A water conservancy project cost method, system and medium based on BIM model

Abstract

The present invention relates to the field of electronic digital data processing technology, and specifically to a water conservancy project cost method, system and medium based on a BIM model. The present invention first obtains a BIM model of a water conservancy project, extracts geometric information and engineering information of each engineering component in the water conservancy project according to the BIM model; then performs point cloud sampling according to the geometric information to generate component point cloud data; at the same time, performs encoding preprocessing on the engineering information to generate component engineering parameters; then constructs an engineering quantity allocation model for analyzing component point cloud data and component engineering parameters to obtain component engineering quantities; constructs a real-time unit price database according to market prices and updates it regularly, obtains a real-time unit price set of component engineering quantities according to the real-time unit price database; calculates the component costs of all engineering components of the water conservancy project to obtain the water conservancy project cost. The present invention can improve the accuracy and reliability of engineering quantity calculation in the process of water conservancy project cost, thereby improving the accuracy of engineering cost.

Description

Technical Field

[0001] The present invention relates to the technical field of electronic digital data processing, and in particular to a water conservancy project cost estimation method, system and medium thereof based on a BIM model. Background Art

[0002] Water conservancy projects are an important part of national infrastructure construction, playing an important role in flood control and drought relief, water supply and irrigation, and hydropower generation. With the continuous expansion of the scale of water conservancy project construction and the continuous improvement of technical level, the structure of engineering projects has become more and more complex, and the difficulty of cost management has also increased. Accurate engineering cost calculation is of great significance for project budget preparation, cost control and investment decision-making.

[0003] In recent years, Building Information Modeling (BIM) has been widely used in the field of engineering construction. BIM technology integrates the geometric information of buildings and detailed information of engineering projects by establishing three-dimensional digital models, providing comprehensive information support for engineering costs, thereby significantly improving the cost management efficiency of water conservancy projects. However, the current cost method mainly relies on the basic parameters provided by the BIM model for engineering quantity calculation, which is difficult to fully describe the complex characteristics of water conservancy project components; in addition, traditional cost methods usually handle each engineering link independently, and cannot fully reflect the correlation between each link during the implementation of the project; limited by the problems of insufficient feature characterization and insufficient consideration of link correlation, the accuracy of the existing cost method is obviously insufficient when dealing with complex water conservancy projects.

[0004] Therefore, a water conservancy project cost method, system and medium based on BIM model are proposed. Summary of the invention

[0005] The purpose of the present invention is to provide a water conservancy project cost method, system and medium based on BIM model, by acquiring the BIM model of the water conservancy project, extracting the geometric information and engineering information of each engineering component in the water conservancy project according to the BIM model; then performing point cloud sampling according to the geometric information to generate component point cloud data; at the same time, performing encoding preprocessing on the engineering information to generate component engineering parameters; then constructing an engineering quantity allocation model for analyzing the component point cloud data and component engineering parameters to obtain the component engineering quantity; constructing a real-time unit price database according to market prices and updating it regularly, and obtaining the real-time unit price set of the component engineering quantity according to the real-time unit price database; calculating the component costs of all engineering components of the water conservancy project to obtain the water conservancy project cost.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] A water conservancy project cost method based on BIM model, including:

[0008] Obtain the BIM model of the water conservancy project, and extract the geometric information and engineering information of each engineering component in the water conservancy project according to the BIM model; the engineering information includes the engineering category, engineering sub-items, machinery required for the engineering sub-items, and materials required for the engineering sub-items; the construction process of each engineering component includes Project categories and Project sub-item, Project items require Machinery and materials;

[0009] Performing point cloud sampling according to the geometric information of each engineering component to generate component point cloud data;

[0010] Preprocessing the engineering information to obtain an engineering category code set, an engineering sub-item code set, a sub-item mechanical code set, and a sub-item material code set to form component engineering parameters;

[0011] Constructing a project quantity allocation model, inputting the component point cloud data and the component engineering parameters into the project quantity allocation model to obtain the component engineering quantity; the component engineering quantity includes a sub-item working time vector, a machine time vector and a material demand vector;

[0012] Constructing a real-time unit price database according to market prices and updating it regularly, and obtaining a real-time unit price set of the component engineering quantity according to the real-time unit price database;

[0013] Calculating the component cost of the engineering component according to the component engineering quantity and the real-time unit price set;

[0014] The total cost of all engineering components of the water conservancy project is calculated to obtain the cost of the water conservancy project.

[0015] Preferably, the step of generating the component point cloud data includes: extracting the surface parameter equation of each surface of the engineering component according to the geometric information of the engineering component; performing uniform sampling within the parameter domain of each surface parameter equation to generate parameter domain sampling points; and mapping the parameter domain sampling points of all surfaces of the engineering component to three-dimensional space to obtain the component point cloud data.

[0016] Preferably, the step of preprocessing the engineering information includes:

[0017] Map each engineering category of the engineering component to a continuous integer interval , get the engineering category code, Indicates the maximum number of engineering categories, combined The engineering category codes are obtained by performing the engineering category code set;

[0018] Map each engineering component of the engineering component to a continuous integer interval , get the engineering sub-item code, Indicates the maximum number of engineering sub-items, combined The engineering sub-item codes are obtained by coding the engineering sub-item codes;

[0019] Map each type of machinery required for each engineering sub-item to a continuous integer interval , get the sub-item mechanical code, Indicates the maximum number of mechanical types, merging the Project sub-items The sub-item mechanical codes are obtained to obtain a sub-item mechanical code list, and further merged The sub-item mechanical code list of each engineering sub-item is obtained to obtain the sub-item mechanical code set;

[0020] Map each material required for each engineering sub-item to a continuous integer interval , get the item material code, Indicates the maximum number of material types, combined Project sub-items The sub-item material codes are obtained to obtain a sub-item material code list, and further merged The sub-item material code list of each engineering sub-item is obtained to obtain the sub-item material code set.

