A curtain wall construction management method and system based on a knowledge graph

Abstract

The application discloses a kind of based on knowledge graph's curtain wall construction management method and system, and the entity and relationship are extracted from curtain wall construction data by knowledge extraction method, construct the multidimensional knowledge network including construction parameter, form semantic association structure. The semantic path between nodes is traversed using the associated query method, and potential conflict points in construction connection are identified. If the conflict involves job coordination, integrate building information model data, determine the coordination requirement level through the monitoring and early warning module, generate the optimized job sequence, and extract the process path from it, combine resource allocation logic to handle material and manpower distribution, and form a draft of the scheduling plan. Finally, integrate the feedback mechanism to update the relationship weight in the knowledge network, and realize dynamic management configuration under risk prediction. The application effectively solves the conflict identification, coordination optimization and resource scheduling problems in curtain wall construction through the construction and iterative update of knowledge network, improves construction efficiency, reduces risk, and realizes intelligent management.

Description

Technical Field

[0001] This invention relates to the field of curtain wall engineering technology, and in particular discloses a curtain wall construction management method and system based on knowledge graph. Background Technology

[0002] As an important external maintenance structure of modern buildings, curtain wall engineering directly affects the building's aesthetics, energy efficiency, and safety. Its construction management involves multiple aspects such as design, material selection, process execution, and on-site coordination, and plays a key role in project quality and schedule control.

[0003] As building forms become increasingly complex, curtain wall construction requires handling a large amount of heterogeneous information, including design parameters, material properties, construction specifications, and historical experience, all of which have intricate relationships. Current management methods largely rely on independent document systems, BIM models, or experience tables. While this information can be stored and retrieved, it is difficult to effectively utilize the implicit relationships in practical applications. For example, parameter adjustments during the design phase often fail to quickly translate into construction process selection, and changes in material properties are not promptly reflected in risk assessments, resulting in a lack of comprehensive support for on-site decision-making. During overlapping operations, conflicts between different trades are easily overlooked, and resource allocation is primarily based on manual experience, making precise optimization difficult. These problems stem from a core factor: the lack of a multi-dimensional semantic relationship network for various types of knowledge during construction. Information such as design parameters, material properties, process standards, and historical cases, which should form close connections, are often stored and used in a fragmented manner. When a design parameter changes, it is impossible to automatically link it to related material compatibility, construction procedure constraints, and risk points from similar past cases; when multiple trades are working simultaneously, it is impossible to quickly identify potential conflict areas and resource competition through semantic relationships. This lack of connection makes it difficult for knowledge to be deeply explored and reused, thus creating a prominent contradiction in complex projects: on the one hand, a large amount of data and experience are accumulated, but on the other hand, it is difficult to quickly integrate this scattered knowledge into actionable guidance at key decision-making nodes, resulting in delayed risk prediction, poor process connection, and uneven resource allocation.

[0004] Therefore, how to construct and utilize a multi-dimensional knowledge semantic association network throughout the entire curtain wall construction process to achieve early risk identification, dynamic optimization of procedures, and precise resource allocation has become a key issue in improving construction management. Summary of the Invention

[0005] This invention provides a knowledge graph-based method and system for curtain wall construction management, aiming to solve at least one of the defects in the prior art.

[0006] One aspect of this invention relates to a knowledge graph-based curtain wall construction management method, comprising the following steps: S100. Using knowledge extraction methods, entities and relationships are extracted from construction data to construct a multi-dimensional knowledge network, resulting in a semantic association structure containing construction parameters. S200. Based on the multidimensional knowledge network, the semantic path between nodes is traversed using the association query method to identify potential conflict points for construction coordination issues. S300. If potential conflict points involve coordination of curtain wall construction operations, the monitoring and early warning module integrates building information model data to determine the level of coordination needs and obtain an optimized work sequence. S400: Extract process paths from the optimized work sequence, combine them with resource allocation logic to process material and manpower distribution, and determine a draft scheduling plan; S500: For the draft scheduling scheme, integrate the application loop in the feedback mechanism, update the relation weights in the multi-dimensional knowledge network, and obtain the management configuration under risk prediction.

[0007] Further, step S100 includes: S110. Obtain the construction text sequence from the curtain wall construction data, and use the named entity recognition model to extract construction entity objects from the construction text sequence. S120. Analyze the contextual dependency paths of construction entities to determine the relationships between construction procedures and construction parameter indicators. S130. Construct a multidimensional knowledge graph based on the relationship between construction procedures and construction parameter indicators, and analyze the topological structure of the multidimensional knowledge graph. S140. Transform the topological structure into a semantic association matrix to obtain a semantic association structure containing construction parameters.

[0008] Further, step S200 includes: S210. Obtain the construction process entities in the multidimensional knowledge network and extract the set of construction connection node pairs with logical succession relationships; S220. Traverse the set of construction connection nodes to retrieve multi-hop semantic paths. If there is spatiotemporal overlap between nodes under the multi-hop semantic path, generate a spatiotemporal conflict feature vector. S230. Calculate the comprehensive conflict index based on the spatiotemporal conflict feature vector to obtain a set of high-risk connection paths where the comprehensive conflict index exceeds the limit. S240. Based on the high-risk connection path set, reverse index to the corresponding construction process entity to determine potential conflict points in construction connection.

[0009] Further, step S300 includes: S310. Obtain potential conflict points and associated building information model data, and map the three-dimensional coordinates of potential conflict points to geometric components within the building information model to establish spatial index relationships. S320. Based on the spatial index relationship, access the real-time sensor stream of the monitoring and early warning module, analyze the geometric components bound to dynamic environmental parameters to quantify the interference intensity of curtain wall construction operations and determine the coordination requirement level. S330. If the coordination requirement level is higher than the preset operation safety threshold, then construct a constraint adjustment matrix that satisfies the spatiotemporal avoidance conditions. S340. Input the constraint adjustment matrix into the original schedule network to perform topology rearrangement and obtain the optimized job sequence.

[0010] Further, step S400 includes: S410. Parse the optimized job sequence to extract the process path with time constraints; S420. Based on the process path and after verifying the yard capacity, generate a material distribution matrix that meets the logistics constraints. S430. Combine the material distribution matrix to correct the execution time of process nodes and determine the operation time window that meets the manpower distribution requirements; S440: Integrate the spatiotemporal parameters of the operation time window and the process path, calculate the precise start and end times of the process nodes to generate a draft scheduling scheme containing resource allocation details.

[0011] Further, step S500 includes: S510. Analyze the draft scheduling scheme to extract the execution deviation feature vector that shows a tendency for timing delays or resource conflicts. S520. Input the execution deviation feature vector into the multidimensional knowledge network model, and update the relation weights in the multidimensional knowledge network according to the historical interaction frequency. S530. Construct a risk transmission matrix based on the updated relationship weights and calculate the cumulative failure probability to mark high-risk warning objects; S540 generates a set of control parameters, including instructions for resource replenishment and process adjustment, for high-risk early warning targets, and obtains the management configuration under risk prediction.

