014] A BIM data-driven engineering cost adjustment system, the system being used to implement the aforementioned BIM data-driven engineering cost adjustment method, the system comprising: The model parsing and topology construction module is used to read BIM model files, process building components using entity parsing algorithms, obtain three-dimensional geometric coordinates and physical material parameters, parse the connection status between building components, obtain the adjacency relationship and hierarchical dependency attributes of components, construct the engineering entity topology network, and define primitive nodes. The mapping matching and bill of quantities generation module is used to semantically associate graphic element nodes with the cost quota database according to the standard engineering quantity calculation specifications, generate logical mapping indexes, execute the full quantity statistics program to calculate the engineering quantity of sub-items, match consumption quotas, and generate benchmark cost bill of quantities. The version comparison and impact identification module is used to respond to the model version update signal, compare the model data through Boolean difference operation, generate a change difference matrix, perform deep traversal analysis based on the engineering entity topology network, identify the affected nodes, and lock the cascaded affected component set; The dynamic recalculation and cost adjustment module is used to perform dynamic recalculation of the quantities of the project for the change difference matrix and the cascaded affected component set, obtain the dynamic recalculation results, connect to the real-time material price data stream, update the comprehensive unit price value, use the comprehensive unit price value to implement incremental adjustment of the benchmark cost list, and output a cost adjustment report.

[0015] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, by analyzing the geometric attributes and topological relationships of the BIM model, a dynamic mapping map of component entities and cost quotas is constructed. When design changes are detected, differentiated calculation logic is automatically triggered to accurately locate and update the engineering quantity data affected by cascading changes. At the same time, historical data is used to correct the cost prediction model, realizing the automation and real-time nature of cost adjustment. This effectively solves the problems of response lag and calculation omissions in traditional manual calculation, ensures data consistency in a multi-disciplinary collaborative environment, and significantly improves the accuracy of engineering cost control and the dynamic adaptability of the whole process management. Attached Figure Description

[0016] Figure 1 This is a flowchart of a BIM data-driven engineering cost adjustment method according to the present invention; Figure 2 A flowchart for constructing the engineering entity topology network for this invention; Figure 3 A flowchart for generating a benchmark cost list for this invention; Figure 4 A flowchart for constructing the cascaded affected component set for this invention; Figure 5 This is a flowchart of the dynamic recalculation of engineering quantities and cost adjustment for this invention. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of this invention clearer, the software-based technical solution is described in detail below with reference to system architecture diagrams and embodiments. It should be understood that the specific embodiments described herein are only for explaining the technical solutions of this invention and do not constitute a limitation on the scope of protection.

[0018] In the description of this invention, the system architecture relationships or data processing flows indicated by terms such as "layer," "module," "interface," "data flow," "client," and "server" are all defined based on the architecture diagram or flowchart corresponding to the embodiments. This way of describing is only used to clearly illustrate the logical relationships between the elements in the technical solution, and not to limit the physical deployment form. The term "multiple" includes two or more technical units, including but not limited to multiple data nodes, processing threads, service instances, or functional components and other scalable elements. The specific number is determined according to the actual business scenario and needs to be specifically specified.

[0019] Please see Figure 1 and Figure 2 This invention provides a technical solution: a method for adjusting engineering costs based on BIM data, comprising the following steps: S1: Read BIM model files through a dedicated data interface, process building components using entity parsing algorithms, extract three-dimensional geometric coordinates and physical material parameters, parse the connection status between building components, extract component adjacency relationships and hierarchical dependency attributes, construct engineering entity topology network, and define primitive nodes. The specific steps of S1 are as follows: S11: Obtain BIM model files and start entity parsing algorithm, traverse all building components to perform multi-dimensional structural feature scanning operation, use spatial vector matrix transformation logic to extract coordinates of spatial distribution dimensions, generate three-dimensional geometric coordinates, and simultaneously use material mapping engine to parse surface texture features, internal thermal conductivity and compressive strength index of structural elements, perform structured cleaning and standardized alignment of spatial coordinates and material information, and generate physical material parameters. Specifically, the entity parsing algorithm is a systematic method for extracting and structuring component information from BIM model files. This algorithm takes all building components in the BIM model as its processing object and transforms the original model data into standardized primitive node data suitable for subsequent engineering cost calculations through three core steps: geometric feature parsing, semantic attribute extraction, and topological relationship construction. The algorithm first reads the BIM model file through a dedicated data interface to obtain the model's underlying data structure. For IFC format files, it traverses the IfcProduct class instances to identify all building component entities and uses entity inheritance relationship parsing technology to establish a three-level classification system including geometric attributes, semantic parameters, and relationship topology. Simultaneously, it performs structured cleaning and standardized alignment processing on noise and redundant information in the original data to ensure the accuracy of subsequent parsing. In the geometric feature analysis stage, the algorithm performs refined analysis of the 3D geometric contour of each building component. It extracts the boundary representation data of the component from the geometric expression chain of the IFC file, identifies the corresponding geometric description method for different types of components, extracts the cross-sectional contour and tensile direction parameters for stretched components, and obtains surface point cloud data through discrete sampling for complex curved surface components, calculating the spatial curvature gradient and normal vector deflection angle. Based on this, the sampling density is dynamically adjusted to preserve edge feature details. Subsequently, a Gaussian mixture model is used to perform nonlinear regression fitting on the sampling points to eliminate spatially isolated outliers. Finally, a rigid body transformation from the local coordinate system to the global coordinate system is achieved through spatial vector matrix transformation, outputting standardized 3D geometric coordinates. In the semantic attribute analysis dimension, the algorithm reads the component's category label, material identifier, and physical performance parameters (such as thermal conductivity and compressive strength) from the IFC model. It analyzes the surface texture features and internal physical indicators of structural elements through a material mapping engine, and associates the material code with material entries in the standard quota database. Simultaneously, it extracts key measurement parameters such as the component's spatial dimensions, volume, and surface area, providing a data foundation for subsequent quantity calculations. In terms of topological relationship construction, the algorithm filters out invalid contact surfaces and identifies valid physical connection relationships by calculating the spatial intersection area and coplanar distance between components and setting an adjacency determination threshold. At the same time, it extracts the parent-child node nesting structure in the model structure tree to analyze the subordinate hierarchy of components in the load-bearing system and spatial distribution. Based on this, independent components are mapped to graph primitive nodes according to graph theory principles. The nodes are connected by edges representing the degree of adjacency and dependency using a directed acyclic graph data structure, and finally a complete engineering entity topological network is constructed.The standardized data output by the algorithm includes a unique identifier for each element node, three-dimensional geometric coordinates, physical material parameters, component adjacency relationships, and hierarchical dependency attributes. This data is organized in a structured format for subsequent modules such as semantic matching, quantity calculation, and cost adjustment. This achieves automatic conversion from the original BIM model to structured engineering data, providing complete and accurate element-level data support for engineering cost management.