[0021] Preferably, the maximum number of engineering categories, the maximum number of engineering sub-items, the maximum number of machinery types and the maximum number of material types are determined based on engineering information of completed water conservancy projects.

[0022] Preferably, the engineering quantity allocation model includes a feature extraction network and an engineering quantity allocation network;

[0023] The feature extraction network includes a geometric feature extraction layer, an engineering feature extraction layer and a feature fusion layer; the geometric feature extraction layer extracts component geometric features according to the component point cloud data; the engineering feature extraction layer extracts component engineering features according to the component engineering parameters; the feature fusion layer processes the component geometric features and the component engineering features to obtain component fusion features;

[0024] The engineering quantity allocation network includes a node initialization layer, a node feature update layer and an output layer; the node initialization layer initializes the sub-item nodes, mechanical nodes and material nodes according to the component fusion features to obtain sub-item node features, mechanical node features and material node features; the node feature update layer performs node feature update on the node features. The output layer obtains the item-by-item working time vector, the machine time vector and the material demand vector according to the node comprehensive characteristics.

[0025] Preferably, the step of extracting component geometric features by the geometric feature extraction layer includes:

[0026] According to the component point cloud data, the three-dimensional coordinates of each point are obtained as point features, and the nearest neighbor points, extracting the spatial features of each point and the nearest neighbor points;

[0027] Combined with the spatial features, the point features of each point are Update once to obtain the geometric features of the primary components;

[0028] Performing maximum pooling processing on the primary component geometric features of all points to obtain component geometric features;

[0029] The step of extracting component engineering features in the engineering feature extraction layer includes:

[0030] Each engineering category code, engineering sub-item code, sub-item machinery code and sub-item material code are embedded in dimensional vector, obtain the engineering category feature vector, engineering sub-item feature vector, sub-item mechanical feature vector and sub-item material feature vector;

[0031] The engineering sub-item feature vector, the sub-item mechanical feature vector and the sub-item material feature vector of each engineering sub-item are spliced ​​to obtain a first comprehensive feature sequence;

[0032] Through Transformer, The first comprehensive feature sequence is processed to obtain Primary component engineering features;

[0033] Will The engineering category feature vector and splicing the engineering features of the primary components to obtain a second comprehensive feature sequence;

[0034] Processing the second comprehensive feature sequence through Transformer to obtain the component engineering feature;

[0035] The step of calculating the component fusion feature by the feature fusion layer includes:

[0036] Mapping the component geometric features to a query vector, mapping the component engineering features to a key vector and a value vector, and calculating a first cross attention;

[0037] Mapping the component engineering features to a query vector, mapping the component geometric features to a key vector and a value vector, and calculating a second cross attention;

[0038] The first cross attention and the second cross attention are fused through layer normalization to obtain the component fusion feature.

[0039] Preferably, the node initialization layer includes a first initialization unit, a second initialization unit and a third initialization unit:

[0040] The component fusion feature initialization is processed by a first initialization unit Item nodes, and obtain item node features;

[0041] Between the sub-item nodes, directed edges are established according to the construction process dependency relationship of the project sub-items;

[0042] The component fusion feature initialization is processed by the second initialization unit mechanical nodes, and obtain mechanical node features;

[0043] Establish an undirected edge between the sub-item node and the machinery node according to the demand relationship between the engineering sub-item and the machinery;

[0044] The component fusion feature initialization is processed by the third initialization unit material nodes, and obtain the material node characteristics;

[0045] Establish an undirected edge between the item node and the material node according to the demand relationship of the engineering item for the material;

[0046] The node feature update layer includes:

[0047] Extraction and Node All nodes that are connected by edges are obtained The set of adjacent nodes of ;

[0048] Calculate the node according to the adjacent node set Aggregation characteristics of

[0049] Combined with the aggregated features, the node is updated through the gated recurrent unit The node characteristics of Update once to obtain the comprehensive characteristics of the node;

[0050] The output layer includes a first engineering quantity allocation unit, a second engineering quantity allocation unit and a third engineering quantity allocation unit:

[0051] Processing the node comprehensive characteristics of the sub-item node through the first engineering quantity allocation unit to obtain the sub-item working time vector;

[0052] Processing the node comprehensive characteristics of the mechanical node through the second engineering quantity allocation unit to obtain the mechanical table time vector;

[0053] The material demand vector is obtained by processing the node comprehensive characteristics of the material node through the third engineering quantity allocation unit.

[0054] Preferably, the total loss function of the engineering quantity allocation model includes the loss of sub-item working hours, the loss of machine hours and the loss of material demand, and the calculation formula is as follows:

[0055] ;

[0056] in, Indicates the loss of working hours by item; Indicates the loss of mechanical time; Indicates loss of material demand; , and Represents the weight coefficient of each loss.

[0057] A water conservancy project cost system based on BIM model, including:

[0058] The information extraction module obtains the BIM model of the water conservancy project and extracts the geometric information and engineering information of each engineering component in the water conservancy project based on the BIM model;

[0059] A point cloud generation module performs point cloud sampling according to the geometric information of each engineering component to generate component point cloud data;

[0060] An information coding module pre-processes the engineering information to obtain an engineering category code set, an engineering sub-item code set, a sub-item mechanical code set and a sub-item material code set to form component engineering parameters;

[0061] A project quantity allocation module constructs a project quantity allocation model, inputs the component point cloud data and the component engineering parameters into the project quantity allocation model, and obtains the component engineering quantity; the component engineering quantity includes a sub-item working time vector, a machine time vector, and a material demand vector;

[0062] A unit price acquisition module constructs a real-time unit price database according to market prices and updates it regularly, and acquires a real-time unit price set of the component engineering quantity according to the real-time unit price database;

[0063] The cost calculation module calculates the component cost of the engineering component according to the component engineering quantity and the real-time unit price set; calculates the sum of the component costs of all the engineering components of the water conservancy project to obtain the water conservancy project cost.