[0012] Another aspect of the present invention relates to a knowledge graph-based curtain wall construction management system for executing the aforementioned knowledge graph-based curtain wall construction management method, comprising: The semantic association structure acquisition module is used to extract entities and relationships from building construction data using knowledge extraction methods, construct a multi-dimensional knowledge network, and obtain a semantic association structure containing construction parameters. The potential conflict point identification module is used to identify potential conflict points for construction coordination issues by traversing the semantic paths between nodes based on a multidimensional knowledge network and using an association query method. The optimized work sequence acquisition module is used to determine the coordination requirement level and obtain an optimized work sequence if potential conflict points involve the coordination of curtain wall construction operations. This is achieved by integrating building information model data with the monitoring and early warning module. The scheduling plan draft determination module is used to extract process paths from the optimized work sequence, combine resource allocation logic to process material and manpower distribution, and determine the scheduling plan draft. The management configuration acquisition module is used to update the relation weights in the multidimensional knowledge network based on the application loop in the integrated feedback mechanism for the draft scheduling scheme, and obtain the management configuration under risk prediction.

[0013] The beneficial effects achieved by this invention are as follows: This invention provides a knowledge graph-based method and system for curtain wall construction management. Addressing the challenges of complex work coordination, frequent conflicts during construction, uneven resource allocation, and difficulty in timely prediction of potential risks in curtain wall construction, this invention extracts entities and relationships from curtain wall construction data using knowledge extraction methods. This constructs a multi-dimensional knowledge network containing construction parameters, forming a semantically related structure. Based on this, an association query method traverses the semantic paths between nodes to identify potential conflict points in construction coordination. If a conflict involves work coordination, building information model (BIM) data is integrated, and a monitoring and early warning module determines the coordination requirement level, generating an optimized work sequence. Work process paths are extracted from this sequence, and material and manpower distribution are processed using resource allocation logic to form a draft scheduling plan. Finally, an integrated feedback mechanism updates the relation weights in the knowledge network, enabling dynamic management configuration under risk prediction. This invention effectively solves the problems of conflict identification, coordination optimization, and resource scheduling in curtain wall construction through the construction and iterative updating of the knowledge network, improving construction efficiency, reducing risks, and achieving intelligent management. Attached Figure Description

[0014] Figure 1 This is a flowchart illustrating an embodiment of the knowledge graph-based curtain wall construction management method of the present invention. Figure 2 This is a functional block diagram of an embodiment of the knowledge graph-based curtain wall construction management system of the present invention.

[0015] Explanation of icon numbers: 10. Semantic association structure acquisition module; 20. Potential conflict point identification module; 30. Optimized job sequence acquisition module; 40. Scheduling scheme draft identification module; 50. Management configuration acquisition module. Detailed Implementation

[0016] To better understand the above technical solutions, the following will provide a detailed explanation of the technical solutions in conjunction with the accompanying drawings and specific implementation methods.

[0017] like Figure 1 As shown, the first embodiment of the present invention proposes a curtain wall construction management method based on knowledge graphs, including the following steps: Step S100: Use knowledge extraction methods to extract entities and relationships from construction data, construct a multi-dimensional knowledge network, and obtain a semantic association structure containing construction parameters.

[0018] Taking the entire lifecycle data of curtain wall construction as the core, knowledge extraction methods (such as Named Entity Recognition (NER), relation extraction, and attribute extraction) are used to mine key information from multi-source building construction data and construct a semantic multi-dimensional knowledge network (i.e., curtain wall construction knowledge graph).

[0019] 1. Data sources: covering curtain wall design drawings (such as component dimensions, material parameters, and installation details), construction process standards (such as keel installation process and glass hoisting specifications), resource data (such as material models / inventory, personnel skills / shifts, and equipment types / status), environmental data (such as construction site conditions and weather warnings), and historical construction cases (such as cases of process conflicts and records of quality issues).

[0020] 2. Entity and Relationship Extraction: The core entities extracted include construction procedures (keel installation, glass installation, sealing and caulking, etc.), material resources (aluminum alloy profiles, tempered glass, sealant, etc.), personnel and equipment (installation teams, hoisting machinery, measuring instruments, etc.), and technical parameters (installation accuracy, load-bearing capacity requirements, sealing performance indicators, etc.). At the same time, the relationships between entities are extracted (e.g., the prerequisite for "keel installation" is "embedded part positioning", the matching equipment for "tempered glass" is "vacuum suction cup lifting tool", and the process parameter for "sealing and caulking" is "temperature ≥ 5℃").

[0021] 3. Semantic association structure generation: The extracted entities, relations, and attributes are organized in the form of knowledge graph triples (entity 1-relation-entity 2) to form a semantic association structure containing construction parameters, i.e., a multi-dimensional knowledge network, which provides knowledge support for subsequent conflict query and path analysis.

[0022] Step S200: Based on the multidimensional knowledge network, the semantic path between nodes is traversed using the association query method to identify potential conflict points for construction coordination issues.

[0023] Based on the multidimensional knowledge network constructed in step S100, the semantic paths between nodes are traversed using association query methods (such as SPARQL (SPARQL Protocol and RDF Query Language) query language and graph traversal algorithms) to uncover potential contradictions between various processes, resources, and parameters in curtain wall construction and to locate construction connection problems.

[0024] 1. Semantic path traversal logic: Taking the key processes of curtain wall construction as the core nodes, query the dependencies, resource occupation conflicts, and parameter matching of their preceding / following processes (e.g., the "glass hoisting" process requires the use of "hoisting machinery". If the hoisting machinery is also used by the "keel transportation" process, a resource conflict will occur; "sealing and caulking" requires an ambient temperature ≥5℃. If the temperature is lower than the threshold during the construction plan period, a process parameter conflict will occur).

[0025] 2. Identification of potential conflict points: Based on the core pain points of curtain wall construction (such as conflicts in the sequence of work processes, conflicts in resource allocation, conflicts in process parameters, and conflicts in the work area), the contradictions found during the process are marked as potential conflict points and classified according to the degree of impact (e.g., Level A: serious conflicts that lead to work stoppages; Level B: general conflicts that affect construction efficiency; Level C: minor quality defects).

[0026] Step S300: If potential conflict points involve coordination of curtain wall construction operations, the monitoring and early warning module integrates building information model data to determine the level of coordination needs and obtain an optimized work sequence.

[0027] If potential conflict points involve coordination of curtain wall construction operations (such as conflicts in process connection, overlapping work areas, and team cooperation), the monitoring and early warning module will be activated to conduct in-depth analysis by integrating Building Information Modeling (BIM) data.

[0028] 1. BIM Data Fusion Analysis: The conflict point information in the multi-dimensional knowledge network is integrated with the curtain wall BIM model (including the three-dimensional coordinates of components, installation space, and process sequence). The conflict location, the involved processes, and the scope of impact are visualized. Based on the severity of the conflict, the duration of the impact on the construction period, and the difficulty of resource allocation, the coordination requirement level is determined (e.g., Level 1 coordination: process sequence needs to be adjusted; Level 2 coordination: spare resources need to be allocated; Level 3 coordination: work surface division needs to be optimized).

[0029] 2. Work Sequence Optimization: Based on the coordination requirement level, combined with construction process constraints and resource availability, rearrange the execution sequence of work processes to eliminate conflict points (such as staggering overlapping work periods, prioritizing critical path processes, and allocating auxiliary equipment to resolve resource conflicts). Finally, output the optimized work sequence, clarifying the execution time, resource allocation, and work area of ​​each process.

[0030] Step S400: Extract process paths from the optimized work sequence, combine them with resource allocation logic to process material and manpower distribution, and determine a draft scheduling plan.