[0020] The specific implementation mechanism of the material mapping engine is as follows: First, the engine parses the material definition associated with each building component in the IFC file. Based on the IfcMaterial and IfcMaterialProperties entity specifications, it extracts the material name identifier associated with the component (e.g., "C30 concrete"). For composite components, it further parses the material layer definitions and their thickness proportions in the IfcMaterialLayerSet. The engine has a built-in standard parameter mapping library with material name as the primary key. It establishes a primary index by material category (concrete, steel, etc.) and a secondary index by strength grade or specification. It pre-stores benchmark values ​​for physical parameters such as density, thermal conductivity (W / (m·K)), and compressive strength (MPa). Data sources cover national building material standards and specifications and industry design manuals. During mapping, the engine first performs a coarse match in the primary index by material category, and then precisely retrieves the corresponding parameters in the secondary index based on specification information (e.g., strength grade C30). For unconventional materials not pre-set in the mapping library, a fuzzy matching mechanism is activated. The optimal candidate is selected by calculating the edit distance and keyword overlap between the material name and the library entry, and a matching confidence flag is output. After matching is complete, the engine converts thermal conductivity to W / (m·K) and compressive strength to MPa. For composite components, it calculates equivalent physical parameters based on the volume ratio of each material layer. Finally, it structurally associates the standardized physical parameters with the three-dimensional geometric coordinates according to the component's unique identifier, providing a reliable data foundation for topology network construction and cost calculation. This mechanism achieves high-precision automated conversion from abstract BIM material identifiers to specific physical parameters, avoiding the inefficiency and errors of manual table lookups.

[0021] S12: For building components, calculate the spatial intersection area and coplanar distance, set the adjacency determination threshold, perform the filtering operation of invalid contact surfaces according to the adjacency determination threshold, determine the spatial position compactness and extract the force transmission direction, generate the component adjacency relationship, and at the same time extract the parent-child node nesting structure in the model structure tree, analyze the subordinate level of building components in the load-bearing system and spatial distribution, and generate hierarchical dependency attributes. S13: Obtain the adjacency relationship and hierarchical dependency attribute of the components, map the independent building components to the corresponding primitive nodes according to the principle of graph theory, connect the primitive nodes through the edges representing the degree of adjacency and dependency using the directed acyclic graph data structure, perform global topology verification to eliminate isolated unrelated nodes, and construct the engineering entity topology network. The process of obtaining three-dimensional geometric coordinates specifically includes: For the surface of building components, obtain the coordinates of discrete sampling points, calculate the spatial curvature change gradient and normal vector deflection angle between adjacent sampling points, and dynamically adjust the sampling density according to the curvature change to preserve edge feature details; For the coordinates of discrete sampling points, the distribution density and noise interference weights are obtained. A Gaussian mixture model is used to perform nonlinear regression fitting on the original coordinate data. Spatially isolated outliers are removed based on the fitting residuals, and the core skeleton information of the main components is retained to obtain the core skeleton information. Obtain core skeleton information using coordinate transformation matrix: ; Perform translation and rotation alignment of the global coordinate system to output three-dimensional geometric coordinates; in, This represents the transformed standard three-dimensional coordinate vector. Represents the spatial rotation transformation matrix. This represents the original local coordinate vector before the transformation. This represents the translation offset vector.

[0022] The system reads BIM model files in industrial basic format and extracts entity data for 15,230 building components. For each extracted component, a multi-dimensional structural feature scan is performed, iterating through and reading the underlying parameter list of each component to extract the specific identifiers of structural columns, load-bearing walls, and floor slabs. Using spatial vector matrix transformation logic, the system extracts the 3D bounding box vertex coordinates of each component in the local coordinate system, multiplying the X, Y, and Z axis vectors of the local coordinate system by a quaternion rotation matrix and a global translation vector, respectively. During the extraction of 3D geometric coordinates, discrete sampling points are placed on the surface of the building components at initial intervals of 20 mm, resulting in a total of 350,000 discrete sampling point coordinate data. The spatial curvature gradient and normal vector deflection angle between adjacent sampling points are calculated. Eight adjacent sampling points are selected to form a local neighborhood, and the eigenvalues ​​of the covariance matrix of the point set within this neighborhood are calculated. The ratio of the minimum eigenvalue to the sum of all eigenvalues ​​is defined as the spatial curvature gradient. Calculate the dot product between the normal vector of the current sampling point and the normal vector of the center point in the neighborhood, and use the inverse cosine of the dot product as the normal vector deflection angle. Regions with a spatial curvature gradient greater than 0.15 and a normal vector deflection angle greater than 15 degrees are identified as edge feature regions. The sampling density in these regions is increased to a spacing of 5 mm, while the sampling density in flat regions with a curvature gradient less than 0.05 is reduced to a spacing of 50 mm. This dynamic adjustment of sampling density preserves edge feature details.