[0064] A storage medium stores computer program instructions, and when the computer program is executed by a processor, a water conservancy project cost estimation method based on a BIM model is implemented.

[0065] Compared with the prior art, the present invention has the following beneficial effects:

[0066] 1. The present invention provides a method for extracting component geometric features based on point clouds. First, point cloud sampling is performed on engineering components, and three-dimensional point cloud data is obtained through parametric equation mapping; then, the nearest neighbor point is calculated for each point and local spatial features are extracted; component geometric features are obtained through multiple feature updates and maximum pooling combined with spatial features; the component geometric feature extraction method of the present invention can accurately capture and better express the spatial features of complex components, improve the accuracy and robustness of geometric feature extraction, make subsequent engineering quantity allocation more accurate, and thus improve the accuracy of water conservancy project cost.

[0067] 2. The present invention encodes engineering categories, sub-items, machinery and material information into high-dimensional feature vectors through the engineering feature extraction layer, and uses Transformer to model the relationship between features; in the feature fusion layer, a two-way cross-attention mechanism is adopted to realize the two-way information flow of geometric features and engineering features through the interaction of query vectors, key vectors and value vectors; finally, a unified component feature representation is obtained through layer normalization fusion; the interaction and influence between features are fully considered, the complementarity of geometric information and engineering information is realized, the comprehensiveness and accuracy of component feature expression are improved, and the accuracy of the cost of water conservancy projects can be improved.

[0068] 3. The engineering quantity allocation model proposed in the present invention respectively initializes the sub-item nodes, mechanical nodes and material nodes through the node initialization layer, and adopts the gated loop unit to perform multiple rounds of node feature updates to realize information interaction and feature evolution between nodes; and comprehensively considers the three dimensions of sub-item working hours, mechanical hours and material requirements in the loss function; the engineering quantity allocation model of the present invention can express the correlation between engineering links, improve the accuracy and reliability of engineering quantity allocation, and make the cost results of water conservancy projects more accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] Figure 1 A schematic diagram of a water conservancy project cost estimation method based on a BIM model provided in an embodiment of the present invention;

[0070] Figure 2 A schematic diagram of the structure of a water conservancy project cost system based on a BIM model is provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0071] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0072] Embodiment 1

[0073] A sluice is an important water conservancy structure, which consists of multiple buildings such as a sluice chamber, upstream and downstream connecting sections, and energy dissipation and anti-scouring facilities. It contains hundreds of engineering components. A single component involves multiple engineering categories, and each engineering category contains multiple engineering sub-items. The engineering quantity calculation is complicated. In order to improve the accuracy and efficiency of engineering quantity calculation and improve the accuracy of engineering cost, a water conservancy engineering cost method based on BIM model is applied. Figure 1 As shown, including:

[0074] Obtain the BIM model of the water conservancy project, and extract the geometric information and engineering information of each engineering component in the water conservancy project according to the BIM model; the engineering information includes the engineering category, engineering sub-items, machinery required for the engineering sub-items, and materials required for the engineering sub-items; the construction process of each engineering component includes Project categories and Project sub-item, Project items require Machinery and materials;

[0075] Performing point cloud sampling according to the geometric information of each engineering component to generate component point cloud data;

[0076] Preprocessing the engineering information to obtain an engineering category code set, an engineering sub-item code set, a sub-item mechanical code set, and a sub-item material code set to form component engineering parameters;

[0077] Constructing a project quantity allocation model, inputting the component point cloud data and the component engineering parameters into the project quantity allocation model to obtain the component engineering quantity; the component engineering quantity includes a sub-item working time vector, a machine time vector and a material demand vector;

[0078] Constructing a real-time unit price database according to market prices and updating it regularly, and obtaining a real-time unit price set of the component engineering quantity according to the real-time unit price database;

[0079] Calculating the component cost of the engineering component according to the component engineering quantity and the real-time unit price set;

[0080] The total cost of all engineering components of the water conservancy project is calculated to obtain the cost of the water conservancy project.

[0081] Component information is extracted based on the BIM model, providing a comprehensive data foundation for subsequent calculations; the precise geometric features of components are obtained through point cloud sampling, and combined with engineering parameter coding, the standardized expression of component features is achieved; the engineering quantity allocation model automatically learns the relationship between component features and engineering quantities through deep learning methods, avoiding the limitations of traditional methods that require manual formulation of calculation rules, and taking into account the construction process dependency and resource demand relationship, so that the engineering quantity allocation is more in line with the actual construction situation and improves the accuracy of cost calculation; the dynamic update of the real-time unit price database ensures the timeliness of cost calculation.

[0082] Furthermore, the step of generating the component point cloud data includes: extracting the surface parameter equation of each surface of the engineering component according to the geometric information of the engineering component; performing uniform sampling within the parameter domain of each surface parameter equation to generate parameter domain sampling points; and mapping the parameter domain sampling points of all surfaces of the engineering component to three-dimensional space to obtain the component point cloud data.

[0083] In an embodiment of the present application, the geometric information includes spatial position information, shape information and topological relationships between engineering components; and the surface parameter equation is extracted by a surface fitting method.