[0031] Key process paths (especially critical paths in curtain wall construction, such as embedded part installation → keel installation → glass installation → sealing acceptance) are extracted from the optimized work sequence. Combined with the preset resource allocation logic, materials, manpower, and equipment are dynamically planned to generate a draft scheduling plan.

[0032] 1. Resource allocation logic: Follow the principles of "prioritizing key processes, maximizing resource utilization, and optimizing costs". For example, prioritize the allocation of high-skilled teams and core equipment for critical path processes; match process execution nodes according to material arrival time to avoid material backlog or shortages; optimize personnel scheduling based on work area distribution to reduce time spent traveling across regions.

[0033] 2. Draft scheduling plan: Includes process execution schedule, resource allocation list (material type / usage / arrival time, personnel team / division of labor / shift, equipment type / usage period / location), cost budget, risk contingency plan, etc., to ensure the plan is feasible.

[0034] Step S500: For the draft scheduling scheme, integrate the application loop in the feedback mechanism, update the relation weights in the multidimensional knowledge network, and obtain the management configuration under risk prediction.

[0035] For the draft scheduling plan, the feedback mechanism application loop of the integrated construction process (such as actual conflict feedback, resource utilization efficiency feedback, and project completion feedback) is used to dynamically update the multi-dimensional knowledge network and generate management configuration under risk prediction.

[0036] 1. Feedback Data Integration: Collect actual data during the implementation of the plan (such as the deviation between the actual time of the process and the plan, the actual utilization rate of resources, and the types of new conflict points), and transform this data into the basis for updating the knowledge graph.

[0037] 2. Relationship weight update: Adjust the relationship weights between entities in the multidimensional knowledge network (e.g., if a certain type of resource conflict occurs frequently in historical cases, increase the weight of that relationship so that subsequent queries can prioritize identifying such conflicts; if a certain process has a high success rate in connection, decrease its conflict warning weight).

[0038] 3. Management Configuration Generation: Based on the updated multidimensional knowledge network, the conflict prediction rules, resource allocation logic, and coordination requirement classification standards are optimized to form a management configuration under risk prediction, enabling subsequent construction management to more accurately identify high-risk points and allocate resources more efficiently.

[0039] Furthermore, the knowledge graph-based curtain wall construction management method proposed in this embodiment includes step S100 as follows: Step S110: Obtain the construction text sequence in the curtain wall construction data, and use the named entity recognition model to extract construction entity objects from the construction text sequence.

[0040] The following formula is used to evaluate the overall performance of extracting construction entity objects from the curtain wall construction text sequence: (1) In formula (1), Represents the extraction of construction entity objects Fraction, This represents the recall rate for identifying construction entities. This represents the precision rate of construction entity identification. The control logic of formula (1) is based on recall rate. and accuracy The weighted combination comprehensively evaluates the performance of extracted construction entities, avoiding the one-sidedness of a single indicator.

[0041] The accuracy of construction entity identification is calculated using the following formula: (2) In formula (2), Represents a set of construction text sequences. Indicates the first The number of words in a text sequence In the context Vocabulary under conditions Identified as an entity The probability, The total number of text sequences is represented. The control logic of formula (2) is to take the set of text sequences as a whole, first calculate the average value of "the probability of each word being identified as an entity in the corresponding context" in each text sequence, and then take the overall average of the average value of all text sequences to measure the "accuracy" of the construction entity recognition result (i.e. the probability level of words identified as entities actually meeting the requirements).

[0042] In practical applications of curtain wall construction projects, the first step is to obtain relevant construction text sequences from sources such as project documents, construction logs, and design drawings. These text sequences include detailed construction descriptions such as "the strength of the fixing bolts of the steel frame needs to be checked before installing the glass panels." When using a named entity recognition model, which is based on natural language processing technology, it can automatically identify and extract key entity objects from the text. Specifically, named entity recognition models typically utilize pre-trained deep learning frameworks, such as models based on BERT (Bidirectional Encoder Representations from Transformers), to label entity categories by performing word vector encoding and sequence labeling on the text sequence. In one embodiment, for a text describing "using 10mm thick Low-E glass fixed to an aluminum alloy frame," the named entity recognition model will extract "Low-E glass" as a material entity, "10mm" as a parameter entity, and "aluminum alloy frame" as a component entity, thus forming a list of entity objects for subsequent analysis. This extraction process helps transform unstructured text into structured data, improving the efficiency of construction management.

[0043] Step S120: Analyze the contextual dependency path of the construction entity object to determine the construction process relationship and construction parameter indicators.

[0044] After extracting entity objects, it is necessary to analyze the contextual dependency paths of these objects. This involves examining the grammatical and semantic relationships of entities within sentences to determine the relationships between construction procedures and related parameter indicators. For example, in a text sequence containing the phrase "After the steel frame is installed, the glass panels are sealed, and the sealant thickness must reach 5mm," dependency path analysis identifies the sequential relationship between "steel frame installation" and "glass panel sealing," meaning the former is a prerequisite for the latter. Simultaneously, "sealant thickness 5mm" is extracted as a parameter indicator. Specifically, contextual dependency path analysis employs a dependency parser, such as the Stanford Parser tool, which constructs a sentence dependency tree, displaying subject-verb-object relationships and modification paths, thereby inferring the causal chain of procedures such as "installation-sealing," and the quantitative value of parameters such as thickness. This analysis not only reveals the logical order of procedures but also quantifies parameter indicators, providing a basis for optimizing the construction process. In one embodiment, assuming the entity objects include "curtain wall support" and "wind load test", path analysis will determine that "support fixation" depends on "load calculation" and associate parameters such as "wind pressure value 0.5kN / m²", which helps to identify potential risk points.

[0045] Step S130: Construct a multidimensional knowledge graph based on the construction process relationship and construction parameter indicators, and analyze the topological structure of the multidimensional knowledge graph.

[0046] The overall structure of a multidimensional knowledge graph is derived from the following formula: (3) In formula (3), This represents the overall structure of a multidimensional knowledge graph. Represents the set of construction process nodes. Indicates the first Each construction process node This represents the set of edges representing relationships between processes. Indicate process to process The connecting edge, Represents a set of relation types. Indicates the first There are various types of construction process relationships. The control logic of formula (3) is to clarify the core components of the multidimensional knowledge graph in the construction field through the triple form of "node set + edge set + relationship type set", and to clearly construct the relationship structure between construction processes.

[0047] Based on the aforementioned process relationships and parameter indicators, a multidimensional knowledge graph is constructed. This multidimensional knowledge graph is a graph-structured representation with entities as nodes and relationships as edges, capable of capturing construction knowledge from multiple dimensions. For example, "steel frame" can be used as a node, connected to "glass panel" through an "installation" relationship, and labeled with parameters such as "bolt torque 50Nm". In one embodiment, the construction process involves using a graph database such as Neo4j to import entities and relationships, forming a multi-layered graph that includes materials, processes, and parameters.

[0048] Subsequently, when analyzing its topology, it is necessary to examine the graph's connectivity, centrality, and cluster structure to understand the overall layout of the knowledge. Specifically, topological analysis calculates the degree of nodes and the shortest path; for example, identifying the high centrality of the "sealing process" node indicates that it is a critical process. This analysis helps to identify bottlenecks and improve project coordination.