[0023] The distribution density and noise interference weights were obtained. The sampling point distribution density was defined as the number of sampling points contained within a 100 mm radius unit sphere. The noise interference weight was set as the product of the inverse of the distribution density and the laser scanner hardware error calibration value of 0.02. A Gaussian mixture model was used to perform nonlinear regression fitting on the original coordinate data. The number of Gaussian distribution components was set to 5. The mean vector, covariance matrix, and mixture weight parameters of each component were initialized, and the expectation-maximization algorithm was used for iteration. In the expectation step, the posterior probability of each sampling point belonging to each Gaussian component was calculated. In the maximization step, the mean, covariance, and mixture weights were updated using the posterior probability. The convergence threshold was set to a log-likelihood function change of less than 0.001, and the maximum number of iterations was limited to 100. After 85 iterations, the model converged. The Mahalanobis distance from each discrete sampling point to the fitted Gaussian surface was calculated as the fitting residual. Coordinate points with fitting residuals greater than 3 standard deviations were identified as spatially isolated outliers and removed, while retaining the core skeletal information of the main components.

[0024] Using coordinate transformation matrix: ; The global coordinate system is aligned through translation and rotation. The spatial rotation transformation matrix is ​​calculated using three Euler angles. Since the X-axis rotation angle is 0 degrees, the Y-axis rotation angle is 90 degrees, and the Z-axis rotation angle is 0 degrees, the rotation matrix possesses orthogonal properties. The original local coordinate vectors are provided by the denoised core skeleton point set described above. The translation offset vector is set to the geodetic coordinates of the project reference point, with an X-axis offset of 5000 mm, a Y-axis offset of 8000 mm, and a Z-axis offset of 0 mm. Substituting these parameters into the formula, the original local coordinate vectors are multiplied by the spatial rotation transformation matrix, and then the translation offset vector is added to complete the rigid body transformation of the coordinates. The three-dimensional geometric coordinates are output. The advantage of this formula is that by introducing Gaussian mixture model denoising and adaptive sampling density combined with spatial vector matrix transformation, the interference of local burr data on the surface of irregular components in the BIM model is reduced, keeping the accuracy error of the coordinate transformation within 2 mm. Experimental data show that when the maximum number of iterations is set to 100 and the residual elimination standard is set to 3 standard deviations, the accuracy of core skeleton information extraction reaches 98.5%, which is 12% higher than that of the traditional principal component analysis method. This result indicates that the purity of the geometric coordinate data meets the standard for subsequent engineering quantity calculations.

[0025] in, This represents the transformed standard three-dimensional coordinate vector. Represents the spatial rotation transformation matrix. This represents the original local coordinate vector before the transformation. Represents the translation offset vector; The surface texture features, internal thermal conductivity, and compressive strength of structural elements are analyzed, and the material texture file path, RGB color values, and roughness parameters are read. The internal thermal conductivity is read and structured into a watt-per-meter Kelvin value, and the compressive strength is extracted and standardized to a megapascal value. The spatial coordinates and material information are structured, cleaned, and standardized for alignment, and null and duplicate values ​​are removed to generate physical material parameters.

[0026] For building components, the spatial intersection volume of the 3D bounding box of the components is calculated using the separating axis theorem, and the contact surfaces are extracted and the spatial intersection area is calculated. The coplanar distance is obtained by calculating the minimum Euclidean distance between the adjacent polygons of two components. Adjacency determination thresholds are set; based on statistical analysis of 100 historical projects, the lower limit for the intersection area is set to 0.05 square meters, and the upper limit for the coplanar distance is set to 10 millimeters. When the calculated results between components do not simultaneously satisfy the conditions of an intersection area greater than 0.05 square meters and a coplanar distance less than 10 millimeters, an invalid contact surface filtering operation is performed, classifying it as a non-structural contact. The ratio of the intersection polygon area to the minimum cross-sectional area of ​​the component is calculated to determine the spatial compactness; when the ratio exceeds 30%, it is considered a strong and tight connection. The force transmission direction is extracted, obtaining the gravity vector and bearing capacity line of action of the component, and determining the load transfer path from the floor slab to the secondary beam, the secondary beam to the main beam, and the main beam to the load-bearing column. Component adjacency relationships are then generated. Extract the nested parent-child node structure from the model's tree structure, parse the aggregated and related entity tags in the IFC file, and analyze the hierarchical relationship of building components in the load-bearing system and spatial distribution. Generate hierarchical dependency attributes.

[0027] The adjacency relationships and hierarchical dependencies of building components are obtained, and 15,230 independent building components are mapped to corresponding 15,230 primitive nodes based on graph theory principles. A directed acyclic graph (DAG) data structure is used to connect these primitive nodes through edges representing adjacency and dependency, assigning each edge a corresponding connection strength weight. Global topology verification is performed by traversing all nodes using a depth-first search algorithm, checking for nodes with a sum of in-degree and out-degree of 0. Twenty-five isolated and unrelated nodes are removed, ultimately constructing an engineering entity topology network containing 15,205 nodes.

[0028] Table 1. Characteristics of Curvature Changes at Sampling Points and Filtering Data

[0029] Table 1 lists the specific values ​​for the local curvature gradient, normal vector deflection angle, and Mahalanobis distance of the sampling points, demonstrating the quantitative filtering standards for the core skeleton information extraction stage.

[0030] The aforementioned watts per meter Kelvin refers to the international unit for measuring the thermal conductivity of a material; The aforementioned megapascal (MPa) refers to a pressure unit that characterizes the compressive strength of a material. The aforementioned separating axis theorem refers to a collision detection algorithm that determines whether two convex polygons intersect by judging whether their projections on any axis overlap.