[0084] By performing parametric equation modeling and uniform sampling on the component surface and then mapping it to three-dimensional space to generate point cloud data, the complete and accurate capture of the geometric features of the engineering components is ensured, which can better express the geometric features of the components and provide a more reliable data basis for subsequent engineering quantity calculations.

[0085] The embodiment of the present application takes the spillway pier, a typical engineering component in a sluice project, as an example, and shows some detailed engineering information in Table 1. The engineering component involves multiple engineering categories such as earthwork engineering, steel engineering and concrete engineering.

[0086] Table 1 Engineering information of spillway side pier components

[0087]

[0088] Furthermore, the step of preprocessing the engineering information includes:

[0089] Map each engineering category of the engineering component to a continuous integer interval , get the engineering category code, Indicates the maximum number of engineering categories, combined The engineering category codes are obtained by performing the engineering category code set;

[0090] Map each engineering component of the engineering component to a continuous integer interval , get the engineering sub-item code, Indicates the maximum number of engineering sub-items, combined The engineering sub-item codes are obtained by coding the engineering sub-item codes;

[0091] Map each type of machinery required for each engineering sub-item to a continuous integer interval , get the sub-item mechanical code, Indicates the maximum number of mechanical types, merging the Project sub-items The sub-item mechanical codes are obtained to obtain a sub-item mechanical code list, and further merged The sub-item mechanical code list of each engineering sub-item is obtained to obtain the sub-item mechanical code set;

[0092] Map each material required for each engineering sub-item to a continuous integer interval , get the item material code, Indicates the maximum number of material types, combined Project sub-items The sub-item material codes are obtained to obtain a sub-item material code list, and further merged The sub-item material code list of each engineering sub-item is obtained to obtain the sub-item material code set.

[0093] By mapping engineering categories, engineering sub-items, machinery and material information to continuous integer intervals for encoding, the standardization and structured expression of engineering information is achieved, the data processing efficiency is improved, while the integrity and relevance of the information are guaranteed, providing standardized input data for the engineering quantity allocation model.

[0094] Furthermore, the maximum number of engineering categories, the maximum number of engineering sub-items, the maximum number of machinery types and the maximum number of material types are determined based on engineering information of completed water conservancy projects;

[0095] The maximum amount of various types of information is determined based on the historical data of completed water conservancy projects, which makes the cost method have practical guidance significance and scalability. It not only ensures the adequacy of the coding space, but also avoids the waste of resources, and can adapt to the needs of water conservancy projects of different scales and types.

[0096] Further, the engineering quantity allocation model includes a feature extraction network and an engineering quantity allocation network;

[0097] The feature extraction network includes a geometric feature extraction layer, an engineering feature extraction layer and a feature fusion layer; the geometric feature extraction layer extracts component geometric features according to the component point cloud data; the engineering feature extraction layer extracts component engineering features according to the component engineering parameters; the feature fusion layer processes the component geometric features and the component engineering features to obtain component fusion features;

[0098] The engineering quantity allocation network includes a node initialization layer, a node feature update layer and an output layer; the node initialization layer initializes the sub-item nodes, mechanical nodes and material nodes according to the component fusion features to obtain sub-item node features, mechanical node features and material node features; the node feature update layer performs node feature update on the node features. The output layer obtains the item-by-item working time vector, the machine time vector and the material demand vector according to the node comprehensive characteristics.

[0099] Through the feature extraction network and the engineering quantity allocation network, the deep integration of geometric features and engineering features is achieved; the node-based intelligent allocation of engineering quantities can make full use of the geometric and engineering information of components, improving the accuracy and intelligence level of engineering quantity calculation.

[0100] Furthermore, the step of extracting component geometric features in the geometric feature extraction layer includes:

[0101] According to the component point cloud data, the three-dimensional coordinates of each point are obtained as point features, and the nearest neighbor points, extract the spatial features of each point and the nearest neighbor points; the spatial features are , Indicate point With the nearest neighbor The spatial characteristics of and Respectively indicate points and Point The three-dimensional coordinates of represents a multi-layer perceptron; Indicates that and Splice to get six-dimensional features; calculate each point The nearest neighbor step includes: calculating the distance matrix of all point pairs in the point cloud data ,in, and They are points and Point The three-dimensional coordinates of represents the Euclidean norm; take the minimum from each row of the distance matrix The point corresponding to the index of the value is taken as the nearest neighbor of each point;

[0102] Combined with the spatial features, the point features of each point are The first update obtains the geometric features of the primary components; Updated points The geometrical characteristics of the primary component are ,in Represents maximum pooling processing; represents the activation function; Indicate point of A set of nearest neighbor points; Indicates The updated weight matrix; Indicate point After Point features obtained by the update;

[0103] The primary component geometric features of all points are processed by maximum pooling to obtain component geometric features ;

[0104] The step of extracting component engineering features in the engineering feature extraction layer includes:

[0105] Each engineering category code, engineering sub-item code, sub-item machinery code and sub-item material code are embedded in dimensional vector, obtain the engineering category feature vector, engineering sub-item feature vector, sub-item mechanical feature vector and sub-item material feature vector;

[0106] The engineering sub-item feature vector, the sub-item mechanical feature vector and the sub-item material feature vector of each engineering sub-item are spliced ​​to obtain a first comprehensive feature sequence;

[0107] Through Transformer, The first comprehensive feature sequence is processed to obtain Primary component engineering features;

[0108] Will The engineering category feature vector and splicing the engineering features of the primary components to obtain a second comprehensive feature sequence;

[0109] Processing the second comprehensive feature sequence through Transformer to obtain the component engineering feature;

[0110] Furthermore, the step of calculating the component fusion feature by the feature fusion layer includes:

[0111] The component geometric features are mapped to the query vector, the component engineering features are mapped to the key vector and the value vector, and the first cross attention is calculated. The calculation formula is as follows:

[0112] ;

[0113] in, Indicates the first cross attention; A query vector representing the geometric features of the component; and The key vector and value vector represent the engineering features of the component respectively; express The transpose of express Dimensions; , and represents the weight matrix; Represents the engineering characteristics of components; represents the normalization function;

[0114] The component engineering features are mapped to the query vector, the component geometric features are mapped to the key vector and the value vector, and the second cross attention is calculated. The calculation formula is as follows:

[0115] ;

[0116] in, indicates the second cross attention; A query vector representing the engineering characteristics of the component; and The key vector and value vector represent the geometric features of the component respectively; , and represents the weight matrix; express Dimensions;

[0117] The first cross attention and the second cross attention are fused by layer normalization to obtain the component fusion feature , Representation layer normalization.

[0118] A nearest neighbor-based spatial feature extraction method is adopted to accurately capture the local geometric features of the component by calculating the point pair distance matrix and selecting the nearest neighbor points. The point features are iteratively updated multiple times to achieve feature extraction from local to global. Transformer and cross-attention mechanism are used to achieve deep fusion of geometric features and engineering features, making full use of the two types of features and improving the accuracy of subsequent engineering quantity calculations.

[0119] Furthermore, the node initialization layer includes a first initialization unit, a second initialization unit and a third initialization unit:

[0120] The component fusion feature initialization is processed by a first initialization unit Item nodes, and obtain item node features;

[0121] Between the sub-item nodes, directed edges are established according to the construction process dependency relationship of the project sub-items;

[0122] The component fusion feature initialization is processed by the second initialization unit mechanical nodes, and obtain the mechanical node characteristics. Indicates The number of machinery types required for each engineering item;

[0123] Establish an undirected edge between the sub-item node and the machinery node according to the demand relationship between the engineering sub-item and the machinery;

[0124] The component fusion feature initialization is processed by the third initialization unit material nodes, and obtain the material node characteristics. Indicates The number of material types required for each engineering item;

[0125] An undirected edge is established between the sub-item node and the material node according to the demand relationship of the engineering sub-item for the material.

[0126] Furthermore, the node feature update layer includes:

[0127] Extraction and Node All nodes that are connected by edges are obtained The set of adjacent nodes ;

[0128] Calculate the node according to the adjacent node set Aggregation features ;in, Indicates The aggregated features updated by the time; Representation Node The adjacent nodes of In the The node characteristics of the update; Representation Node In the The node characteristics of the update; Indicates that the node features and To splice;

[0129] Combined with the aggregated features, the node is updated through the gated recurrent unit Node features of the node In the Node features after update , represents a gated recurrent unit; Update the node to obtain the comprehensive characteristics ;

[0130] The output layer includes a first engineering quantity allocation unit, a second engineering quantity allocation unit and a third engineering quantity allocation unit:

[0131] Processing the node comprehensive characteristics of the sub-item node through the first engineering quantity allocation unit to obtain the sub-item working time vector;

[0132] Processing the node comprehensive characteristics of the mechanical node through the second engineering quantity allocation unit to obtain the mechanical table time vector;

[0133] The material demand vector is obtained by processing the node comprehensive characteristics of the material node through the third engineering quantity allocation unit.

[0134] In an embodiment of the present application, the first initialization unit, the second initialization unit and the third initialization unit are based on a multilayer perceptron including two hidden layers; the first engineering quantity allocation unit, the second engineering quantity allocation unit and the third engineering quantity allocation unit are based on a multilayer perceptron including two hidden layers.

[0135] Through the node initialization and feature update mechanism, a graph structure considering the construction process dependency and resource demand relationship is established, which can show the complex relationship between various elements in the actual construction process, make the engineering quantity allocation more in line with the actual situation, and make the calculation results more reliable.

[0136] Table 2 shows the engineering quantity data of some components obtained by the engineering quantity allocation model of the embodiment of the present application.

[0137] Table 2 Engineering quantity data of spillway side pier components

[0138]

[0139] Furthermore, the total loss function of the engineering quantity allocation model includes the loss of sub-item working hours, the loss of machine hours and the loss of material demand, and the calculation formula is as follows:

[0140] ;

[0141] in, Indicates the loss of working hours by item; Indicates the loss of mechanical time; Indicates loss of material demand; , and Represents the weight coefficient of each loss.

[0142] Furthermore, the calculation formula for itemized working time loss is as follows:

[0143] ;

[0144] in, Indicates the total quantity of engineering sub-items of the engineering component; Indicates the first Itemized working hours for each project item; Indicates The actual working hours of each project item;

[0145] The calculation formula of mechanical time loss is as follows:

[0146] ;

[0147] in, Indicates The number of machinery types required for each engineering item; It means that the The first The mechanical hours of the machine; Indicates The first The actual mechanical stage of the machine;

[0148] The material demand loss calculation formula is as follows:

[0149] ;

[0150] in, Indicates The number of material types required for each engineering item; It means that the The first Material requirements for the materials; Indicates The first The actual material demand for this material.

[0151] The multi-objective loss function includes item-by-item labor time loss, machine hour loss and material requirement loss. It adjusts the importance of each part through weights and comprehensively considers all aspects of engineering quantity calculation. It can effectively guide the convergence of engineering quantity allocation model training and improve the accuracy of engineering quantity allocation results.