[0049] Step S140: Transform the topology into a semantic association matrix to obtain a semantic association structure containing construction parameters.

[0050] The semantic association matrix containing construction parameters is derived using the following formula: (4) In formula (4), This represents a semantic association matrix containing construction parameters. Indicates the first The construction parameters and the first The numerical value of the correlation between semantic elements This indicates the total number of construction parameters. The total number of semantic elements is represented by the number of elements in the semantic association matrix. Each element in the semantic association matrix quantifies the degree of association between the parameters and the semantics. The control logic of formula (4) quantifies the degree of association between the two types of objects, "construction parameters" and "semantic elements", one by one through a two-dimensional matrix, thereby presenting the semantic association strength between the two in a systematic and intuitive way.

[0051] After transforming the topological structure into a semantic association matrix, a matrix form is obtained, where rows and columns represent entities, and element values ​​represent association strength with embedded parameters. For example, in the semantic association matrix, the intersection of the "steel frame" row and the "glass panel" column can be assigned an association weight of 0.8, with the parameter "fixed strength". In one embodiment, the transformation process uses an adjacency matrix to represent the edges of the graph and incorporates semantic vectors to enhance associations. This structure facilitates further processing by machine learning algorithms, enabling techniques such as predictive maintenance.

[0052] Furthermore, in the knowledge graph-based curtain wall construction management method proposed in this embodiment, step S200 includes: Step S210: Obtain the construction process entities in the multidimensional knowledge network and extract the set of construction connection node pairs with logical succession relationships.

[0053] The set of construction connection nodes is derived using the following formula: (5) In formula (5), This represents the set of connection nodes of construction process entities. and This represents a construction process node in a multidimensional knowledge network. This represents the set of all construction process nodes. This function represents the logical connection between nodes; a value of 1 indicates that a connection exists. and These represent the temporal attributes of the process nodes, ensuring that the preceding process is executed before the subsequent process. The control logic of formula (5) is to select legal construction connection node pairs from all construction process nodes that simultaneously satisfy the two conditions of "logical succession relationship" and "temporal sequence constraint", thereby clarifying the reasonable connection relationship between processes.

[0054] When extracting construction process entities from a multidimensional knowledge network, the first step is to identify specific process nodes related to curtain wall installation, such as "aluminum alloy keel fixing" and "unit panel hoisting." These entities are extracted from the previously constructed graph through a query interface, forming a list of process entities. Subsequently, a set of construction connection node pairs with logical succession relationships is extracted. This involves analyzing the edge relationships between entities; for example, "keel fixing" must precede "unit panel hoisting" because the latter depends on the structural stability of the former. Specifically, this extraction process uses a graph traversal algorithm to scan adjacent nodes, filtering out pairs with sequential dependencies. For example, "keel fixing - unit panel hoisting" is treated as a node pair, and similar pairs such as "sealant application - waterproofing test" are collected, resulting in a set of construction connection node pairs for subsequent conflict detection. This method helps to systematically capture the inherent logic between processes, ensuring complete coverage of connection relationships.

[0055] Step S220: Traverse the set of construction connection nodes to retrieve multi-hop semantic paths. If there is spatiotemporal overlap among the nodes under the multi-hop semantic path, generate a spatiotemporal conflict feature vector.

[0056] The spatiotemporal conflict feature vector is generated using the following formula: (6) In formula (6), Represents the spatiotemporal conflict feature vector. arrive This represents the weight coefficients for each dimension. Indicates the intensity of temporal overlap. Indicates the spatial overlap strength. Indicates the degree of resource priority conflict. Indicates the intensity of resource competition. This indicates the transpose operation. The control logic of formula (6) is the weighted integration logic of spatiotemporal conflict characteristics. The core is to integrate the key indicators of the three types of construction conflicts, namely "time, space and resources", into a unified feature vector through weight allocation, which is used to quantify the comprehensive degree of spatiotemporal conflict.

[0057] When traversing the set of construction connection node pairs to retrieve multi-hop semantic paths, it is necessary to start from each node pair and expand to more distant related paths. For example, from "keel fixing - unit panel hoisting" to a multi-hop chain involving "glass fitting", the path is "keel fixing → unit panel hoisting → glass fitting". If the nodes under the multi-hop semantic path have spatiotemporal overlap, a spatiotemporal conflict feature vector is generated. Specifically, spatiotemporal overlap refers to multiple processes occupying the same spatial resources within the same time period. For example, if "unit panel hoisting" and "glass fitting" are planned to be carried out on the same floor and on the same day, it will lead to conflicts in lifting equipment. In this case, the spatiotemporal conflict feature vector is represented as a multi-dimensional array, including dimensions such as time overlap degree and space occupancy rate. For example, the time overlap degree is calculated based on the intersection of the duration of the processes, and the space occupancy rate is obtained by matching the floor coordinates. This generation process provides a data foundation for conflict assessment by quantifying the overlap situation.

[0058] Step S230: Calculate the comprehensive conflict index based on the spatiotemporal conflict feature vector to obtain a set of high-risk connection paths where the comprehensive conflict index exceeds the limit.

[0059] The comprehensive index of multidimensional spatiotemporal conflict characteristics is calculated using the weighted Euclidean norm using the following formula: (7) In formula (7), This indicates the overall conflict index. The dimension of the spatiotemporal conflict feature vector. Indicates the first Weight coefficients for each feature dimension Indicates the number of conflict feature samples. Indicates the first The first dimension Each conflict feature value. The control logic of formula (7) is to first perform local aggregation (Euclidean norm) on the feature values ​​within each conflict feature dimension, and then perform global weighted integration of the aggregation results of different dimensions through weights, so as to finally obtain a comprehensive index reflecting the severity of the construction time and space conflict.

[0060] The following formula defines the criteria for screening high-risk paths that exceed the safety threshold: (8) In formula (8), This represents a set of high-risk connection paths. Indicates the first Connecting paths, Indicates the first The comprehensive conflict index corresponding to each path, This indicates that the conflict index has exceeded the threshold. This represents the total number of connection paths to be evaluated. The control logic of formula (8) is to select the paths whose "comprehensive conflict index exceeds the preset safety threshold" from all the construction connection paths to be evaluated, form a set of high-risk paths, and identify the risk links that need to be focused on.

[0061] When calculating the comprehensive conflict index based on the spatiotemporal conflict feature vector, the various dimensions of the spatiotemporal conflict feature vector are weighted and summed. For example, the time dimension has a higher weight because delays affect the overall progress.

[0062] Specifically, assuming the spatiotemporal conflict feature vector includes temporal overlap (0.7), spatial overlap (0.5), and resource sharing (0.3), the index is calculated as a weighted average of these values, resulting in a value such as 0.55. If this index exceeds a preset threshold, such as 0.5, it is considered out of bounds, thus obtaining a set of high-risk connection paths, for example, including the path "keel fixing → unit panel hoisting → glass fitting". This calculation helps quantify the risk level, facilitating the prioritization of potential problems.

[0063] Step S240: Based on the high-risk connection path set, reverse index to the corresponding construction process entity to determine the potential conflict points of construction connection.

[0064] The following formula is used to find the construction process entity with the greatest conflict impact through reverse indexing: (9) In formula (9), This indicates the location of the largest potential conflict point identified. Represents the set of all possible construction process entities. This represents a set of high-risk connection paths. Represents the entity of the process Does it belong to a path? Indicator functions, Represents the entity of the process In the path The conflict impact value in the formula (9) is determined by statistically analyzing the total conflict impact of construction process entities in all high-risk paths, identifying the process entity with the largest total impact, and locating the potential risk point that contributes the most to the conflict impact.