[0031] Please see Figure 1 and Figure 3S2: Based on the standard engineering quantity calculation specifications, semantically associate the graphic element nodes with the cost quota database, generate a logical mapping index, execute the full statistical program, calculate the engineering quantity of each part, compare and filter the resource consumption standards, match the consumption quota, summarize various cost indicators, and generate a benchmark cost list. The specific steps of S2 are as follows: S21: Obtain the graphic element nodes in the engineering entity topology network and the preset cost quota database. Based on the standard engineering quantity calculation specifications, extract the category labels and size features of the graphic element nodes, use the semantic matching engine to search in the quota item system, obtain the quota number and work instruction code, and generate a logical mapping index. S22: Obtain the logical mapping index and load the full statistics program. Gather the volume, area and length data of the same type of primitive nodes one by one in the order of bottom to top of the network hierarchy. Deduct the duplicate engineering quantity caused by the spatial intersection of components. After accuracy verification and unit conversion, calculate the engineering quantity of each component. S23: Obtain the resource consumption standards from the database of sub-item engineering quantities and cost quotas, use the quota matching algorithm to compare the engineering categories and construction process requirements one by one, filter out the matching labor, material and mechanical operation consumption indicators, generate consumption quotas, and summarize the total consumption amount and various cost rates to construct a benchmark cost list. The process of creating a logical mapping index specifically includes: For the primitive nodes, category labels, material identifiers and spatial size features are obtained. The applicable scope and process feature text of the quota items are extracted from the cost quota database. Natural language processing technology is used to perform part-of-speech tagging on the process feature text to obtain word segmentation results. Obtain word segmentation results and category labels, calculate the cosine similarity between the two in the semantic vector space and set a numerical judgment threshold, filter out candidate quota items with cosine similarity higher than the numerical judgment threshold, analyze the labor consumption weight and mechanical construction operation coefficient of the candidate quota items, and obtain the labor consumption weight and mechanical construction operation coefficient. Obtain the labor consumption weight, mechanical construction operation coefficient and category label, establish the association between candidate quota items and graphic element nodes, use the conflict resolution mechanism to eliminate mutually exclusive quota association items, and generate a logical mapping index.

[0032] Based on standard engineering quantity calculation specifications, category labels and dimensional features of 15,205 graphical element nodes were extracted from the engineering entity topology network. For category labels, the corresponding structural name text was read, such as "reinforced concrete rectangular column". For dimensional features, the corresponding three-dimensional dimensions of length, width, and height were extracted, for example, length 400 mm, width 400 mm, and height 3000 mm. The process of establishing a logical mapping index specifically includes: obtaining category labels, material identifiers, and spatial dimensional features; extracting the applicable scope and process feature text of quota entries from a cost quota database containing 50,000 standard data entries; using natural language processing technology to perform part-of-speech tagging on the process feature text; and using a hidden Markov model to perform part-of-speech tagging on the text sequence, segmenting the text into noun, verb, and adjective combinations to obtain the word segmentation results.

[0033] Obtain word segmentation results and category labels, and use the term frequency-inverse document frequency algorithm to map the word segmentation results and category labels into 256-dimensional semantic vectors. Calculate the cosine similarity between the two in the semantic vector space, using the vector inner product divided by the product of their respective vector magnitudes. Set a numerical judgment threshold; based on the optimization results of the semantic matching test set, the numerical judgment threshold is set to 0.85. Compare the calculated cosine similarity and filter out candidate quota items with a cosine similarity higher than 0.85. For the selected candidate quota items, analyze the labor consumption weight and mechanical construction operation coefficient, and read the base price composition ratio from the quota item's backend. Obtain the labor consumption weight, mechanical construction operation coefficient, and category label. If the category label is "cast-in-place concrete beam," and the labor consumption weight is 0.35 and the mechanical construction operation coefficient is 1.2, then establish the association between the candidate quota item and the primitive node. A conflict resolution mechanism is used to eliminate mutually exclusive quota association items. When the same graphic element node is associated with both "precast concrete beam" and "cast-in-place concrete beam" quotas, the on-site casting marker in the graphic element node attributes is checked. On-site casting association items are retained, while pre-defined quota association items are eliminated, generating a logical mapping index. This mapping index provides a unique quota number and work guidance coding benchmark for subsequent quantity statistics and price matching. Experimental data shows that when the numerical judgment threshold is set to 0.85, the semantic matching accuracy reaches 96.8%, which is 25% higher than the accuracy of the direct keyword matching method.

[0034] The full statistical program is loaded and traverses from the basic components to the floor slabs and roof components in a bottom-up order according to the network hierarchy. Volume, area, and length data of similar element nodes are aggregated one by one, and the geometric data of element nodes mapped to the same quota number are accumulated. The duplicate quantities caused by spatial intersections of components are deducted, and the spatial intersection volumes calculated in the previous steps are retrieved. For the intersections, the deduction rules of the standard engineering quantity calculation specifications are applied. For example, when a main beam and a secondary beam intersect, the volume of the secondary beam in the intersection area is deducted by 1.5 cubic meters, retaining the complete volume of the main beam. After accuracy verification and unit conversion, the volume data in cubic millimeters is divided by 1,000,000,000 to convert to cubic meters, and three decimal places are retained to calculate the quantities of each component.

[0035] The process involves obtaining resource consumption standards from the database of project quantities and cost quotas, and then comparing each item against the project category and construction process requirements using a quota matching algorithm. For each project quantity, the corresponding labor man-day consumption, material unit consumption, and machinery shift consumption are retrieved from the resource consumption standard table. Matching labor, material, and machinery operation consumption indicators are selected to generate consumption quotas. The project quantity is multiplied by the corresponding indicator value of the consumption quota, and the total consumption amount is calculated. Based on the pricing standards of the project location, an 8% enterprise management fee rate, a 5% profit margin, and an 11% regulatory fee and tax rate are extracted. These rates are then added together and summarized to construct a benchmark cost list.

[0036] Table 2. Comparison of semantic matching of cost quotas

[0037] See Table 2, which details the labor consumption weights and mechanical construction operation coefficients for different quota items, and shows the calculated cosine similarity values, demonstrating the quantitative basis for using 0.85 as the numerical judgment threshold to screen valid quota items.