[0152] Embodiment 2

[0153] The present invention also provides a water conservancy project cost system based on the BIM model, such as Figure 2 As shown, including:

[0154] The information extraction module obtains the BIM model of the water conservancy project and extracts the geometric information and engineering information of each engineering component in the water conservancy project based on the BIM model;

[0155] A point cloud generation module performs point cloud sampling according to the geometric information of each engineering component to generate component point cloud data;

[0156] An information coding module pre-processes the engineering information to obtain an engineering category code set, an engineering sub-item code set, a sub-item mechanical code set and a sub-item material code set to form component engineering parameters;

[0157] A project quantity allocation module constructs a project quantity allocation model, inputs the component point cloud data and the component engineering parameters into the project quantity allocation model, and obtains the component engineering quantity; the component engineering quantity includes a sub-item working time vector, a machine time vector, and a material demand vector;

[0158] A unit price acquisition module constructs a real-time unit price database according to market prices and updates it regularly, and acquires a real-time unit price set of the component engineering quantity according to the real-time unit price database;

[0159] The cost calculation module calculates the component cost of the engineering component according to the component engineering quantity and the real-time unit price set; calculates the sum of the component costs of all the engineering components of the water conservancy project to obtain the water conservancy project cost.

[0160] Furthermore, the step of generating the component point cloud data includes: extracting the surface parameter equation of each surface of the engineering component according to the geometric information of the engineering component; performing uniform sampling within the parameter domain of each surface parameter equation to generate parameter domain sampling points; and mapping the parameter domain sampling points of all surfaces of the engineering component to three-dimensional space to obtain the component point cloud data.

[0161] Furthermore, the step of preprocessing the engineering information includes:

[0162] Map each engineering category of the engineering component to a continuous integer interval , get the engineering category code, Indicates the maximum number of engineering categories, combined The engineering category codes are obtained by performing the engineering category code set;

[0163] Map each engineering component of the engineering component to a continuous integer interval , get the engineering sub-item code, Indicates the maximum number of engineering sub-items, combined The engineering sub-item codes are obtained by coding the engineering sub-item codes;

[0164] Map each type of machinery required for each engineering sub-item to a continuous integer interval , get the sub-item mechanical code, Indicates the maximum number of mechanical types, merging the Project sub-items The sub-item mechanical codes are obtained to obtain a sub-item mechanical code list, and further merged The sub-item mechanical code list of each engineering sub-item is obtained to obtain the sub-item mechanical code set;

[0165] Map each material required for each engineering sub-item to a continuous integer interval , get the item material code, Indicates the maximum number of material types, combined Project sub-items The sub-item material codes are obtained to obtain a sub-item material code list, and further merged The sub-item material code list of each engineering sub-item is obtained to obtain the sub-item material code set.

[0166] Furthermore, the maximum number of engineering categories, the maximum number of engineering sub-items, the maximum number of machinery types and the maximum number of material types are determined based on engineering information of completed water conservancy projects.

[0167] Further, the engineering quantity allocation model includes a feature extraction network and an engineering quantity allocation network;

[0168] The feature extraction network includes a geometric feature extraction layer, an engineering feature extraction layer and a feature fusion layer; the geometric feature extraction layer extracts component geometric features according to the component point cloud data; the engineering feature extraction layer extracts component engineering features according to the component engineering parameters; the feature fusion layer processes the component geometric features and the component engineering features to obtain component fusion features;

[0169] The engineering quantity allocation network includes a node initialization layer, a node feature update layer and an output layer; the node initialization layer initializes the sub-item nodes, mechanical nodes and material nodes according to the component fusion features to obtain sub-item node features, mechanical node features and material node features; the node feature update layer performs node feature update on the node features. The output layer obtains the item-by-item working time vector, the machine time vector and the material demand vector according to the node comprehensive characteristics.

[0170] Table 3 shows the test results of the project quantity allocation model on small irrigation channel projects, medium-sized sluice projects and large reservoir dam projects. It can be seen that the model shows high accuracy for water conservancy projects of different scales.

[0171] Table 3 Results of project quantity allocation

[0172]

[0173] Furthermore, the step of extracting component geometric features in the geometric feature extraction layer includes:

[0174] According to the component point cloud data, the three-dimensional coordinates of each point are obtained as point features, and the nearest neighbor points, extracting the spatial features of each point and the nearest neighbor points;

[0175] Combined with the spatial features, the point features of each point are Update once to obtain the geometric features of the primary components;

[0176] The primary component geometric features of all points are subjected to maximum pooling processing to obtain component geometric features.

[0177] Furthermore, the step of extracting component engineering features by the engineering feature extraction layer includes:

[0178] Each engineering category code, engineering sub-item code, sub-item machinery code and sub-item material code are embedded in dimensional vector, obtain the engineering category feature vector, engineering sub-item feature vector, sub-item mechanical feature vector and sub-item material feature vector;

[0179] The engineering sub-item feature vector, the sub-item mechanical feature vector and the sub-item material feature vector of each engineering sub-item are spliced ​​to obtain a first comprehensive feature sequence;

[0180] Through Transformer, The first comprehensive feature sequence is processed to obtain Primary component engineering features;

[0181] Will The engineering category feature vector and splicing the engineering features of the primary components to obtain a second comprehensive feature sequence;

[0182] Processing the second comprehensive feature sequence through Transformer to obtain the component engineering feature;

[0183] The step of calculating the component fusion feature by the feature fusion layer includes:

[0184] Mapping the component geometric features to a query vector, mapping the component engineering features to a key vector and a value vector, and calculating a first cross attention;

[0185] Mapping the component engineering features to a query vector, mapping the component geometric features to a key vector and a value vector, and calculating a second cross attention;

[0186] The first cross attention and the second cross attention are fused through layer normalization to obtain the component fusion feature.