[0065] When the high-risk connection path set is reverse-indexed to the corresponding construction process entity, the source entity is located through a path backtracking mechanism. For example, the high-risk path is reversed to "keel fixing" and "unit panel hoisting" to identify these as potential conflict points in the construction connection. Specifically, the reverse indexing involves graph database queries to match path endpoints to entities, thereby identifying conflict points such as those caused by improper equipment scheduling. This identification process can accurately locate the source of the problem and support adjustments to the work process to avoid delays.

[0066] Preferably, in the knowledge graph-based curtain wall construction management method proposed in this embodiment, step S300 includes: Step S310: Obtain potential conflict points and associated building information model data, and map the three-dimensional coordinates of potential conflict points to geometric components within the building information model to establish spatial index relationships.

[0067] The spatial index relationship is established using the following formula: (10) In formula (10), The function representing the mapping from three-dimensional coordinates to geometric components. Represents a three-dimensional spatial coordinate system. Represents a set of geometric components in a building information model. Represents three-dimensional coordinate values. Indicates the first A geometric component, Indicates the first The spatial volume occupied by each geometric component The total number of geometric components is represented. The control logic of formula (10) is to establish the correspondence between three-dimensional spatial coordinates and building geometric components, and to associate any three-dimensional coordinate with its corresponding geometric component through a mapping function, so as to achieve accurate indexing of spatial location to physical component.

[0068] When acquiring potential conflict points and related building information model (BIM) data, we first start with previously identified conflict points, such as the connection between aluminum alloy keel fixing and unit panel hoisting. We extract the three-dimensional coordinate data of these points; for example, the coordinates of the keel fixing point are (120, 45, 30), and the coordinates of the unit panel hoisting point are (120, 45, 35). These coordinates are imported using BIM software such as Revit. The BIM model includes floor geometry components such as steel structural beams and glass curtain wall panels.

[0069] The coordinates of potential conflict points are mapped to geometric components within the model, establishing spatial indexing relationships. Specifically, this mapping process involves coordinate system alignment. First, the coordinates of the conflict points are converted to the model's global coordinate system. Then, spatial query algorithms such as R-tree indexing are used to associate the joist fixing points with specific beam components, forming an index table where each conflict point is linked to a corresponding geometric entity ID (Identity number), such as beam ID B001. This relationship establishment helps quickly locate the conflict within the model, supporting subsequent visualization and analysis. In this way, potential conflict points are no longer isolated but embedded in the overall framework of the Building Information Model, ensuring data consistency and traceability.

[0070] Step S320: Access the real-time sensor stream of the monitoring and early warning module based on the spatial index relationship, analyze the geometric components bound to dynamic environmental parameters to quantify the interference intensity of curtain wall construction operations and determine the coordination requirement level.

[0071] The coordination requirement level is determined using the following formula: (11) In formula (11), This indicates the hierarchical determination result of the coordination requirement level. This represents the overall risk assessment value. Indicates the low-risk threshold boundary. Indicates the threshold boundary of medium risk. Indicates the high-risk threshold boundary, numerical value arrive These correspond to four coordination need levels: low, medium, high, and emergency. The control logic of formula (11) is to map the comprehensive risk assessment value to different coordination need levels based on the threshold range in which it falls, thereby achieving a quantitative classification and determination of risk level to coordination priority.

[0072] When accessing the real-time sensor stream of the monitoring and early warning module based on spatial index relationships, the monitoring and early warning module is a system integrating a sensor network, including IoT sensors installed at the construction site, such as temperature sensors and vibration sensors, which collect data streams in real time. The access process binds geometric components to the sensor stream through an index table; for example, beam component B001 is bound to nearby vibration sensor data. The geometric components bound to dynamic environmental parameters are then analyzed. Dynamic environmental parameters refer to factors that change in real time on site, such as wind speed and humidity. Specifically, the analysis process first extracts parameter values ​​from the sensor stream, such as a wind speed of 5 m / s. Then, an interference intensity quantification formula is applied to map the parameters to the components to calculate the interference value. For example, the interference intensity of wind speed on curtain wall operations is quantified by multiplying the wind speed value by an empirical coefficient; a score of 0.6 indicates moderate interference. The coordination requirement level is then determined, based on the interference intensity classification; for example, a score exceeding 0.5 indicates a high requirement level. This quantification helps identify the severity of operational interference, providing a basis for safety management.

[0073] Step S330: If the coordination requirement level is higher than the preset operation safety threshold, then construct a constraint adjustment matrix that satisfies the spatiotemporal avoidance conditions.

[0074] The following formula is used to construct the constraint adjustment matrix that satisfies the spatiotemporal avoidance conditions: (12) In formula (12), Represents the constraint adjustment matrix. This indicates the number of parameter dimensions that need to be adjusted in the system. Indicates the first The constraint condition applies to the first... The adjustment coefficients of each control variable. The control logic of formula (12) uses a matrix to structure the adjustment relationship between "constraints" and "control variables". The coefficients in the matrix clarify the adjustment strength of each constraint on different control variables, providing a quantitative constraint basis for parameter optimization under spatiotemporal avoidance conditions.

[0075] If the coordination requirement level is higher than the preset work safety threshold, such as 0.7, a constraint adjustment matrix that satisfies the spatiotemporal avoidance conditions is constructed. For example, the constraint adjustment matrix can be a two-dimensional array, where rows represent processes such as keel fixing, and columns represent time slots and spatial zones. Specifically, the construction process first identifies avoidance conditions, such as avoiding the overlap of multiple processes on the same floor at the same time, and then fills the matrix values. For example, a matrix element (1, 1) of 1 indicates that keel fixing in time slot 1 is feasible, and 0 indicates that it is not feasible. Through this constraint adjustment matrix, sufficient spatiotemporal intervals are ensured between the adjusted processes.

[0076] Step S340: Input the constraint adjustment matrix into the original schedule network to perform topology rearrangement and obtain the optimized job sequence.

[0077] The following formula is used to determine the optimal position of each job in the rearranged sequence: (13) In formula (13), Indicate homework At the new position in the optimized sequence, Indicate homework With homework Dependency weights between them Indicate homework In position The cost of time delay, Indicates position Priority penalty factor, Represents the balance parameters. This indicates the total number of assignments. Representation and assignment The total number of jobs with dependencies. The control logic of formula (13) is to comprehensively weigh the "weight of the dependency relationship between jobs" and the "delay / priority cost of job position", and find the optimal position of the job in the rearranged sequence by minimizing the weighted sum of the two, so as to achieve reasonable optimization of the construction sequence.

[0078] When the constraint adjustment matrix is ​​input into the original schedule network for topology rearrangement, the original schedule network is a directed graph where nodes represent work processes and edges represent dependencies. After inputting the constraint adjustment matrix, a topology sorting algorithm is used to adjust the node order. For example, the hoisting of unit panels is moved to a later stage than the fixing of the keel, resulting in an optimized work sequence such as fixing the keel first and then hoisting the panels. This rearrangement improves construction efficiency.