[0038] Please see Figure 1 and Figure 4 S3: Responding to the model version update signal, compare the model data through Boolean difference operation to generate a change difference matrix, perform deep traversal analysis based on the engineering entity topology network, identify the affected nodes, aggregate the core changed components and derived components, and construct a cascaded affected component set; The specific steps for S3 are as follows: S31: Obtain the model version update signal from external input, obtain historical model data and the latest model data, perform Boolean difference operation on primitive attributes and geometric contours within the same spatial coordinate range for both, extract geometric shape addition and deletion and material replacement status, perform numerical conversion calculation based on the degree of difference, and generate change difference matrix; S32: Obtain the change difference matrix and the engineering entity topology network, anchor the difference points to specific primitive nodes, start the breadth-first depth traversal analysis along the topology edge with the changed node as the starting point, determine whether the connection strength and dependency weight exceed the preset wave propagation threshold, and identify the affected nodes. S33: Obtain the affected nodes and their associated edge data in the engineering entity topology network, extract the structural force transmission path and spatial conflict avoidance conditions, screen out the derivative components that must be simultaneously modified in design and adjusted in engineering quantity, aggregate the core changed components and derivative components at the system level, and establish a cascaded set of affected components. The process of generating the change difference matrix specifically includes: Obtain historical model data and the latest model data, extract the overlapping areas of their spatial locations to obtain voxel mesh units, compare the material status and existence markers of voxel mesh units under the same spatial coordinates to determine whether there is an incremental reduction of geometric volume or a complete replacement of material properties, and identify the changed voxel mesh. Obtain the changed voxel mesh, calculate the volume change, surface area change ratio, and material unit price difference coefficient, and convert various change indicators into dimensionless characteristic values ​​with unified dimensions to obtain dimensionless characteristic values. Obtain dimensionless feature values, aggregate the voxel-level difference data upwards according to the component affiliation, construct a multidimensional matrix structure with the component unique identifier code as the row index and the change index type as the column index, and generate the change difference matrix.

[0039] In response to the model version update signal, the system acquires the externally input model version update signal, which includes an update timestamp and a unique identifier for the update package. It retrieves historical and latest model data, reading the pre-update version 1 dataset and the updated version 2 dataset. The process of generating the change difference matrix specifically includes: extracting the overlapping spatial regions of the two datasets; dividing the 3D space into a 3D orthogonal mesh with a fixed side length of 500 mm; and obtaining voxel mesh elements containing model entities. It compares the material status and existence markers of voxel mesh elements at the same spatial coordinates, extracting the voxel element at coordinates X=1000, Y=2000, Z=3000 in version 1, reading its material code as C30 concrete, and its existence marker as 1. Simultaneously, it extracts voxel elements at the same coordinates in version 2, reading their material code as changed to C40 concrete, and their existence marker as 1. It then determines whether there is an incremental reduction in geometric volume or a complete replacement of material properties. If the existence marker changes from 1 to 0, it is determined to be a volume reduction; if the existence marker changes from 0 to 1, it is determined to be a volume increase; if the material code changes, it is determined to be a material replacement. A total of 1850 voxel grids with changed states were identified.

[0040] To calculate the volume change, multiply the number of newly added voxels by the volume of a single voxel (0.125 cubic meters) and subtract the number of deleted voxels multiplied by the volume of a single voxel. To calculate the surface area change ratio, extract the number of faces of the changed voxels exposed to air and calculate their area, then divide by the total surface area of ​​the original component. To calculate the material unit price difference coefficient, subtract the old material unit price of 400 yuan per cubic meter from the new material unit price of 500 yuan per cubic meter, then divide by the old material unit price to obtain 0.25. Transform all change indicators into dimensionless feature values ​​with a unified dimension. Using the range standardization method, subtract the minimum sample value from each indicator and divide by the sample range to map the data to the interval between 0 and 1. Obtain the dimensionless feature values. Aggregate the voxel-level difference data upwards according to the component affiliation, and calculate the average of all changed voxel grid indicators belonging to the same unique identifier. Construct a multidimensional matrix structure with the component unique identifier as the row index and the change indicator type as the column index, generating a change difference matrix containing 150 rows and 4 columns.

[0041] The differences are anchored to specific primitive nodes. By matching the row index of the change difference matrix with the unique identifier of each primitive node, the 150 core primitive nodes that have undergone changes are identified. Starting from the changed nodes, a breadth-first depth-first traversal analysis is initiated along the topology edges to read the connection strength weight of each edge in the network. It is then determined whether the connection strength and dependency weight exceed a preset sweep propagation threshold. Based on 50 simulated propagation tests, the sweep propagation threshold is set to 0.65. When the connection strength between nodes is 0.8, which is greater than the sweep propagation threshold of 0.65, it is identified as a affected node.

[0042] Data on affected nodes and their associated edges in the engineering entity topology network were acquired, and attribute parameters of the affected nodes were extracted. The structural force transmission path and spatial conflict avoidance conditions were extracted, and it was analyzed whether the affected nodes were located on the direct load-bearing path of the core modified component and whether there were physical collisions involving space occupation. For example, when the cross-sectional size of the core modified component increased from 400 mm to 500 mm, a 50 mm spatial overlap conflict occurred at the location of the pre-reserved opening for the adjacent pipe. Derivative components that required simultaneous design modifications and quantity adjustments were identified. The core modified component and the affected derivative components were aggregated at the system level, duplicate records were removed, and a cascaded affected component set containing 320 components was finally established. Experimental data showed that when the impact transmission threshold was set to 0.65, the accuracy of identifying the scope of change impact reached 95%, reducing the false negative rate by 18% compared to manual screening methods.

[0043] The range standardization method mentioned above refers to a data preprocessing algorithm that scales all values ​​to between 0 and 1 by subtracting the minimum value from a set of data values ​​and dividing by the difference between the maximum and minimum values ​​in the set.