[0187] Furthermore, the node initialization layer includes a first initialization unit, a second initialization unit and a third initialization unit:

[0188] The component fusion feature initialization is processed by a first initialization unit Item nodes, and obtain item node features;

[0189] Between the sub-item nodes, directed edges are established according to the construction process dependency relationship of the project sub-items;

[0190] The component fusion feature initialization is processed by the second initialization unit mechanical nodes, and obtain the mechanical node characteristics. Indicates The number of machinery types required for each engineering item;

[0191] Establish an undirected edge between the sub-item node and the machinery node according to the demand relationship between the engineering sub-item and the machinery;

[0192] The component fusion feature initialization is processed by the third initialization unit material nodes, and obtain the material node characteristics. Indicates The number of material types required for each engineering item;

[0193] An undirected edge is established between the sub-item node and the material node according to the demand relationship of the engineering sub-item for the material.

[0194] Furthermore, the node feature update layer includes:

[0195] Extraction and Node All nodes that are connected by edges are obtained The set of adjacent nodes ;

[0196] Calculate the node according to the adjacent node set Aggregation characteristics of

[0197] Combined with the aggregated features, the node is updated through the gated recurrent unit The node characteristics of Update the node to obtain the comprehensive characteristics ;

[0198] The output layer includes a first engineering quantity allocation unit, a second engineering quantity allocation unit and a third engineering quantity allocation unit:

[0199] Processing the node comprehensive characteristics of the sub-item node through the first engineering quantity allocation unit to obtain the sub-item working time vector;

[0200] Processing the node comprehensive characteristics of the mechanical node through the second engineering quantity allocation unit to obtain the mechanical table time vector;

[0201] The material demand vector is obtained by processing the node comprehensive characteristics of the material node through the third engineering quantity allocation unit.

[0202] Furthermore, the total loss function of the engineering quantity allocation model includes the loss of sub-item working hours, the loss of machine hours and the loss of material demand, and the calculation formula is as follows:

[0203] ;

[0204] in, Indicates the loss of working hours by item; Indicates the loss of mechanical time; Indicates loss of material demand; , and Represents the weight coefficient of each loss.

[0205] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A water conservancy project cost method based on BIM model, characterized in that: include: Obtain the BIM model of the water conservancy project, and extract the geometric information and engineering information of each engineering component in the water conservancy project based on the BIM model; The engineering information includes engineering category, engineering sub-items, machinery required for engineering sub-items and materials required for engineering sub-items; the construction process of each engineering component includes Project categories and Project sub-item, Project items require Machinery and materials; Performing point cloud sampling according to the geometric information of each engineering component to generate component point cloud data; Preprocessing the engineering information to obtain an engineering category code set, an engineering sub-item code set, a sub-item mechanical code set, and a sub-item material code set to form component engineering parameters; Constructing a project quantity allocation model, inputting the component point cloud data and the component engineering parameters into the project quantity allocation model to obtain the component engineering quantity; the component engineering quantity includes a sub-item working time vector, a machine time vector and a material demand vector; Constructing a real-time unit price database according to market prices and updating it regularly, and obtaining a real-time unit price set of the component engineering quantity according to the real-time unit price database; Calculating the component cost of the engineering component according to the component engineering quantity and the real-time unit price set; Calculate the total cost of all engineering components of the water conservancy project to obtain the cost of the water conservancy project; The engineering quantity allocation model includes a feature extraction network and an engineering quantity allocation network; the feature extraction network includes a geometric feature extraction layer, an engineering feature extraction layer and a feature fusion layer; the geometric feature extraction layer extracts component geometric features according to the component point cloud data; the engineering feature extraction layer extracts component engineering features according to the component engineering parameters; the feature fusion layer processes the component geometric features and the component engineering features to obtain component fusion features; the engineering quantity allocation network includes a node initialization layer, a node feature update layer and an output layer; the node initialization layer initializes sub-item nodes, mechanical nodes and material nodes according to the component fusion features to obtain sub-item node features, mechanical node features and material node features; the node feature update layer processes the node features The output layer obtains the item-by-item working time vector, the machine time vector and the material demand vector according to the node comprehensive characteristics.

2. A water conservancy project cost method based on BIM model according to claim 1, characterized in that: The steps of generating the component point cloud data include: extracting the surface parameter equation of each surface of the engineering component according to the geometric information of the engineering component; performing uniform sampling in the parameter domain of each surface parameter equation to generate parameter domain sampling points; mapping the parameter domain sampling points of all surfaces of the engineering component to three-dimensional space to obtain the component point cloud data.

3. A water conservancy project cost method based on BIM model according to claim 1, characterized in that: The step of preprocessing the engineering information includes: Map each engineering category of the engineering component to a continuous integer interval , get the engineering category code, Indicates the maximum number of engineering categories, combined The engineering category codes are obtained by performing the engineering category code set; Map each engineering component of the engineering component to a continuous integer interval , get the engineering sub-item code, Indicates the maximum number of engineering sub-items, combined The engineering sub-item codes are obtained by coding the engineering sub-item codes; Map each type of machinery required for each engineering sub-item to a continuous integer interval , get the sub-item mechanical code, Indicates the maximum number of mechanical types, merging the Project sub-items The sub-item mechanical codes are obtained to obtain a sub-item mechanical code list, and further merged The sub-item mechanical code list of each engineering sub-item is obtained to obtain the sub-item mechanical code set; Map each material required for each engineering sub-item to a continuous integer interval , get the item material code, Indicates the maximum number of material types, combined Project sub-items The sub-item material codes are obtained to obtain a sub-item material code list, and further merged The sub-item material code list of each engineering sub-item is obtained to obtain the sub-item material code set.

4. A water conservancy project cost method based on BIM model according to claim 3, characterized in that: The maximum number of engineering categories, the maximum number of engineering sub-items, the maximum number of machinery types and the maximum number of material types are determined based on the engineering information of the completed water conservancy projects.