[0079] Furthermore, in the knowledge graph-based curtain wall construction management method proposed in this embodiment, step S400 includes: Step S410: Parse the optimized job sequence to extract the process path with time constraints.

[0080] When analyzing optimized work sequences to extract time-constrained process paths, the process begins with the optimized sequences. For example, in curtain wall construction, the sequence might involve steel structure installation followed by glass panel assembly. These work sequences are represented as paths using graph theory, with nodes representing processes such as welding and hoisting, and edges representing time constraints such as welding needing to be completed before hoisting. Specifically, this analysis process involves traversing the sequences and identifying critical paths. For instance, PERT network analysis can be used to find paths from the start point to the end point, ensuring that the start time of each process on the path is no earlier than the end time of the previous process. Through this extraction, the process paths not only reflect the temporal order but also incorporate dependencies, providing a foundation for subsequent logistics planning.

[0081] Step S420: Based on the process path, generate a material distribution matrix that meets the logistics constraints after verifying the yard capacity.

[0082] When generating a material distribution matrix that conforms to logistical constraints based on the process path and after yard capacity verification, yard capacity verification refers to checking the available space in the on-site storage area. For example, in a curtain wall project, the yard capacity is 500 square meters for storing aluminum alloy frames and glass panels. Specifically, a material list is first compiled based on the process requirements along the path, such as 10 tons of steel required for welding. Then, a verification algorithm compares the requirements with the capacity; if the requirements exceed the capacity, the distribution is adjusted. The generated material distribution matrix is ​​a table where rows represent material types (e.g., frames), columns represent yard zones, and element values ​​are the allocated quantities. This material distribution matrix ensures that materials arrive in a timely manner during path execution, avoiding logistical bottlenecks and thus improving the overall construction smoothness.

[0083] Step S430: Combine the material distribution matrix to correct the execution time of the process nodes and determine the operation time window that meets the manpower distribution requirements.

[0084] The execution time after the process node is corrected is calculated using the following formula: (14) In formula (14), Indicates the first The revised execution time for each process node Indicates the first The basic execution time of each process node Represents the first element in the material distribution matrix. The first material for the first The influence coefficient of each process node Indicates the first Weighting factors for various materials The total number of material types is indicated. The control logic of formula (14) is based on the basic execution time of the process. By multiplying the product term of "material influence coefficient × material weight", the correction effect of various materials on the execution time of the process is comprehensively quantified, so as to obtain the process time that is more in line with the actual construction conditions.

[0085] The optimal work time window that satisfies the requirements of manpower distribution is derived by the following formula: (15) In formula (15), This represents the optimal work time window that satisfies the requirements for manpower distribution. Indicates the first Manpower demand for a given time period Indicates the first The first process in the Staffing levels for a given time period Indicates the first A set of procedures executed within a time period. Indicates the total number of time periods. The set of time windows is represented. The control logic of formula (15) is to minimize the deviation between "the demand for manpower in a time period" and "the total manpower configuration of all execution procedures in that time period", and select the optimal operation time window that meets the manpower distribution requirements from the set of time windows to achieve the supply and demand matching of construction manpower.

[0086] When adjusting the execution time of process nodes by combining the material distribution matrix to determine the work time window that meets the manpower distribution requirements, the adjustment of execution time involves adjusting the material availability in the matrix. For example, if the supply of frame materials is delayed, the welding node time will be extended from the original 8 hours to 10 hours. Specifically, the time window is the range in which the process can be executed, such as from 9:00 AM to 5:00 PM, which needs to meet the manpower distribution requirements, ensuring that the number of workers in each window does not exceed the safety limit. By integrating matrix data and calculating the adjusted time, the manpower is ensured to be evenly distributed within the time window. This determination process helps reduce waiting time and improve resource utilization efficiency.

[0087] Step S440: Integrate the spatiotemporal parameters of the operation time window and the process path, and calculate the precise start and end times of the process nodes to generate a draft scheduling scheme containing resource configuration details.

[0088] The precise start time of a process node can be calculated using the following formula: (16) In formula (16), Indicates the first The start time of each process node Indicates the earliest start time of the task time window. Indicate process The set of precursor processes, Indicates the precursor process End time, Indicates from process to process The control logic of formula (16) is to simultaneously consider the "earliest limit of the operation time window" and the "completion of the predecessor process + transportation connection requirements", and take the larger value of the two as the precise start time of the process, so as to ensure that the process starts neither earlier than the time window requirements, and can connect with the results of the predecessor process.

[0089] The precise end time of a process node can be calculated using the following formula: (17) In formula (17), Indicates the first The end time of each process, Indicates the first The start time of each process, Indicate process Standard processing time Indicate process The workload, Indicates assignment to process The control logic of formula (17) is based on the start time of the process, combined with the "standard processing time" and the "matching degree between workload and resource capacity", to dynamically adjust the processing time, thereby obtaining the process end time that fits the actual resource conditions.

[0090] The following formula is used to generate a draft scheduling scheme containing resource configuration details: (18) In formula (18), Represents a complete set of scheduling schemes. Indicates the first Each process operation The first allocation One resource, and Each represents a process. The start and end times, This represents the set of all processes. This represents the latest completion time of the operation time window. The control logic of formula (18) integrates three core information categories: "operation process, resource allocation, and time interval". Through the constraint that "the end time of the operation process is not later than the latest completion time of the time window", it constructs a set of construction scheduling schemes that include both resource allocation details and time requirements.

[0091] When integrating the time windows and spatiotemporal parameters of the work process paths to calculate the precise start and end times of each process node and generate a draft scheduling plan containing resource allocation details, the spatiotemporal parameters include the location coordinates and time constraints of the processes. For example, the location of a welding node on the path is at coordinates (100, 50, 20) on floor 3. Specifically, the calculation process first merges the time windows and the path, then solves for the start and end times for each node, such as welding starting at 9:00 and ending at 19:00. Then, a draft scheduling plan is generated, with details including resource allocation such as 5 workers and 2 cranes. This integration ensures a comprehensive draft scheduling plan, and its application significantly reduces the risk of construction delays and promotes on-time project completion.

[0092] Preferably, in the knowledge graph-based curtain wall construction management method proposed in this embodiment, step S500 includes: Step S510: Analyze the draft scheduling scheme and extract the execution deviation feature vector that shows a tendency for timing delays or resource conflicts.

[0093] The following formula is used to extract time-lag feature vectors: (19) In formula (19), Represents the timing lag bias vector. Indicates the total number of scheduled tasks. Indicates the first The weighting coefficients of each task. Indicates the first The actual execution time of each task Indicates the first The planned execution time for each task Indicates the first The unit vector corresponding to each task. The control logic of formula (19) is to calculate the "lag difference between actual execution time and planned time (only retaining the case of actual timeout)" for each scheduled task, and then combine the task weight and unit vector to integrate the lag deviation of all tasks into a structured feature vector to quantify the overall time lag degree.

[0094] The following formula is used to identify resource conflict feature vectors: (20) In formula (20), Represents the resource conflict deviation vector. Indicates the number of resource types. Representing resources and resources Conflict indication functions between them Representing resources and resources Overlap measure The vector represents the direction of resource conflict. The control logic of formula (20) is to identify the conflict relationship between different resources, combine the degree of overlap and direction of the conflict, integrate the conflict information of all resource pairs into a structured feature vector, and quantify the overall situation of resource conflict.