[0044] Please see Figure 1 and Figure 5 S4: Perform dynamic recalculation of quantities for the change difference matrix and the cascaded affected component set, obtain the dynamic recalculation results, connect to the real-time material price data stream, calculate and update the comprehensive unit price value, use the comprehensive unit price value to implement incremental adjustment of the benchmark cost list, and output the cost adjustment report; The specific steps for S4 are as follows: S41: Obtain the change difference matrix and the cascaded affected component set, re-extract the geometric parameters for components that have undergone shape modification or material replacement, use the local update algorithm to replace the global recalculation, calculate the net engineering quantity changes of the increment, reduction and remodeling parts, and obtain the dynamic recalculation results; S42: Obtain real-time material price data streams and historically stored price archives, use time series smoothing technology to perform smoothing operations on discrete price data, calculate the current market average price and procurement channel weighted cost for core building materials, replace the original static quota unit price, and generate a comprehensive unit price value; S43: Obtain dynamic recalculation results, comprehensive unit price values, and benchmark cost list; perform matrix multiplication of the changed quantities and updated unit prices to determine the increase or decrease in cost; add the increase or decrease to the benchmark cost items to implement incremental adjustments; and compile a cost adjustment report. The process of generating the composite unit price value specifically includes: Based on the records of real-time material price data streams over a past observation window, discrete price data is obtained, and the variance of price fluctuations and the distribution of purchasing frequency for various materials at different time points are calculated. Based on the degree of fluctuation, smoothing weights are dynamically allocated to obtain the average price and market share. To obtain the average price and market share, use the weighted fusion formula: ; The base price of the material is calculated; in, This represents the weighted average benchmark price of the materials. Representing the Market share of each supplier Representing the The average price offered by each supplier This represents a price risk compensation coefficient set in conjunction with macroeconomic inflation expectations. Represents the total number of suppliers; Obtain the benchmark price of materials, include it in the preset labor cost and machinery depreciation and amortization expenses for construction operations, and perform additional calculations according to the rate standards issued by the cost management department of the project location to generate a comprehensive unit price value.

[0045] Obtain the change difference matrix and the cascaded affected component set. For the 320 components that underwent shape modifications or material replacements, re-extract geometric parameters and use a local update algorithm instead of a global recalculation. Only extract component model data within the cascaded affected component set, skipping the 14885 unchanged components. Calculate the net changes in volume for increments, decrements, and remodeled portions. For a concrete column with an enlarged cross-section, calculate the difference between the new cross-sectional area of ​​0.25 square meters and the old cross-sectional area of ​​0.16 square meters, multiply by the column height of 3000 mm, and obtain an increase in volume of 0.27 cubic meters. Summarize the volume, area, and length changes of all affected components to obtain the dynamic recalculation results.

[0046] The process involves acquiring real-time material price data streams and historical price archives, and capturing daily material price data streams published by the local construction cost management station via a network interface. Time series smoothing techniques are used to smooth the discrete price data. The process of generating a comprehensive unit price value specifically includes: obtaining 300 daily discrete price quotes from the real-time material price data stream records within the past 30-day observation window; calculating the price fluctuation variance and purchase frequency distribution for various materials at different time points; calculating the sample variance of C40 concrete prices over 30 days as 15.6; and statistically analyzing the purchase frequency share of five major suppliers. Smoothing weights are dynamically allocated based on the degree of volatility; when the volatility variance exceeds 10, the weight of recent time points is increased to 0.7, and the weight of longer-term time points is decreased to 0.3. The average price and the market share of each supplier are then obtained.

[0047] Using the weighted fusion formula: ; The benchmark price of materials was calculated. A price risk compensation coefficient, based on macroeconomic inflation expectations, was set at 0.03. The total number of suppliers was set to 5. The market share of a certain supplier (0.4) and its average price of 520 yuan were obtained from the previous calculation. The market share of each supplier was multiplied by its average price, and the products of the five suppliers were summed to obtain a benchmark weighted value of 515 yuan. Multiplying 515 yuan by 1 plus the compensation coefficient of 0.03 yielded 530.45 yuan. The advantage of this formula is that by introducing market share weighting and a price risk compensation coefficient, it smooths out abnormal fluctuations in short-term material market prices, maintaining the stability of the unit price. Experimental data shows that the material benchmark price calculated using this weighted fusion formula has a deviation rate of less than 2.5% from the actual purchase settlement price, a reduction of 4.2% compared to the arithmetic mean method. This result indicates that the generated material price possesses market fit and inflation resistance capabilities, providing a price benchmark for cost updates.

[0048] in, This represents the weighted average benchmark price of the materials. This represents the market share of the i-th supplier. This represents the average price offered by the i-th supplier. This represents a price risk compensation coefficient set in conjunction with macroeconomic inflation expectations. Represents the total number of suppliers; The cost is included in the pre-set construction labor cost of 150 yuan per cubic meter and the machinery depreciation and amortization cost of 80 yuan per cubic meter. An additional 8% management fee and a 5% profit margin are extracted according to the fee rate standards issued by the local cost management department. Adding the labor and machinery costs to the base material price of 530.45 yuan, and then multiplying by the fee rate multiplier of 1.13, the final comprehensive unit price is 859.3 yuan.

[0049] Obtain the dynamic recalculation results, comprehensive unit price values, and benchmark cost list. Perform a matrix multiplication operation on the column vector composed of the changed quantities and the row vector composed of the updated unit prices. Multiply the increased volume of 0.27 cubic meters by the corresponding comprehensive unit price of 859.3 yuan, and determine the cost increase to be 232.01 yuan. Summarize the increases and decreases for all affected components, and calculate the total cost increase to be 125,000 yuan. Add this increase or decrease to the benchmark cost item to implement incremental adjustments, compile a cost adjustment report, and output detailed incremental details, unit price adjustment basis, and explanation of the total cost change range.

[0050] Table 3. Price Smoothing and Unit Price Update of Core Building Materials

[0051] See Table 3, which lists the variance of core building materials, the average market price, and the final generated comprehensive unit price, demonstrating the actual calculation results and data distribution of the multi-factor price smoothing fusion algorithm.