5. The water conservancy project cost method based on the BIM model according to claim 1 is characterized in that: The step of extracting component geometric features in the geometric feature extraction layer includes: According to the component point cloud data, the three-dimensional coordinates of each point are obtained as point features, and the nearest neighbor points, extracting the spatial features of each point and the nearest neighbor points; Combined with the spatial features, the point features of each point are Update once to obtain the geometric features of the primary components; Performing maximum pooling processing on the primary component geometric features of all points to obtain component geometric features; The step of extracting component engineering features in the engineering feature extraction layer includes: Each engineering category code, engineering sub-item code, sub-item machinery code and sub-item material code are embedded in dimensional vector, obtain the engineering category feature vector, engineering sub-item feature vector, sub-item mechanical feature vector and sub-item material feature vector; The engineering sub-item feature vector, the sub-item mechanical feature vector and the sub-item material feature vector of each engineering sub-item are spliced ​​to obtain a first comprehensive feature sequence; Through Transformer, The first comprehensive feature sequence is processed to obtain Primary component engineering features; Will The engineering category feature vector and splicing the engineering features of the primary components to obtain a second comprehensive feature sequence; Processing the second comprehensive feature sequence through Transformer to obtain the component engineering feature; The step of calculating the component fusion feature by the feature fusion layer includes: Mapping the component geometric features to a query vector, mapping the component engineering features to a key vector and a value vector, and calculating a first cross attention; Mapping the component engineering features to a query vector, mapping the component geometric features to a key vector and a value vector, and calculating a second cross attention; The first cross attention and the second cross attention are fused through layer normalization to obtain the component fusion feature.

6. A water conservancy project cost method based on BIM model according to claim 1, characterized in that: The node initialization layer includes a first initialization unit, a second initialization unit and a third initialization unit: The component fusion feature initialization is processed by a first initialization unit Item nodes, and obtain item node features; Between the sub-item nodes, directed edges are established according to the construction process dependency relationship of the project sub-items; The component fusion feature initialization is processed by the second initialization unit mechanical nodes, and obtain mechanical node features; Establish an undirected edge between the sub-item node and the machinery node according to the demand relationship between the engineering sub-item and the machinery; The component fusion feature initialization is processed by the third initialization unit material nodes, and obtain the material node characteristics; Establish an undirected edge between the item node and the material node according to the demand relationship of the engineering item for the material; The node feature update layer includes: Extraction and Node All nodes that are connected by edges are obtained The set of adjacent nodes of ; Calculate the node according to the adjacent node set Aggregation characteristics of Combined with the aggregated features, the node is updated through the gated recurrent unit The node characteristics of Update once to obtain the comprehensive characteristics of the node; The output layer includes a first engineering quantity allocation unit, a second engineering quantity allocation unit and a third engineering quantity allocation unit: Processing the node comprehensive characteristics of the sub-item node through the first engineering quantity allocation unit to obtain the sub-item working time vector; Processing the node comprehensive characteristics of the mechanical node through the second engineering quantity allocation unit to obtain the mechanical table time vector; The material demand vector is obtained by processing the node comprehensive characteristics of the material node through the third engineering quantity allocation unit.

7. A water conservancy project cost method based on BIM model according to claim 1, characterized in that: The total loss function of the engineering quantity allocation model includes the loss of sub-item working hours, machine hour loss and material demand loss, and the calculation formula is as follows: ； in, Indicates the loss of working hours by item; Indicates the loss of mechanical time; Indicates loss of material demand; , and Represents the weight coefficient of each loss.

8. A water conservancy project cost system based on BIM model, characterized in that: include: The information extraction module obtains the BIM model of the water conservancy project and extracts the geometric information and engineering information of each engineering component in the water conservancy project based on the BIM model; A point cloud generation module performs point cloud sampling according to the geometric information of each engineering component to generate component point cloud data; An information coding module pre-processes the engineering information to obtain an engineering category code set, an engineering sub-item code set, a sub-item mechanical code set and a sub-item material code set to form component engineering parameters; A project quantity allocation module constructs a project quantity allocation model, inputs the component point cloud data and the component engineering parameters into the project quantity allocation model, and obtains the component engineering quantity; the component engineering quantity includes a sub-item working time vector, a machine time vector, and a material demand vector; A unit price acquisition module constructs a real-time unit price database according to market prices and updates it regularly, and acquires a real-time unit price set of the component engineering quantity according to the real-time unit price database; A cost calculation module, which calculates the component cost of the engineering component according to the component engineering quantity and the real-time unit price set; Calculate the total cost of all engineering components of the water conservancy project to obtain the cost of the water conservancy project; The engineering quantity allocation model includes a feature extraction network and an engineering quantity allocation network; the feature extraction network includes a geometric feature extraction layer, an engineering feature extraction layer and a feature fusion layer; the geometric feature extraction layer extracts component geometric features according to the component point cloud data; the engineering feature extraction layer extracts component engineering features according to the component engineering parameters; the feature fusion layer processes the component geometric features and the component engineering features to obtain component fusion features; the engineering quantity allocation network includes a node initialization layer, a node feature update layer and an output layer; the node initialization layer initializes sub-item nodes, mechanical nodes and material nodes according to the component fusion features to obtain sub-item node features, mechanical node features and material node features; the node feature update layer processes the node features The output layer obtains the item-by-item working time vector, the machine time vector and the material demand vector according to the node comprehensive characteristics.

9. A storage medium having computer program instructions stored thereon, characterized in that: When the computer program is executed by a processor, a water conservancy project cost estimation method based on a BIM model as described in any one of claims 1 to 7 is implemented.

Images