[0095] When analyzing a draft scheduling plan to extract execution deviation feature vectors that exhibit timing lags or resource conflicts, the system first identifies potential problems within the draft plan. For example, in a curtain wall installation project, the draft plan might show that the glass panel assembly process starts two hours later than the steel structure welding finishes, constituting a timing lag. Specifically, this extraction process involves scanning all process nodes in the draft scheduling plan and quantifying deviations such as lag duration or resource overlap. For instance, if welding and hoisting simultaneously require the same crane, it is marked as a resource conflict. The deviation feature vector is a multi-dimensional array whose elements include lag values ​​(e.g., 2.5 hours) and conflict indices (e.g., 0.8), representing the severity of the conflict. In this way, the system can systematically capture execution deviations in the draft plan, laying the foundation for subsequent analysis.

[0096] Step S520: Input the execution deviation feature vector into the multidimensional knowledge network model, and update the relation weights in the multidimensional knowledge network according to the historical interaction frequency.

[0097] The relation weights in the updated multidimensional knowledge network are derived using the following formula: (twenty one) In formula (21), Indicates the updated node To the node Relationship weights This indicates the relation weight at the current moment. Indicates historical interaction factors. Represents a node With nodes The frequency of historical interactions between them Represents a node With nodes The frequency of historical interactions between them Represents nodes The number of all connected nodes. The control logic of formula (21) is based on the current relationship weight, combined with the "relative proportion of historical interaction frequency between nodes" and the "historical interaction influence factor", to dynamically adjust the relationship weight between nodes so that the weight is more in line with the actual interaction intensity between nodes.

[0098] When the execution deviation feature vector is input into a multidimensional knowledge network (MMN) model, and the relation weights in the MMN are updated based on historical interaction frequencies, the MMN model is a graph-based framework. Nodes represent processes such as welding or assembly, and edges represent dependencies between them, such as timing or resource sharing. Specifically, after inputting the execution deviation feature vector, the MMN model queries historical data. For example, if the interaction frequency of welding delays and hoisting conflicts in a past project was 15 times, the weight of the corresponding edge would be updated from 0.6 to 0.75. This update is achieved through a weighted average algorithm to ensure that the network reflects the latest risk patterns.

[0099] Step S530: Construct a risk transmission matrix based on the updated relationship weights and calculate the cumulative failure probability to mark high-risk warning objects.

[0100] The cumulative failure probability is derived using the following formula: (twenty two) In formula (22), This represents the comprehensive risk score of high-risk warning targets. Indicates the total number of objects being evaluated. Indicates the first Risk weight coefficient for each object Indicates the first The current risk score of each object. Indicates the warning threshold. Indicates the first The urgency adjustment factor for each object. The control logic of formula (22) is to first screen out the objects whose "current risk score exceeds the warning threshold" for each high-risk warning object, and then combine their risk weight and urgency adjustment factor to accumulate the risk contribution of these objects to obtain the comprehensive cumulative failure probability and quantify the overall risk accumulation.

[0101] When constructing a risk transmission matrix based on the updated relationship weights and calculating the cumulative failure probability to mark high-risk warning objects, the risk transmission matrix is ​​a square matrix where rows and columns correspond to process nodes, and element values ​​are weights, such as 0.75, representing the probability that a failure at one node will propagate to another. Specifically, calculating the cumulative failure probability involves matrix multiplication. For example, the initial failure vector is multiplied by the transmission matrix to obtain a probability, such as 0.9, where a welding failure leads to overall path failure. Nodes with a probability exceeding 0.7 are then marked as high-risk objects, such as glass assembly nodes.

[0102] Step S540: Generate a set of control parameters containing resource replenishment and process adjustment instructions for high-risk early warning objects to obtain the management configuration under risk prediction.

[0103] The control parameter set is obtained through the following formula: (twenty three) In formula (23), Represents the optimal set of control parameters. This indicates the total number of high-risk warning targets. Indicates the first The weighting coefficient of each risk object Indicates the first Risk rating of each object, Indicates the first The amount of additional resources required for each object, Indicates the first The process adjustment cost for each object This represents all possible control parameter schemes. The control logic of formula (23) is to comprehensively weigh the "risk level of the risk object" and the "corresponding resource supplementation and process adjustment costs", and to select the optimal set of control parameters from the available schemes by minimizing the weighted sum of risk and cost, so as to achieve a balance between risk control and cost / resources.

[0104] For high-risk early warning targets, a set of control parameters is generated, including instructions for resource replenishment and process adjustment. When obtaining the management configuration under risk prediction, the control parameter set is a collection, including instructions such as adding 3 workers to the welding node or adjusting the assembly start time by 1 hour. Specifically, the management configuration under risk prediction refers to a forward-looking, structured, and dynamic allocation and adjustment plan for management elements (resources, processes, personnel, time, costs, etc.) based on the pre-identified potential risks. This generation process first assesses the target's needs. For example, if the assembly node has a high risk, the process adjustment instruction specifies that additional glass plates be transferred from the spare stockpile to ensure that the configuration covers the predicted risk.

[0105] Please see Figure 2This embodiment provides a knowledge graph-based curtain wall construction management system for executing the aforementioned knowledge graph-based curtain wall construction management method. It includes a semantic association structure acquisition module 10, a potential conflict point determination module 20, an optimized work sequence acquisition module 30, a scheduling plan draft determination module 40, and a management configuration acquisition module 50. The semantic association structure acquisition module 10 uses knowledge extraction methods to extract entities and relationships from building construction data, constructs a multi-dimensional knowledge network, and obtains a semantic association structure containing construction parameters. The potential conflict point determination module 20 uses an association query method to traverse the nodes based on the multi-dimensional knowledge network. The semantic path is used to identify potential conflict points for construction coordination issues; the optimized work sequence acquisition module 30 is used to determine the coordination requirement level by integrating building information model data through the monitoring and early warning module if the potential conflict point involves the coordination of curtain wall construction operations, and obtain the optimized work sequence; the scheduling plan draft determination module 40 is used to extract the process path from the optimized work sequence, combine the resource allocation logic to process the distribution of materials and manpower, and determine the scheduling plan draft; the management configuration acquisition module 50 is used to integrate the application loop in the feedback mechanism, update the relation weights in the multidimensional knowledge network, and obtain the management configuration under risk prediction for the scheduling plan draft.

[0106] This embodiment provides a knowledge graph-based curtain wall construction management method and system. Addressing the challenges of complex work coordination, frequent conflicts during construction, uneven resource allocation, and difficulty in timely prediction of potential risks in curtain wall construction, this method extracts entities and relationships from curtain wall construction data using knowledge extraction methods. This constructs a multi-dimensional knowledge network containing construction parameters, forming a semantic association structure. Based on this, an association query method traverses the semantic paths between nodes to identify potential conflict points in construction connections. If a conflict involves work coordination, building information model data is integrated, and a monitoring and early warning module determines the coordination requirement level, generating an optimized work sequence. Work process paths are extracted from this sequence, and material and manpower distribution are processed using resource allocation logic to form a draft scheduling plan. Finally, an integrated feedback mechanism updates the relation weights in the knowledge network, enabling dynamic management configuration under risk prediction. This embodiment effectively solves the problems of conflict identification, coordination optimization, and resource scheduling in curtain wall construction through the construction and iterative updating of the knowledge network, improving construction efficiency, reducing risks, and achieving intelligent management.