[0052] A BIM data-driven engineering cost adjustment system is provided. This system is used to execute the aforementioned BIM data-driven engineering cost adjustment method. The system includes: The model parsing and topology construction module is used to read BIM model files, process building components using entity parsing algorithms, obtain three-dimensional geometric coordinates and physical material parameters, parse the connection status between building components, obtain the adjacency relationship and hierarchical dependency attributes of components, construct the engineering entity topology network, and define primitive nodes. The mapping matching and bill of quantities generation module is used to semantically associate graphic element nodes with the cost quota database according to the standard engineering quantity calculation specifications, generate logical mapping indexes, execute the full quantity statistics program to calculate the engineering quantity of sub-items, match consumption quotas, and generate benchmark cost bill of quantities. The version comparison and impact identification module is used to respond to model version update signals, compare model data through Boolean difference operations, generate a change difference matrix, perform deep traversal analysis based on the engineering entity topology network, identify affected nodes, and lock the cascaded affected component set; The dynamic recalculation and cost adjustment module is used to perform dynamic recalculation of the quantities of work for the change difference matrix and the cascaded affected component set, obtain the dynamic recalculation results, connect to the real-time material price data stream, update the comprehensive unit price value, use the comprehensive unit price value to implement incremental adjustment of the benchmark cost list, and output a cost adjustment report.

[0053] The above embodiments illustrate preferred embodiments of the present invention. Any equivalent adjustments to the technical solution based on software engineering methods are within the scope of protection, including but not limited to: implementing algorithm logic using different programming languages, refactoring functional modules into services, adjusting data interaction protocols, and optimizing resource scheduling strategies. Any implementation scheme derived from reasonable modifications to the data processing flow, service call chain, or system architecture layer without departing from the core technology of the present invention should be considered within the protection scope defined by the technical solution of the present invention.

Claims

1. A method for adjusting engineering costs based on BIM data, characterized in that, Includes the following steps: S1: Read BIM model files through a dedicated data interface, process building components using entity parsing algorithms, extract three-dimensional geometric coordinates and physical material parameters, parse the connection status between building components, extract component adjacency relationships and hierarchical dependency attributes, construct engineering entity topology network, and define primitive nodes. S2: Based on the standard engineering quantity calculation specifications, semantically associate the graphic element nodes with the cost quota database, generate a logical mapping index, execute the full statistical program, calculate the engineering quantity of each part, compare and filter the resource consumption standards, match the consumption quota, summarize various cost indicators, and generate a benchmark cost list. S3: In response to the model version update signal, compare the model data through Boolean difference operation to generate a change difference matrix, perform deep traversal analysis based on the engineering entity topology network, identify the affected nodes, aggregate the core changed components and derived components, and construct a cascaded affected component set; S4: Perform dynamic recalculation of the quantities of the change difference matrix and the cascaded affected component set, obtain the dynamic recalculation results, connect to the real-time material price data stream, calculate and update the comprehensive unit price value, use the comprehensive unit price value to implement incremental adjustment of the benchmark cost list, and output a cost adjustment report.

2. The method for adjusting engineering costs based on BIM data according to claim 1, characterized in that, The specific steps of S1 are as follows: S11: Obtain the BIM model file and start the entity parsing algorithm. Traverse all building components to perform multi-dimensional structural feature scanning operations. Use spatial vector matrix transformation logic to extract coordinates of spatial distribution dimensions and generate the three-dimensional geometric coordinates. Simultaneously use the material mapping engine to analyze the surface texture features, internal thermal conductivity and compressive strength index of structural elements. Perform structured cleaning and standardized alignment processing on the spatial coordinates and material information to generate the physical material parameters. S12: For building components, calculate the spatial intersection area and coplanar distance, set an adjacency determination threshold, perform invalid contact surface filtering operation according to the adjacency determination threshold, determine the spatial position compactness and extract the force transmission direction, generate the adjacency relationship of the components, extract the parent-child node nesting structure in the model structure tree, analyze the subordinate level of building components in the load-bearing system and spatial distribution, and generate the hierarchical dependency attribute. S13: Obtain the adjacency relationship of the components and the hierarchical dependency attribute, map the independent building components to the corresponding primitive nodes according to the principle of graph theory, connect the primitive nodes through the edges representing the degree of adjacency and dependency using the directed acyclic graph data structure, perform global topology verification to eliminate isolated unrelated nodes, and construct the engineering entity topology network.

3. The method for adjusting engineering costs based on BIM data according to claim 1, characterized in that, The specific steps of S2 are as follows: S21: Obtain the graphic element nodes in the engineering entity topology network and the preset cost quota database. Based on the standard engineering quantity calculation specification, extract the category labels and size features for the graphic element nodes. Use the semantic matching engine to search in the quota item system, obtain the quota number and work instruction code, and generate the logical mapping index. S22: Obtain the logical mapping index and load the full statistics program. Gather the volume, area and length data of the same type of primitive nodes one by one in the order of bottom to top of the network hierarchy. Deduct the duplicate engineering quantity caused by the spatial intersection of components. After accuracy verification and unit conversion, calculate the engineering quantity of the sub-items. S23: Obtain the resource consumption standards from the sub-item project quantity and cost quota database, use the quota matching algorithm to compare the project category and construction process requirements one by one, filter out the matching labor, material and mechanical operation consumption indicators, generate the consumption quota, and summarize the total consumption amount and the fee rate of each item to construct the benchmark cost list.

4. The method for adjusting engineering costs based on BIM data according to claim 1, characterized in that, The specific steps in S3 are as follows: S31: Obtain the model version update signal from external input, obtain historical model data and the latest model data, perform Boolean difference operation on primitive attributes and geometric contours within the same spatial coordinate range for both, extract geometric shape addition and deletion and material replacement status, perform numerical conversion calculation based on the degree of difference, and generate the change difference matrix; S32: Obtain the change difference matrix and the engineering entity topology network, anchor the difference points to specific primitive nodes, start a breadth-first depth traversal analysis along the topology edge with the change node as the starting point, determine whether the connection strength and dependency weight exceed the preset wave propagation threshold, and identify the affected node. S33: Obtain the affected nodes and their associated edge data in the engineering entity topology network, extract the structural force transmission path and spatial conflict avoidance conditions, screen out the derivative components that must be simultaneously modified in design and adjusted in engineering quantity, aggregate the core changed components and derivative components at the system level, and establish the cascaded affected component set.