[0107] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention. Clearly, those skilled in the art can make various alterations and modifications to the invention without departing from its spirit and scope. Thus, if these modifications and modifications of the invention fall within the scope of the claims and their equivalents, the invention is also intended to include these modifications and modifications.

Claims

1. A knowledge graph-based method for curtain wall construction management, characterized in that, Includes the following steps: S100. Using knowledge extraction methods, entities and relationships are extracted from construction data to construct a multi-dimensional knowledge network, resulting in a semantic association structure containing construction parameters. S200. Based on the multidimensional knowledge network, the semantic path between nodes is traversed using the association query method to identify potential conflict points for construction coordination issues. S300. If the potential conflict point involves the coordination of curtain wall construction operations, the monitoring and early warning module integrates building information model data to determine the coordination requirement level and obtain an optimized work sequence. S400: Extract process paths from the optimized work sequence, combine them with resource allocation logic to process material and manpower distribution, and determine a draft scheduling plan; S500. For the aforementioned draft scheduling scheme, the application loop in the integrated feedback mechanism is used to update the relation weights in the multidimensional knowledge network, thereby obtaining the management configuration under risk prediction.

2. The knowledge graph-based curtain wall construction management method according to claim 1, characterized in that, Step S100 includes: S110. Obtain the construction text sequence from the curtain wall construction data, and use a named entity recognition model to extract construction entity objects from the construction text sequence. S120. Analyze the context dependency path of the construction entity object to determine the construction process relationship and construction parameter indicators; S130. Construct a multidimensional knowledge graph based on the construction process relationship and the construction parameter index, and parse the topological structure of the multidimensional knowledge graph; S140. The topology is transformed into a semantic association matrix to obtain a semantic association structure containing construction parameters.

3. The knowledge graph-based curtain wall construction management method according to claim 1, characterized in that, Step S200 includes: S210. Obtain the construction process entities in the multidimensional knowledge network and extract the set of construction connection node pairs with logical succession relationships; S220. Traverse the set of construction connection nodes to retrieve multi-hop semantic paths. If the nodes under the multi-hop semantic path have spatiotemporal overlap, generate a spatiotemporal conflict feature vector. S230. Calculate the comprehensive conflict index based on the spatiotemporal conflict feature vector to obtain a set of high-risk connection paths where the comprehensive conflict index exceeds the limit; S240. Based on the high-risk connection path set, reverse index to the corresponding construction process entity to determine potential conflict points in construction connection.

4. The knowledge graph-based curtain wall construction management method according to claim 1, characterized in that, Step S300 includes: S310. Obtain the potential conflict points and associated building information model data, and map the three-dimensional coordinates of the potential conflict points to geometric components in the building information model to establish a spatial index relationship. S320. Based on the spatial index relationship, access the real-time sensor stream of the monitoring and early warning module, analyze the geometric components bound to dynamic environmental parameters to quantify the interference intensity of curtain wall construction operations and determine the coordination requirement level. S330. If the coordination requirement level is higher than the preset operation safety threshold, then construct a constraint adjustment matrix that satisfies the spatiotemporal avoidance conditions. S340. Input the constraint adjustment matrix into the original schedule network to perform topology rearrangement and obtain the optimized job sequence.

5. The knowledge graph-based curtain wall construction management method according to claim 1, characterized in that, Step S400 includes: S410. Parse the optimized job sequence to extract the process path with time constraints; S420. Based on the process path, a material distribution matrix that meets logistics constraints is generated after verifying the yard capacity. S430. Based on the material distribution matrix, correct the execution time of the process node to determine the operation time window that meets the manpower distribution requirements; S440. Integrate the spatiotemporal parameters of the operation time window and the process path, and calculate the precise start and end times of the process node to generate a draft scheduling scheme containing resource configuration details.

6. The knowledge graph-based curtain wall construction management method according to claim 1, characterized in that, Step S500 includes: S510. Analyze the draft scheduling scheme to extract the execution deviation feature vector that shows a tendency for timing delays or resource conflicts; The following formula is used to extract time-lag feature vectors: ； in, Represents the timing lag bias vector. Indicates the total number of scheduled tasks. Indicates the first The weighting coefficients of each task. Indicates the first The actual execution time of each task Indicates the first The planned execution time for each task Indicates the first Unit vectors corresponding to each task; The following formula is used to identify resource conflict feature vectors: ； in, Represents the resource conflict deviation vector. Indicates the number of resource types. Representing resources and resources Conflict indication functions between them Representing resources and resources Overlap measure Represents the direction vector of resource conflict; S520. Input the execution deviation feature vector into the multidimensional knowledge network model, and update the relation weights in the multidimensional knowledge network according to the historical interaction frequency. S530. Construct a risk transmission matrix based on the updated relationship weights, and calculate the cumulative failure probability to mark high-risk warning objects; S540. Generate a set of control parameters containing resource replenishment and process adjustment instructions for the high-risk early warning object to obtain the management configuration under risk prediction.

7. The knowledge graph-based curtain wall construction management method according to claim 6, characterized in that, In step S520, the relation weights in the updated multidimensional knowledge network are obtained using the following formula: ； in, Indicates the updated node To the node Relationship weights This indicates the relation weight at the current moment. Indicates historical interaction factors. Represents a node With nodes The frequency of historical interactions between them Represents a node With nodes The frequency of historical interactions between them Represents nodes The number of all connected nodes.

8. The knowledge graph-based curtain wall construction management method according to claim 7, characterized in that, In step S530, the cumulative failure probability is obtained using the following formula: ； in, This represents the comprehensive risk score of high-risk warning targets. Indicates the total number of objects being evaluated. Indicates the first Risk weight coefficient for each object Indicates the first The current risk score of each object. Indicates the warning threshold. Indicates the first An urgency adjustment factor for each object.

9. The knowledge graph-based curtain wall construction management method according to claim 8, characterized in that, In step S540, the control parameter set is obtained using the following formula: ； in, Represents the optimal set of control parameters. This indicates the total number of high-risk warning targets. Indicates the first The weighting coefficient of each risk object Indicates the first Risk rating of each object, Indicates the first The amount of additional resources required for each object, Indicates the first The process adjustment cost for each object This represents all possible control parameter schemes.

10. A knowledge graph-based curtain wall construction management system, used to execute the knowledge graph-based curtain wall construction management method as described in any one of claims 1 to 9, characterized in that, include: The semantic association structure acquisition module (10) is used to extract entities and relationships from building construction data using knowledge extraction methods, construct a multi-dimensional knowledge network, and obtain a semantic association structure containing construction parameters. The potential conflict point determination module (20) is used to determine potential conflict points for construction connection problems by traversing the semantic paths between nodes according to the multidimensional knowledge network and using the association query method. The optimized work sequence acquisition module (30) is used to determine the coordination requirement level by integrating building information model data through the monitoring and early warning module if the potential conflict point involves the coordination of curtain wall construction operations, and to obtain the optimized work sequence. The scheduling scheme draft determination module (40) is used to extract process paths from the optimized work sequence, combine the resource allocation logic to process the distribution of materials and manpower, and determine the scheduling scheme draft. The management configuration acquisition module (50) is used to update the relation weights in the multidimensional knowledge network for the application loop in the integrated feedback mechanism for the draft scheduling scheme, so as to obtain the management configuration under risk prediction.

Images