5. The method for adjusting engineering costs based on BIM data according to claim 1, characterized in that, The specific steps of S4 are as follows: S41: Obtain the change difference matrix and the cascaded affected component set, re-extract the geometric parameters for the components that have undergone shape modification or material replacement, use the local update algorithm to replace the global recalculation, calculate the net engineering quantity changes of the increment, reduction and remodeling parts, and obtain the dynamic recalculation result; S42: Obtain real-time material price data stream and historically stored price archives, use time series smoothing technology to perform smoothing calculations on discrete price data, calculate the current market average price and procurement channel weighted cost for core building materials, replace the original static quota unit price, and generate the comprehensive unit price value; S43: Obtain the dynamic recalculation result, the comprehensive unit price value, and the benchmark cost list; perform matrix multiplication on the changed quantity of work and the updated unit price to determine the increase or decrease in cost; add the increase or decrease to the benchmark cost item to implement incremental adjustment; and compile the cost adjustment report.

6. The method for adjusting engineering costs based on BIM data according to claim 2, characterized in that, The process of obtaining the three-dimensional geometric coordinates specifically includes: For the surface of building components, obtain the coordinates of discrete sampling points, calculate the spatial curvature change gradient and normal vector deflection angle between adjacent sampling points, and dynamically adjust the sampling density according to the curvature change to preserve edge feature details; For the coordinates of discrete sampling points, the distribution density and noise interference weights are obtained. A Gaussian mixture model is used to perform nonlinear regression fitting on the original coordinate data. Spatially isolated outliers are removed based on the fitting residuals, and the core skeleton information of the main components is retained to obtain the core skeleton information. Obtain the core skeleton information using a coordinate transformation matrix: ； Perform translation and rotation alignment of the global coordinate system, and output the three-dimensional geometric coordinates; in, This represents the transformed standard three-dimensional coordinate vector. Represents the spatial rotation transformation matrix. This represents the original local coordinate vector before the transformation. This represents the translation offset vector.

7. The method for adjusting engineering costs based on BIM data according to claim 3, characterized in that, The process of establishing the logical mapping index specifically includes: For the primitive nodes, category labels, material identifiers and spatial size features are obtained. The applicable scope and process feature text of the quota items are extracted from the cost quota database. Natural language processing technology is used to perform part-of-speech tagging on the process feature text to obtain word segmentation results. Obtain the word segmentation results and the category labels, calculate the cosine similarity between the two in the semantic vector space and set a numerical judgment threshold, filter out candidate quota items with cosine similarity higher than the numerical judgment threshold, analyze the labor consumption weight and mechanical construction operation coefficient of the candidate quota items, and obtain the labor consumption weight and mechanical construction operation coefficient. Obtain the labor consumption weight, the mechanical construction operation coefficient, and the category label; establish the association between candidate quota items and graphic element nodes; use the conflict resolution mechanism to eliminate mutually exclusive quota association items; and generate the logical mapping index.

8. The method for adjusting engineering costs based on BIM data according to claim 4, characterized in that, The process of generating the change difference matrix specifically includes: Obtain historical model data and the latest model data, extract the overlapping areas of their spatial locations to obtain voxel mesh units, compare the material status and existence markers of voxel mesh units under the same spatial coordinates to determine whether there is an incremental reduction of geometric volume or a complete replacement of material properties, and identify the changed voxel mesh. The changed voxel mesh is obtained, and the volume change, surface area change ratio and material unit price difference coefficient are calculated. Various change indicators are converted into dimensionless characteristic values ​​with unified dimensions to obtain dimensionless characteristic values. The dimensionless feature values ​​are obtained, and the voxel-level difference data are aggregated and summarized upwards according to the component affiliation relationship. A multidimensional matrix structure is constructed with the component unique identifier code as the row index and the change index type as the column index, and the change difference matrix is ​​generated.

9. The method for adjusting engineering costs based on BIM data according to claim 5, characterized in that, The process of generating the comprehensive unit price value specifically includes: Based on the records of real-time material price data streams over a past observation window, discrete price data is obtained, and the variance of price fluctuations and the distribution of purchasing frequency for various materials at different time points are calculated. Based on the degree of fluctuation, smoothing weights are dynamically allocated to obtain the average price and market share. Obtain the average price and the market share ratio, and then use the weighted fusion formula: ； The base price of the material is calculated; in, This represents the weighted average benchmark price of the materials. Representing the Market share of each supplier Representing the The average price offered by each supplier This represents a price risk compensation coefficient set in conjunction with macroeconomic inflation expectations. Represents the total number of suppliers; Obtain the benchmark price of the material, include it in the preset labor cost and machinery depreciation and amortization expenses for construction operations, and perform additional calculations according to the rate standards promulgated by the cost management department of the project location to generate the comprehensive unit price value.

10. A BIM data-driven engineering cost adjustment system, characterized in that, The system is used to implement the BIM data-driven engineering cost adjustment method according to any one of claims 1-9, and the system includes: The model parsing and topology construction module is used to read BIM model files, process building components using entity parsing algorithms, obtain three-dimensional geometric coordinates and physical material parameters, parse the connection status between building components, obtain the adjacency relationship and hierarchical dependency attributes of components, construct the engineering entity topology network, and define primitive nodes. The mapping matching and bill of quantities generation module is used to semantically associate graphic element nodes with the cost quota database according to the standard engineering quantity calculation specifications, generate logical mapping indexes, execute the full quantity statistics program to calculate the engineering quantity of sub-items, match consumption quotas, and generate benchmark cost bill of quantities. The version comparison and impact identification module is used to respond to the model version update signal, compare the model data through Boolean difference operation, generate a change difference matrix, perform deep traversal analysis based on the engineering entity topology network, identify the affected nodes, and lock the cascaded affected component set; The dynamic recalculation and cost adjustment module is used to perform dynamic recalculation of the quantities of the project for the change difference matrix and the cascaded affected component set, obtain the dynamic recalculation results, connect to the real-time material price data stream, update the comprehensive unit price value, use the comprehensive unit price value to implement incremental adjustment of the benchmark cost list, and output a cost adjustment report.

Images